Hindawi Publishing CorporationInternational Journal of Distributed Sensor NetworksVolume 2013 Article ID 642986 13 pageshttpdxdoiorg1011552013642986

Research ArticleResearch on Migration Strategy of Mobile Agent in WirelessSensor Networks

Dai Ting12 Huang Haiping123 Lu Yang12 Wang Ruchuan124 and Pan Xinxing12

1 College of Computer Nanjing University of Posts and Telecommunications Nanjing 210003 China2 Jiangsu High Technology Research Key Laboratory for Wireless Sensor Networks Nanjing 210003 China3 College of Computer Science and Technology Nanjing University of Aeronautics and Astronautics Nanjing 210016 China4Key Lab of Broadband Wireless Communication and Sensor Network Technology of Ministry of EducationNanjing University of Posts and Telecommunications Nanjing 210003 China

Correspondence should be addressed to Huang Haiping hhpnjupteducn

Received 16 July 2013 Accepted 9 September 2013

Academic Editor Chongsheng Zhang

Copyright copy 2013 Dai Ting et alThis is an open access article distributed under the Creative Commons Attribution License whichpermits unrestricted use distribution and reproduction in any medium provided the original work is properly cited

Big data and distributed computing are of great importance in wireless sensor networks (WSNs)They are always bonded togetherand the latter one upholds the former one In distributed computing mobile agent model is the mainstream technology Withautonomy communicativeness mobility and role mobile agent model is more suitable for large-scale resources-restrained WSNto deal with big dataThis paper mainly studies migration schemes for mobile agents determines core factors of migration strategyand proposes SMLA and DMLA algorithms And aiming at revealing the characteristics of target tracking this paper puts forwardpid-DMLA algorithm and considering multiple agentsrsquo cooperation it presents Mpid-DMLA algorithm Moreover this paperevaluates and analyzes the advantages and disadvantages of the above-mentioned four algorithms by simulations

1 Introduction

In wireless sensor networks (WSNs) big data is the intensi-fication of information architecture while distributed com-puting is the intensification of infrastructure and resource[1] Distributed computing models can bring benefits to bigdata And there are two categories of distributed computingclientserver model and mobile code model Mobile agentmodel consists of remote computation code on demand andmobile agent model [2]

In clientserver model client asks for services whileserver provides resources services and methods for execut-ing services There are many distributed systems adoptingthis model [3 4] Mainwaring proposed a habitat monitor-ing system which is accomplished by transmitting data togateway [3] Ivy is the distributing system inWSN studied bytheUniversity of California Berkeley CAUSA [4] Althoughthe clientserver model has been widely used its drawbacksare very obvious especially in WSN [5] Firstly it needsa series of supernodes as a processing center Secondly inmulticlient systems overdependence on bandwidth reduces

the transmitting performance Thirdly the nodes which arecloser to the server need to transmit more data thus theyconsume more energy

Mobile agent model supports distributed computing tooThe processing centercentralized information system (CIS)is in charge of sending and recovering mobile agents whileagents migrate between sensor nodes and are responsiblefor data exchange A typical exemplar is the data fusionalgorithm based on mobile agent to track objects in WSN[6] In general mobile agent is a special program whichcan be executed autonomously And it is also defined asan entity with the following four attributes agent identifierdata space method and migration table Agent identifierdescribes the uniqueness of each agent data space carriespart of the integrated results method is used to deal withmissions or execute codes and migration table stores therouting information of an agent and it can be preset ormodified dynamically

Compared with traditional computing models whichneed amount of data interaction between nodes and CIS(or base station) mobile agent model has many merits As

2 International Journal of Distributed Sensor Networks

illustrated above inWSNmobile agents dynamicallymigrateto the CIS with service request With CIS accomplishing thecomputing operations locally and returning the results to thenodes agents can be devoid of the transmission of big dataand the reliance on the network bandwidth which improvesthe efficiencyThe specific advantages of mobile agent modelespecially in terms of energy consumption reliability andlarge-scale distribution are declared in [7] However theconsequent challenges are not negligible like the high failurerate unreliable communication link high dependence onbandwidth and so on

2 Related Work

Although mobile-agent-based distributed model has excep-tional advantages in WSN [8] dynamic migration is neces-sary in practical deployment Dynamic migration presentedby migration table has two behaviors nodes selection andmigration sequences Therefore migration tablersquos efficiencyin some sense decides accuracy of data integrations timeof migrations energy consumption and performance of thewhole network

In response to the challenges embodied in the specialtyof WSN the design of mobile agentrsquos migration table needsto adequately consider the following three aspects Firstlyagent entity must be as small as possible to reduce migrat-ing energy consumption Secondly migration table has toguarantee rational precision of data integrations to obtainsufficient sensor nodes Thirdly in the precondition of thesecond aspect migration table needs to be short enough Inconclusion a suitable migration table should balance the costand precision

Thus in this paper we mainly concentrate on the designof the migration table which can reduce agentrsquos migrationenergy consumption while guaranteeing reliable migrationmethods

Briefly mobile agent migration has two categories staticmigration list algorithm (SMLA) and dynamic migrationlist algorithm (DMLA) In SMLA CIS analyzes the wholenetworkrsquos resources and designs migration paths that areinserted in agent entities initially However static SMLAcannot respond to the changes of the network timely thechanges happen when one or more links between nodesbreak while in DMLA migration table is updated at eachmigration interval according to the sensibility of the network

Without loss of generality mobile agent migration algo-rithms can be illustrated by target tracking Consideringmovements of the target entity Zhao et al proposed theinformation-driven-based tracking method [9] Liu et alproposed the collaborative in-networkmethod to track target[10] which effectively reduced overheads and energy con-sumption Kim et al transferred the distributed tracking issueto the continuous Bayesian estimation problem and adoptedthe information-driven-based sensor querying scheme toselect the next hop [11]Wang et al proposed a heuristic targetlocalization method based on entropy [12] which is moreefficient than the information-based computing methodIn this paper we will propose the dynamic migration list

algorithm based on the predicting information drive (pid-DMLA) to solve the motion problems of migrations of thetarget entity

Considering the condition of using multiple agents totrack an application in practice we will also propose thedynamic migration list algorithm based on the predictinginformation drive of multiple agents (Mpid-DMLA)

In this paper we would propose themigration algorithmsin different scenarios single mobile agent and multiplemobile agents which is valuable and applicable in WSNsThe rest of this paper is organized as follows Section 3describes the preliminary of migration Section 4 presentssingle-agent-based migration algorithms including SMLADMLA and pid-DMLA Section 5 illustrates Mpid-DMLASimulation results and analyses are presented in Section 6and conclusions are drawn in Section 7

3 Preliminary

31 Assumptions The following assumptions are considered

(1) The location of each sensor node is known(2) Each node has at least two neighbor nodes(3) The valid perception radius of sensor node is more

than half the distance between two neighboringnodes

(4) All nodes are synchronized(5) All sensors are deployed in a two-dimensional region(6) There is only one target entity in the region and it

moves uniformly and linearly

32 Evaluation Indicators of the Agent Migration AlgorithmIn order to evaluate the existing agent migration algorithmswe adopt the three indicators proposed by Xu et al [7] energyconsumption network lifetime and migration hops

Each migration step of mobile agent is called a hop Andthe energy consumption of each hop consists of three aspectstransmission consumption 119864

119886119887(119864119886119887

represents the energyconsumption of an agent migrating from node 119886 to node119887) initialization consumption 119864

119894(119864119894is constant) and data

processing consumption The network energy consumptionis primarily concentrated in transmissions

Amigration is primarily composed of four parts sendingan agent receiving an agent data processing and decidingthe next hop And each consumption time of the fourprocesses is fixed Thus evaluating the migration time of anagent is equivalent to calculating the migration hops

33 Nodersquos Sensor Model We adopt the following wirelesstransmission model [13] proposed by Rappaport

119911119896(119905) prop

119864

1003817100381710038171003817119883 (119905) minus 119883119896

1003817100381710038171003817

2 (1)

where 119911119896(119905) is a measure value of sensor 119896 related to the

location and time 119864 is the signal power of the target entity119883(119905) represents the location of the target entity in time

International Journal of Distributed Sensor Networks 3

A A AB B BD D

r1r1

r1

r2 r2r2

D1

D2

C C C

Figure 1 The target localization algorithm

119905 119883119896represents the location of node 119896 119883(119905) and 119883

119896are

two-dimensional vectors and 119883(119905) minus 119883119896 represents the

Euclidean distance between node 119896 and target entity intime 119905

34 Nodersquos Information CollectionModel In order to quantifythe data provided by sensors we need to define the amountof information According to (1) at any time 119911

119896(119905) is only

related to 119883(119905) minus 119883119896 Thus adopting the Gaussian model

the amount of information when the agent migrates to node119896 is the following

119868119896(119905) =

1

radic2120587120590

119890minus(119905)minus119883119896

2

21205902

(2)

where 120590 is the standard deviation of the amount of infor-mation which indicates the decrease of speed of 119868

119896(119905) when

the distance between sensor node and target entity increasesand 119883(119905) represents the location of the target entity which iscalculated by the target localization algorithm in Section 36

35 Beacon Frames We use beacons to termly exchangeinformation of the neighboring nodes A beacon frame con-sists of the following five fields 119883

119896represents the location of

node 119896 119890119896(119905) represents the residual energy of node 119896 in time

119905 119911119896(119905) represents the sensor measurement of node 119896 in time

119905 and the last two fields are sending time 119905119904and receiving time

119905119903 Beacon frames are used to provide information of location

andmeasurement to the target localization algorithm and theagentmigration algorithmMoreover they can be used to testthe survivability of neighbor nodes

36 Target Localization Algorithm As illustrated in (2) thelocation of target entity is calculated by the target localizationalgorithm At present target localization algorithms mostlyadopt the method of centroid weighting And the basic ideais to make nodersquos beacons take control of the centroidrsquoscoordinate and to reflect the general relations of the locationsby using weighting factors which influence the positions ofthe nodesrsquo centroids And according to the RSSIs of the threesensorsrsquo beacons it is easy to obtain the other nodersquos locationby the trilateration

As shown in Figure 1 the locations of 119860(119909119860 119910119860)

119861(119909119861 119910119861) and 119862(119909

119862 119910119862) are known and one of 119863(119909

119863 119910119863)

needs to be tested According to the RSSIs we can calculate

the distance of 119860119863 119861119863 and 119862119863 denoted by 1199031 1199032 and 119903

3

respectively Thus we can obtain the following equation set

(119909119861minus 119909119863)2

+ (119910119861minus 119910119863)2

= 1199032

2

(119909119860minus 119909119863)2

+ (119910119860minus 119910119863)2

= 1199032

1

(3)

Apparently the distance of 119860 and 119861 is less than or equalto 1199031+ 1199032 So (3) has two real solutions that are identical or

not which correspond to the coordinates of the two nodes1198631

and 1198632that are identical or not Select the one that is nearer

to 119862 as the approximate location of 119863 denoted as (1199091 1199101)

In the same way we can obtain the other two approximatelocations of119863 denoted as (119909

2 1199102) and (119909

3 1199103)Therefore the

coordinate of 119863 is (119909119863 119910119863) which can be calculated by the

following

119909119863

=1199091 (1199031+ 1199032) + 1199092 (1199032+ 1199033) + 1199093 (1199033+ 1199031)

1 (1199031+ 1199032) + 1 (119903

2+ 1199033) + 1 (119903

3+ 1199031)

119910119863

=1199101 (1199031+ 1199032) + 1199102 (1199032+ 1199033) + 1199103 (1199033+ 1199031)

1 (1199031+ 1199032) + 1 (119903

2+ 1199033) + 1 (119903

3+ 1199031)

(4)

As a result according to the previous target localizationalgorithm we can obtain 119909(119905)

4 Migration Algorithms Based on SingleMobile Agent

This section would propose SMLA and DMLA algorithmsfrom static and dynamic aspects respectively And it wouldthen propose pid-DMLA based on disadvantages of SMLAand DMLA

41 SMLA In SMLA algorithm we use static informa-tion Before distributing mobile agents the CIS collects bycommunicating with other nodes and it then generates amigration plan In thismigration plan information collectionvalues of all nodes in the network decreaseThe specific stepsof SMLA are as follows

Algorithm 1 (SMLA)(1) At time 119905 = 0 the node that has first detected

information of target entity is denoted as cluster headit generates a mobile agent

(2) At time 119905 = 0 calculate location information oftarget entity 119883(0) according to the target locationalgorithm

4 International Journal of Distributed Sensor Networks

Calculate the amount of information collection 1198680

according to (2) and set 119878119906119898119860119903119903119886119910 = 1198680

Select the next hop according to the predeterminedmigration sequence

(3) At time t when agent reaches node k the location oftarget entity is 119883(119905) according to the target locationalgorithmCalculate the amount of information collection ofnode 119896 119868

119896 and update 119878119906119898119860119903119903119886119910 = 119878119906119898119860119903119903119886119910 + 119868

119896

If 119878119906119898119860119903119903119886119910 is greater than the threshold then go tostep 4 otherwise continue to migrate to the next hopin the migration sequence update time 119905 = 119905 + 1 andthen repeat step 3

(4) Return to cluster head to process data

Though sometimes SMLA can gain the ideal migrationsequence there are many drawbacks in practical applica-tions Firstly since SMLA uses static information sensorrsquostransmission distance must be large enough to ensure thatCIS is within one-hop range So we need to adopt thenodes with greater power in communication capabilitiesAnd using these nodes will inevitably result in relativelyhigher energy consumption Secondly since SMLA is a staticalgorithm the static information which is acquired beforemigration lack timeliness and cannot reflect the changes ofthe network in the process of migrating For example thatsome nodes suddenly lose efficacy can result in the failure ofthe deployment of SMLA [9] Moreover most of WSNs arehighly dynamic so SMLA has some limitations in WSNs

42 DMLA SMLA is a centralized algorithm and it hasalready decided the migration sequence before distributingmobile agents Because WSN is a dynamic and distributednetwork static routing information exposes a lot of problemsin practical applications Thus DMLA is proposed DMLAdynamically selects the next hop in the process of migrationbased on characteristics of the network

421 The Migration Model of Mobile Agent in DMLA InDMLA mobile agents decide the nodes that they need tomigrate to In the surviving neighbor nodes not all of thenodes can provide information affecting the final resultTherefore the purpose ofmobile agentmigration algorithm isto produce a set of optimal subsets to determine the optimummigration sequence

In the target entity tracking we construct rules to choosethe next hop as followsmobile agent which is located in node119896 always looks for the next hop with minimal overheadsAs the selection of the next hop is a method of balancingoverheads and profit we define the overheads function whenmobile agent moves from node 119894 to neighbor node 119895 at time119905 as follows

119862119894119895(119905) = 119886

119864119894119895

119864max+ 119887(1 minus

119868119895(119905)

119868max) + radic1 minus 119886 minus 119887(1 minus

119890119895(119905)

119890max)

(5)

Equation (5) consists of three parts including energy con-sumption information collection and residual energy where119864119894119895represents transmission energy consumption when the

agent migrates directly from node 119894 to node 119895 119868119895(119905) represents

the amount of information collection of node 119895 at time 119905119890119895(119905) represents the residual energy of node 119895119864max represents

themaximumenergy consumption ofmigration between twonodes which is determined by the maximum transmissiondistance 119868max represents the maximum amount of informa-tion collection 119890max represents the initial energy of sensornode and it is a constant value and 119886 and 119887 are the balancecoefficients 0 le 119886 and 119887 le 1

Define the transmission energy consumption of the agentmigrating from node 119894 to node 119895 as follows

119864119894119895= 119875119894119895sdot 119879trans = 120572119875

119903-th sdot10038171003817100381710038171003817119883119894minus 119883119895

10038171003817100381710038171003817

2

sdot 119879trans (6)

where 119875119894119895represents the transmission power of sensor nodes

119883119894minus 119883119895 represents the distance between node 119894 and node

119895 119879trans a constant represents the time of transmitting themobile agent 120572 = 4120587

2

1205822

11986611198662 1198661 1198662are all of the

antenna gain factors (set them as 1) 120582 represents the signalwavelength and 119875

119903-th represents the receiving thresholdFrom (2) (5) and (6) we can gain

119862119894119895(119905) = 119886

10038171003817100381710038171003817119883119894minus 119883119895

10038171003817100381710038171003817

2

1198892max+ 119887

10038171003817100381710038171003817119883 (119905) minus 119883

119895

10038171003817100381710038171003817

2

1198891199052

max

+ radic1 minus 119886 minus 119887(1 minus

119890119895(119905)

119890max)

(7)

where 119889119905max represents the maximum possible distancebetween two nodes

Equation (7) can be rewritten as follows

119895lowast

= argmin

119886

10038171003817100381710038171003817119883119894minus 119883119895

10038171003817100381710038171003817

2

1198892max+ 119887

10038171003817100381710038171003817119883 (119905) minus 119883

119895

10038171003817100381710038171003817

2

1198891199052

max

+radic1 minus 119886 minus 119887(1 minus

119890119895(119905)

119890max)

119895 isin 119873119894

(8)

where 119873119894represents a group of neighbors of node 119894 node

119895lowast

isin 119873119894represents the next hop and 119895

lowast

isin 119873119894can ensure less

energy consumption stronger sensing capabilities and moreresidual energy

In summary the total cost in the migration process is

Cost =HOP119878minus1

sum

119901=0

119862119894119901119894119901+1

st 119878119906119898119860119903119903119886119910 =

HOP119878sum

119901=0

119868119901gt threshold

(9)

where 119901 represents the number of nodes migrated throughHOP119878represents the total number of migration hops In

the practical migration once mobile agent gains enoughinformation and it will terminate migration and return tothe CIS

International Journal of Distributed Sensor Networks 5

422The SpecificMigration Procedure of DMLA Themobileagent migrates to the alive node which has not been reachedand with the minimal migration overheads according to (7)In this way we can gain a table close to the optimalmigrationIf all nodes have been reached mobile agent will return thenode that distributes it And when the mobile agent hasacquired enough information it will return to the CIS

Since each node periodically receives beacon frames fromneighbor nodes DMLA can always use the updated localvariables to determine the next hop As only using localvariables sensor nodes can avoid the nodewhich has a greatertransmission distance so part of the transmission energycan be saved In addition when some nodes suddenly loseefficacy the whole migration process will not make mistakesSoDMLAalgorithm is a fault-tolerant algorithmThe specificmigration steps of DMLA are as follows

Algorithm 2 (DMLA)

(1) At time 119905 = 0 the node that has first detectedinformation of target entity is denoted as cluster headit generates a mobile agent

(2) At time 119905 = 0 calculate location information oftarget entity 119883(0) according to the target locationalgorithmCalculate the amount of information collection 119868

0

according to (2) and set 119878119906119898119860119903119903119886119910 = 1198680

Select the next hop according to (8)(3) At time 119905 when agent reaches node 119896 the location of

target entity is 119883(119905) according to the target locationalgorithmCalculate the amount of information collection ofnode 119896 119868

119896 and update 119878119906119898119860119903119903119886119910 = 119878119906119898119860119903119903119886119910 + 119868

119896

If 119878119906119898119860119903119903119886119910 is greater than the threshold then go tostep 4 otherwise continue to migrate to the next hopaccording to (8) update time 119905 = 119905+1 and then repeatstep 3

(4) Return to cluster head to process data

43 pid-DMLA

431 The Proposition of pid-DMLA In the cooperationprocessing applications the movement of target entity isone of the important factors that affect the final result [14ndash19] DMLA algorithm does not take the direction of themovement of target entity into account So in some cases itcannot select the optimumnext hop For example in Figure 2at time 119905 the mobile agent is located in node 119860 Nodes 119861

and 119862 are neighbors of 119860 and they are equidistant to 119860According to DMLA mobile agent calculates node 119861 as thenext hop This is because node 119861 is nearer to the targetentity than node 119862 Then at time 119905 + 1 the mobile agent islocated in node 119861 But the target entity moves to a positioncloser to node 119862 Thus DMLA on the contrary selectsthe node which is farther from the target and has smalleramount of information collection and measurement value

t t + 1

A A

B B

CC

Figure 2 The movement of the target entity

as the next hop Therefore when designing agent migrationalgorithms we need to focus on predicting the movement ofthe target entity and selecting the node with the maximuminformation And we will propose the pid-DMLA algorithmto solve the problems of the movement of the target entity inthe migration process

432 Agent Migration Model in pid-DMLA Suppose thatthe moving direction and speed of the target entity do notchange Then the change of the target entityrsquos displacementis a constant at each identical time interval as follows

119883 (119905 + 1) minus 119883 (119905) = 119883 (119905) minus 119883 (119905 minus 1) (10)

Equivalently

119883 (119905 + 1) = 2119883 (119905) minus 119883 (119905 minus 1) (11)

As illustrated in (11) the position information119883(119905 + 1) attime 119905 + 1 can be deduced from the positions at time 119905 minus 1 andtime 119905 In this way the location information119883(119905minus1) and119883(119905)

can be calculated by target location algorithm and accordingto (11) the position of target entity119883(119905+1) can be estimatedat time 119905 + 1 as follows

119883 (119905 + 1) = 2119883 (119905) minus 119883 (119905 minus 1) (12)

Update the overhead function in (7) and get

119862119894119895(119905) = 119886

10038171003817100381710038171003817119883119894minus 119883119895

10038171003817100381710038171003817

2

1198892max+ 119887

10038171003817100381710038171003817119883 (119905 + 1) minus 119883

119895

10038171003817100381710038171003817

2

1198891199052max

+ radic1 minus 119886 minus 119887(1 minus

119890119895(119905)

119890max)

(13)

where 119883(119905 + 1) minus 119883119895 represents the distance between node

119895 and the target entity at time 119905 + 1And the next hop is

119895lowast

= argmin

119886

10038171003817100381710038171003817119883119894minus 119883119895

10038171003817100381710038171003817

2

1198892max+ 119887

10038171003817100381710038171003817119883 (119905 + 1) minus 119883

119895

10038171003817100381710038171003817

2

1198891199052max

+radic1 minus 119886 minus 119887(1 minus

119890119895(119905)

119890max)

119895 isin 119873119894

(14)

6 International Journal of Distributed Sensor Networks

t

t

t + 1

t + 1

t + 2

t + 2

t minus 1

t minus 1

AB DC

Figure 3 The migration of pid-DMLA

At time 119905 = 0 since the position of the target entity atthe previous time is unknown the amount of informationcollection can be calculated by the DMLA At time 119905 gt 0mobile agent migrates to the node closer to the estimatedposition of the target entity So it can obtain larger sensingmeasurement and information collection which is the coreof pid-DMLA

433 The Specific Migration Procedure of pid-DMLA Theprocess of mobile agent migration of pid-DMLA is shown inFigure 3 In a period of time the target entity moves alongthe dotted line and mobile agent migrates from node 119860 to119863

through 119861 and 119862 As shown in Figure 3 mobile agent alwaysmigrates to the node nearest to the next position of the targetentity

The specific migration procedure of pid-DMLA is asfollows

Algorithm 3 (pid-DMLA)

(1) At time 119905 = 0 the node that has first detectedinformation of target entity is denoted as cluster head

it generates a mobile agent

(2) At time 119905 = 0 calculate location information oftarget entity 119883(0) according to the target locationalgorithm

Calculate the amount of information collection 1198680

according to (2) and set 119878119906119898119860119903119903119886119910 = 1198680

Select the next hop according to (14)

(3) At time 119905 when agent reaches node 119896 the location oftarget entity is 119883(119905) according to the target locationalgorithm

Calculate the amount of information collection ofnode 119896 119868

119896 and update 119878119906119898119860119903119903119886119910 = 119878119906119898119860119903119903119886119910 + 119868

119896

If 119878119906119898119860119903119903119886119910 is greater than the threshold then goto step 4 otherwise continue to migrate to the nexthop according to (14) update time 119905 = 119905 + 1 and thenrepeat step 3

(4) Return to cluster head to process data

5 Migration Algorithm Based on MultipleMobile Agents

51 The Architecture of Mpid-DMLA Many researchers haveproposed and realized feasible ways to organize themigrationof a single mobile agent such as Agilla program [18] andSensorWare program [19] Consider the functional require-ments of these programs Firstly they have not formed anarchitecture to manage the agent migration mechanismsin WSN Secondly they lack migration algorithms whena tracking application needs the collaboration of multipleagents So when dealing with multiple agentsrsquo migration weneed a specific agent to manage the applications

Therefore we will propose Mpid-DMLA to solve thisproblem

Suppose that multiple target entity applications aredeployed in an area of sensors each application is responsiblefor monitoring a part of the area and composed of oneor more mobile agents Therefore we have the followingdefinitions

Definition 4 (code segment application) A centralized pro-gram which consists of multiple mobile agents deployed inthe WSN to accomplish a specific task is called centralizedcode segment application of multiple agents which is calledcode segment application for short

Code segment applications are associated with sensingdata that need to be evaluated and they need to cooperate andinteract based on the need of programs On the other handaccording to the Deluge amp Mate model [20] code segmentapplications need to redeploy themselves so that they can bedistributed by the CIS

To track targets code segment applications require thefollowing four functions [21]

(1) Sensing Sensor nodes are distributed around thetarget node and they test and collect information ofthe surrounding environment

(2) Processing Information After acquiring sensor infor-mation most of the nodes send it to the CIS forprocessing while cluster heads will directly processinformation

(3) Communication In order tomaintain the applicationsensor nodes need to keep exchanging informationwith other nodes or CIS

(4) Migration Mobile agents migrate based on the needof code segment application

Constructing the architecture of multiple mobile agentsalso requires the following definitions

Definition 5 (main component agent) The main componentagent is responsible for managingmigration information andactivity cycle of all component agents processing sensingdata of other agents and collecting and exchanging theinformation of subcomponent agents Each code segmentapplication has one and only main component agent

International Journal of Distributed Sensor Networks 7

Definition 6 (subcomponent agent) The subcomponentagent is used to obtain the data of nodes and report them tothe main component agent Each code segment applicationhas one or more subcomponent agents

Definition 7 (communication component agent) The com-munication component agent is used to collect and exchangethe entity information of WSN by migrating Each codesegment application has one or more communication com-ponent agents

Definition 8 (middleware software) The middleware soft-ware located between the physical and the operating systemlayers is the software entity running in mobile agents

52 The Migration Model of Mobile Agent in Mpid-DMLAWhen a code segment application changes monitoring posi-tions all of the mobile agents in this application needto migrate together to reconstruct a new code segmentapplication In Agilla the main component agent controlsthe entire process by sending the migration sequence whileother mobile agents can communicate by accessing theindependent tuple space [22] And the engine virtualmachineworks as the kernel to provide services and controls all of theconcurrent execution of all of the agents in a node Howeverwith the increase of the mobile agents the possibilities ofthe loss of packet and the failure of migration will increasetoo Suppose that a code segment application is composedof 119873 mobile agents Define 119875 as the failure coefficient of asingle agent And then we can obtain the success coefficientof migration in the entire application as follows

119875success = (1 minus 119875)119873

(15)

Communication overheads based on pid-DMLA aregiven as follows

119862119894119895(119905) = [

[

119886

10038171003817100381710038171003817119883119894minus 119883119895

10038171003817100381710038171003817

2

1198892max+ 119887

10038171003817100381710038171003817119883 (119905 + 1) minus 119883

119895

10038171003817100381710038171003817

2

1198891199052max

+radic1 minus 119886 minus 119887(1 minus

119890119895(119905)

119890max)] times (119875success)

minus1

(16)

And the next hop is

119895lowast

= argmin

[

[

119886

10038171003817100381710038171003817119883119894minus 119883119895

10038171003817100381710038171003817

2

1198892max+ 119887

10038171003817100381710038171003817119883 (119905 + 1) minus 119883

119895

10038171003817100381710038171003817

2

1198891199052max

+radic1 minus 119886 minus 119887(1 minus

119890119895(119905)

119890max)]

]

times (119875success)minus1

(17)

53 The Specific Migration Procedure of Mpid-DMLA In theconstruction of the target entity tracking we assume thatsubcomponent agent is a small program of lightweight codewhich is suitable for migration And it can independently

Figure 4 Migration based on predicting information drive ofmultiple mobile agents

access the target entityrsquos location As shown in Figure 4hollow circles represent sensor nodes which are distributedin the sensing region code segment is composed of one maincomponent agent (solid circle) four subcomponent agents(solid squares) and one communication component agent(solid triangle) Each agent runs on one node executingcode segment applications The migration of code segmentapplication means the migration of the six mobile agentswhich is the migration of multiple mobile agents [19] Thefollowing algorithm can be used to describe the above-mentioned procedure

Algorithm 9 (Mpid-DMLA)

(1) A main component agent a communication com-ponent agent and several subcomponent agents aredeployed in different sensor nodes

(2) Sensor nodes are monitoring around the target entityThe node that has first detected information of targetentity is denoted as cluster head themain componentagent then migrates to it

(3) At time 119905 = 0 calculate location information oftarget entity 119883(0) according to the target locationalgorithmCalculate the amount of information collection 119868

0

according to (2)Define the threshold of the amount of information as119879ℎ119903119890119904ℎ119900119897119889 119880119875 and let 119879ℎ119903119890119904ℎ119900119897119889 119880119875 = 119868

0

The main component agent generates new com-munication component agent and transfers 119868

0and

119879ℎ119903119890119904ℎ119900119897119889 119880119875 to itThen the communication component agent sets tomigrate

(4) At time 119905 when agent reaches node 119896 the location oftarget entity is 119883(119905) according to the target locationalgorithmCalculate the amount of information collectionof node 119896 119868

119896 and update 119879ℎ119903119890119904ℎ119900119897119889 119880119875 =

119879ℎ119903119890119904ℎ119900119897119889 119880119875 + 119868119896

If 119879ℎ119903119890119904ℎ119900119897119889 119880119875 is greater than the threshold thengo to step 5 otherwise select the next hop formigration according to (17) update time 119905 = 119905 + 1and then repeat step 3

8 International Journal of Distributed Sensor Networks

t

t

t + 1

t + 1

t + 3 t + 4 t + 5 t + 6 t + 7t + 2

t + 3

t + 4

t + 5

t + 6

t + 7

t + 2

Figure 5 The migration process of SMLA

t t + 1 t + 3 t + 4 t + 5 t + 6 t + 7t + 2

tt + 1 t + 3

t + 4

t + 5

t + 6t + 7

t + 2

Figure 6 The migration process of DMLA

(5) Communication component agent stores the infor-mation of the migration path (119868

0(119905 = 0) 119868

1(119905 =

1) 119868119896(119905 = 119896) 119883 (119905 = 0) 119883 (119905 = 1) 119883 (119905 =

119896)) and other relevant information

And it interacts with the main component agent

(6) The main component agent generates the migrationsequence 119873

1 1198732 119873

119896 according to location of

target entity

Send it to each subcomponent agent (this step isoptional for different applications)

(7) The main component agent sends complete instruc-tions to the lower-layered middleware software

Method functions which are predefined in the mid-dleware software recycle the communication compo-nent agent and the subcomponent agents

(8) The main component agent goes to the destinationnode based on the migration sequence

If the migration fails send failure information toCIS to terminate this migration otherwise generatenew communication component agent and subcom-ponent agents and this migration ends

6 Simulation and Analyses

61 The Effect of Single Mobile Agent in Target TrackingBefore the simulation experiments we firstly use targettracking as a sample to demonstrate the migration effect ofa single mobile agent of different migration algorithms Asshown in Figures 5 6 and 7 suppose that a car is runningfrom left to right through the area covered byWSNThe smallcircle represents the sensor node solid small circle representsthe cluster head which distributes agents large dotted circlerepresents the communication range of sensor node and eachnode selects the next hop only within its communicationrange The above-mentioned figures illustrate the specificlocations of the car and agent from time 119905 to time 119905 + 7

The migration process of SMLA is shown in Figure 5 Attime 119905 CIS distributes the mobile agent to the node whichis the nearest to the target entity according to the prede-termined migration sequence When the object is stationaryor its motion can be estimated this static approach is verysuitable However it loses efficacy in some conditions Forexample if the object remains in the location at time 119905+1 thenthe migration sequence will be the positions that are aroundthe target entity But once the target entity starts to move notaccording to the predetermination SMLA will fail

DMLA can in some sense optimize the performanceof SMLA As shown in Figure 6 mobile agent tracks themovement of the target entity However the movement of

International Journal of Distributed Sensor Networks 9

t

t + 1 t + 3

t + 4

t + 5

t + 6t + 7

t + 2

t t + 1 t + 3 t + 4 t + 5 t + 6 t + 7t + 2

Figure 7 The migration process of pid-DMLA

Table 1 Basic simulation parameters of single mobile agent migra-tion algorithms

Network area 500 times 500mNumber of nodes 200 or 800Nodersquos initial energy 50 JNodersquos sensing radius 5mEnergy consumption for data processing 03 JEnergy consumption for sending beacon frames 01 JThe interval of node broadcasting beacon frames 01 sThe speed of the target entity 0sim50msThe threshold of the amount of information 18The size of a mobile agent 800 bitsSimulation time 100 s

the target entity is unpredictable so it is not accurate to decidemigration path only by the amount of information which ishowever related to the distance to the target entity

The pid-DMLA can predict movement information oftarget entity dynamically which is a better solution to suchproblems As shown in Figure 7 at time 119905 + 5 mobile agentselects the next hop according to the predicted informationwhich is different from DMLA In brief pid-DMLA rebuildsmigration sequence based on predicting themovement of thetarget entity to achieve the purpose of tracking

62 Performance Analysis on the Single Mobile Agent Migra-tion Algorithms We use simulation software NS2 to eval-uate the three single mobile agent migration algorithmsrsquoperformances The default simulation parameters are shownin Table 1

Suppose that once the agent migrates one hop all sensornodes in the network consume 02 J For SMLA the transmis-sion distance is set to 30m while for DMLA and pid-DMLAthe transmission distance is set to be 10m Therefore theconsumption of transmission for SMLA is three times morethan that of sendingThe reason for this setting is that SMLAneeds bigger transmission range to collect information Thedirection and size of movement of the target entity areconstant denoted as V (ms) Once the target entity reachesthe edge of the sensing area it reverses direction immediatelywhich ensures its being always within the sensing area

0 5 10 15 20 25 30 35 40 45 500

2

4

6

8

10

12

Target speed (ms)

Ener

gy co

nsum

ptio

n (J

)

SMLADMLApid-DMLA

Figure 8 Energy consumption of the single agent migrationalgorithms in Scenario 1

Scenario 1 Set the number of nodes as 200 the target entityspeed as 0sim50ms and the simulation time as 100 s whichcan guarantee a complete migration The simulation resultsare shown in Figures 8 and 9 It is obvious that DMLA andpid-DMLA consume less energy than SMLA in the wholenetwork This is because the communication cost of SMLAis higher than that of DMLA But with the increasing speedof the target entity the effect of SMLA algorithm in trackingthe target entity becomes less and less obvious thereforemigration may terminate and the energy consumption onthe contrary becomes slower On the other hand when thespeed of target entity is very low there are no significantdifferences among the three migration schemes

Scenario 2 Set the speed of target entity as 5ms and thenumber of nodes as 800 Generally the coverage and thenumber of nodes are positively correlated The simulationresults are shown in Figures 10 and 11 It is clear that pid-DMLA has the lowest energy consumption when the numberof nodes is more than 100 This is because with the increaseof the nodes the choices of predicting the next hop become

10 International Journal of Distributed Sensor Networks

0102030405060708090

100

Life

time (

s)

0 5 10 15 20 25 30 35 40 45 50Target speed (ms)

SMLADMLApid-DMLA

Figure 9 Lifetime of the single agent migration algorithms inScenario 1

Ener

gy co

nsum

ptio

n (J

)

SMLADMLApid-DMLA

0 100 200 300 400 500 600 700 8000

1

2

3

4

5

6

7

8

9

Target speed (ms)

Figure 10 Energy consumption of the single agent migrationalgorithms in Scenario 2

more as a result the overheads become less [23] When thenumber of nodes is less than 50 the performance of SMLAis better than those of the other two This is because withless nodes static and dynamic migration algorithms share themore similar migration path however the migration tablehas been predetermined in SMLA thus it on the contraryhas the least overheads On the other hand when the densityof nodes is high there would be more neighbors and moremigration choices Thus DMLA and pid-DMLA outperformSMLA

63 Multiple Mobile Agents Migration Algorithm PerformanceAnalysis Before the assessment of Mpid-DMLA we need toconsider two problems [24 25] The first is the effect of code

SMLADMLApid-DMLA

10

20

30

40

50

60

70

80

Life

time (

s)

0 100 200 300 400 500 600 700 800Target speed (ms)

Figure 11 Lifetime of the single agent migration algorithms inScenario 2

segment applications on the amount of communication inthe migration process To ensure the comprehensiveness ofthe evaluation set the number of communication componentagents as 1 and the number of subcomponent agents as 5 10or 20 and evaluate Mpid-DMLA based on code segmentswith different scales The second is the effect of differentenvironment and running platform in practical operation onthe amount of information and the success rate of migrationTherefore we only consider the relation of the number ofcommunication and energy instead of the influence of theagentrsquos size

The performance of the migration of mobile agent inAgilla system has been evaluated in [19] First Agilla systemis not designed to support multiple mobile agents specificallywhich means that it is not aware of the number of the agentsin the code segment application And the systemrsquos overheadsare related to the size ofmobile agent So inAgilla all kinds ofagents are considered to be identical and there is no discrim-ination between the main and communication componentagents However this does not affect the simulation resultsbecause all agents are uniformly assumed in the simulationin this paper

The purpose of multiple mobile agents migration isto complete the entire application migration with limitedenergy The default simulation parameters are shown inTable 2

And when main component agent recycles and sendsmobile agents suppose that the energy consumption of thenode where main component agent stays is neglected whilethat of other nodes is 005 J

Figures 12 13 and 14 illustrate the relation between thenumber of application code segments and the success rateof one hop migration when the numbers of subcomponentagents are 5 10 and 15 respectivelyThe success rate ofMpid-DMLA is higher when the number of subcomponent agentsis 5 than those when the numbers of subcomponent agents

International Journal of Distributed Sensor Networks 11

Table 2 Basic simulation algorithm parameters of a single mobileagent migration

Network area 100 times 100mNumber of nodes 7 times 7Number of main component agents per codesegment application 1

Number of subcomponent agents per code segmentapplication 5 10 or 15

Number of communication component agents percode segment application 1

Nodersquos initial energy 2 JEnergy consumption of a node when an agentreaches it 005 J

Interaction consumption when an agent migrates 01 J

20

30

40

50

60

70

80

90

100

Mpid-DMLAAgilla

2 25 3 35 45 554 5 6Number of migration applications

Succ

ess r

ate (

)

Figure 12 The relation between the number of code segmentapplications and the success rate of migration (I)

are 10 and 20 It is obvious that the successful rate of Mpid-DMLA is generally not less than 50 which is much largerthan that of Agilla

Moreover Figure 15 demonstrates the relationship be-tween themigration hops and the amount of communicationwhen the subcomponent agents are 5 As shown in Figure 15with the samemigration hops the amount of communicationof Mpid-DMLA is much less than that of Agilla As the num-ber of data packet reflects the amount of energy consumptionMpid-DMLA consumes less energy than Agilla

Although the simulation results show that Mpid-DMLAis a suitable migration scheme for multiple mobile agents itbases on some specific assumptions In general this schemestill encounters the following three issues

(1) When the main component agent loses efficacy inthe process of migration code segment applicationscannot be rebuilt

(2) When subcomponent agents lose efficacy afterdeployment code segment applications will keepwaiting because the applications cannot be recycled

20

30

40

50

60

70

80

90

100

Mpid-DMLAAgilla

2 25 3 35 45 554 5 6Number of migration applications

Succ

ess r

ate (

)

Figure 13 The relation between the number of code segmentapplications and the success rate of migration (II)

10

20

30

40

50

60

70

80

0

Mpid-DMLAAgilla

2 25 3 35 45 554 5 6Number of migration applications

Succ

ess r

ate (

)

Figure 14 The relation between the number of code segmentapplications and the success rate of migration (III)

1 15 2 25 3 35 420

40

60

80

100

120

140

160

180

200

Hops of main agent

Num

ber o

f dat

a pac

kets

Mpid-DMLAAgilla

Figure 15The relation between themigration hops and the amountof communication

12 International Journal of Distributed Sensor Networks

(3) If the main and the subcomponent agents which loseefficacy in issue 1 or 2 are left in the nodes withoutbeing recycled then othermobile agents cannot reachthese nodes

Therefore in designing middleware software it is nec-essary to add the feedback mechanism of main componentagent and it needs to periodically return ACK informationto the higher layer of the middleware software When themiddleware finds that the main component agent losesefficacy it needs to distribute a new one or a specializedindependent agent to recycle the useless agents in the wholenetwork

In Agilla the method of recycling the useless agentsis by adding the unified interface into all of the agentsRemoving all of the mobile agents will affect the function ofcode segment applications Therefore developing the moreappropriate middleware software is the future work whichis also related to the development of the better migrationstrategies [26]

7 Conclusion

As big data is the intensification of information architec-ture while distributed computing is the intensification ofinfrastructure and resource distributed computing is anappropriate method to deal with big data In this paperwe have firstly illustrated the mobile agent model in thedistributed computing And it led to the two considerablefactors of migration the selection of nodes and of thesequence of them Then we have proposed the assumptionsof the migration algorithms Moreover we have designedthe nodersquos sensor model and information collection modelFurthermore we have formulated the beacon frames ofmobile agent for communication Generally mobile agentmigration algorithms can be transferred as the target trackingissue Thus we have described a simple target localizationalgorithm applicable for migration On this basis we havegeneralized the whole migration process of the mobile agent

Considering the single mobile agent we have proposedthe static SMLA and the dynamic DMLA algorithms Afteranalyzing the drawbacks of SMLA and DMLA we thenproposed the pid-DMLA algorithm Section 4 has demon-strated the modeling procedure of the abovementioned threealgorithms Section 6 has emulated and evaluated their per-formances and studied their advantages and disadvantages indifferent scenarios

Considering the multiple mobile agents applied in prac-tical target tracking we have also put forward the Mpid-DMLA algorithm In the process of mobile agent componentconstructing code segment applications we defined threemobile agents with different functions to cooperate with eachother to accomplish the migration Section 5 illustrated thearchitecture modeling procedure and basic process ofMpid-DMLA Section 6 compared it with Agilla and evaluated itsadvantages and disadvantages

Acknowledgments

The paper is sponsored by the National Natural ScienceFoundation of China (61170065 61171053 61003039 6110019961003236 61202355 61201163 and 61373138) the Natu-ral Science Foundation of Jiangsu Province (BK2011755BK2011072 and BK2012436) the Scientific amp Technolog-ical Support Project of Jiangsu Province (BE2012183 andBE2012755) the Natural Science Key Fund for Collegesand Universities in Jiangsu Province (12KJA520002) theScientific Research amp Industry Promotion Project for HigherEducation Institutions (JHB2012-7) the Postdoctoral Foun-dation (2012M511753 1101011B 1202034C and 2013T60536)the Doctoral Fund of Ministry of Education of China(20113223110002 and 20123223120006) the Fund of NanjingUniversity of Posts and Telecommunications (NY212047)and the project funded by the Priority Academic Pro-gram Development of Jiangsu Higher Education Institutions(yx002001)

References

[1] Research Trends special issue on big data httpwwwresearch-trendscomwp-contentuploads201209Research Trends Is-sue30pdf 2012

[2] L Ismail and D Hagimont ldquoA performance evaluation of themobile agent paradigmrdquoACM SIGPLANNotices vol 34 no 10pp 306ndash313 1999

[3] AMainwaring J Polastre R Szewczyk D Culler and J Ander-son ldquoWireless sensor networks for habitat monitoringrdquo inProceedings of the 1st ACM International Workshop on WirelessSensor Networks and Applications pp 88ndash97 September 2002

[4] IVY-A sensor network infrastructure for the college of engi-neering httpwww-bsaceecsberkeleyeduprojectsivy 2008

[5] I F AkyildizW Su Y Sankarasubramaniam and E Cayirci ldquoAsurvey on sensor networksrdquo IEEE Communications Magazinevol 40 no 8 pp 102ndash105 2002

[6] Y-C Tseng S-P Kuo H-W Lee and C-F Huang ldquoLocationtracking in a wireless sensor network by mobile agents and itsdata fusion strategiesrdquoComputer Journal vol 47 no 4 pp 448ndash460 2004

[7] Y Xu and H Qi ldquoMobile agent migration modeling anddesign for target tracking in wireless sensor networksrdquo Ad HocNetworks vol 6 no 1 pp 1ndash16 2008

[8] Y Xu and H Qi ldquoDistributed computing paradigms for collab-orative signal and information processing in sensor networksrdquoJournal of Parallel and Distributed Computing vol 64 no 8 pp945ndash959 2004

[9] F Zhao J Shin and J Reich ldquoInformation-driven dynamicsensor collaborationrdquo IEEE Signal Processing Magazine vol 19no 2 pp 61ndash72 2002

[10] J Liu J Reich and F Zhao ldquoCollaborative in-network pro-cessing for target trackingrdquo Eurasip Journal on Applied SignalProcessing vol 2003 no 4 pp 378ndash391 2003

[11] J-W Kim J-S In K Hur J-W Kim and D-S Eom ldquoAnintelligent agent-based routing structure for mobile sinks inWSNsrdquo IEEE Transactions on Consumer Electronics vol 56 no4 pp 2310ndash2316 2010

[12] H Wang G Pottie K Yao and D Estrin ldquoEntropy-basedsensor selection heuristic for target localizationrdquo in Proceedings

International Journal of Distributed Sensor Networks 13

of the 3rd International Symposium on Information Processing inSensor Networks (IPSN rsquo04) pp 36ndash45 ACM Press April 2004

[13] T S RappaportWireless Communications Principles and Prac-tice IEEE Press 1996

[14] H Mousannif and I Khalil ldquoCooperative data management inwireless sensor networksrdquo International Journal of Computa-tional Intelligence Systems vol 5 no 3 pp 403ndash412 2012

[15] N Labraoui M Guerroui M Aliouat and T Zia ldquoDataaggregation security challenge in wireless sensor networks asurveyrdquo Ad-Hoc and Sensor Wireless Networks vol 12 no 3-4pp 295ndash324 2011

[16] M Chen L T Yang T Kwon L Zhou and M Jo ldquoItineraryplanning for energy-efficient agent communications in wirelesssensor networksrdquo IEEE Transactions on Vehicular Technologyvol 60 no 7 pp 3290ndash3299 2011

[17] Q Wu N S V Rao J Barhen et al ldquoOn computing mobileagent routes for data fusion in distributed sensor networksrdquoIEEE Transactions on Knowledge and Data Engineering vol 16no 6 pp 740ndash753 2004

[18] C-L Fok G-C Roman and C Lu ldquoMobile agent middlewarefor sensor networks an application case studyrdquo in Proceedingsof the 4th International Symposium on Information Processing inSensor Networks (IPSN rsquo05) pp 382ndash387 April 2005

[19] A Boulis C Han andM B Srivastava ldquoDesign and implemen-tation of a framework for efficient and programmable sensornetworksrdquo in Proceedings of the 1st international conference onMobile systems applications and services ACM San FranciscoCalif USA 2003

[20] P Levis and D Culler ldquoMate a tiny virtual machine for sensornetworksrdquo ACM Sigplan Notices vol 37 no 10 pp 85ndash95 2002

[21] S Suenaga N Yoshioka and S Honiden ldquoGroup migration bymobile agents in wireless sensor networksrdquo Computer Journalvol 54 no 3 pp 345ndash355 2011

[22] P Bellavista Y Cai and T Magedanz ldquoRecent advances inmobile middleware for wireless systems and servicesrdquo MobileNetworks and Applications vol 16 no 3 pp 267ndash269 2011

[23] Z Zhirou and C Wengang ldquoModel and optimization ofmobile agentrsquos migration in gridrdquo in Proceedings of the 2ndInternational Conference on Bio-Inspired Computing Theoriesand Applications (BICTA rsquo07) pp 127ndash130 September 2007

[24] J-P Zhang M Jun J Yang and L-L Cheng ldquoAn improvedmigration strategy of mobile agentrdquo in Proceedings of the4th International Conference on Fuzzy Systems and KnowledgeDiscovery (FSKD rsquo07) pp 577ndash581 August 2007

[25] Y Lee and K Kim ldquoOptimal migration path searching usingpath adjustment and reassignment for mobile agentrdquo in Pro-ceedings of the 4th International Conference on NetworkedComputing and Advanced InformationManagement (NCM rsquo08)pp 564ndash569 Gyeongju South Korea September 2008

[26] M Tentori M Rodrıguez and J Favela Vara ldquoAn agent-basedmiddleware for the design of activity-aware applicationsrdquo IEEEIntelligent Systems vol 26 no 3 pp 15ndash23 2011

International Journal of

AerospaceEngineeringHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

RoboticsJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Active and Passive Electronic Components

Control Scienceand Engineering

Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

International Journal of

RotatingMachinery

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporation httpwwwhindawicom

Journal ofEngineeringVolume 2014

Submit your manuscripts athttpwwwhindawicom

VLSI Design

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Shock and Vibration

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Civil EngineeringAdvances in

Acoustics and VibrationAdvances in

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Electrical and Computer Engineering

Journal of

Advances inOptoElectronics

Hindawi Publishing Corporation httpwwwhindawicom

Volume 2014

The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014

SensorsJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Modelling amp Simulation in EngineeringHindawi Publishing Corporation httpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Chemical EngineeringInternational Journal of Antennas and

Propagation

International Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Navigation and Observation

International Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

DistributedSensor Networks

International Journal of

2 International Journal of Distributed Sensor Networks

illustrated above inWSNmobile agents dynamicallymigrateto the CIS with service request With CIS accomplishing thecomputing operations locally and returning the results to thenodes agents can be devoid of the transmission of big dataand the reliance on the network bandwidth which improvesthe efficiencyThe specific advantages of mobile agent modelespecially in terms of energy consumption reliability andlarge-scale distribution are declared in [7] However theconsequent challenges are not negligible like the high failurerate unreliable communication link high dependence onbandwidth and so on

2 Related Work

Although mobile-agent-based distributed model has excep-tional advantages in WSN [8] dynamic migration is neces-sary in practical deployment Dynamic migration presentedby migration table has two behaviors nodes selection andmigration sequences Therefore migration tablersquos efficiencyin some sense decides accuracy of data integrations timeof migrations energy consumption and performance of thewhole network

In response to the challenges embodied in the specialtyof WSN the design of mobile agentrsquos migration table needsto adequately consider the following three aspects Firstlyagent entity must be as small as possible to reduce migrat-ing energy consumption Secondly migration table has toguarantee rational precision of data integrations to obtainsufficient sensor nodes Thirdly in the precondition of thesecond aspect migration table needs to be short enough Inconclusion a suitable migration table should balance the costand precision

Thus in this paper we mainly concentrate on the designof the migration table which can reduce agentrsquos migrationenergy consumption while guaranteeing reliable migrationmethods

Briefly mobile agent migration has two categories staticmigration list algorithm (SMLA) and dynamic migrationlist algorithm (DMLA) In SMLA CIS analyzes the wholenetworkrsquos resources and designs migration paths that areinserted in agent entities initially However static SMLAcannot respond to the changes of the network timely thechanges happen when one or more links between nodesbreak while in DMLA migration table is updated at eachmigration interval according to the sensibility of the network

Without loss of generality mobile agent migration algo-rithms can be illustrated by target tracking Consideringmovements of the target entity Zhao et al proposed theinformation-driven-based tracking method [9] Liu et alproposed the collaborative in-networkmethod to track target[10] which effectively reduced overheads and energy con-sumption Kim et al transferred the distributed tracking issueto the continuous Bayesian estimation problem and adoptedthe information-driven-based sensor querying scheme toselect the next hop [11]Wang et al proposed a heuristic targetlocalization method based on entropy [12] which is moreefficient than the information-based computing methodIn this paper we will propose the dynamic migration list

algorithm based on the predicting information drive (pid-DMLA) to solve the motion problems of migrations of thetarget entity

Considering the condition of using multiple agents totrack an application in practice we will also propose thedynamic migration list algorithm based on the predictinginformation drive of multiple agents (Mpid-DMLA)

In this paper we would propose themigration algorithmsin different scenarios single mobile agent and multiplemobile agents which is valuable and applicable in WSNsThe rest of this paper is organized as follows Section 3describes the preliminary of migration Section 4 presentssingle-agent-based migration algorithms including SMLADMLA and pid-DMLA Section 5 illustrates Mpid-DMLASimulation results and analyses are presented in Section 6and conclusions are drawn in Section 7

3 Preliminary

31 Assumptions The following assumptions are considered

(1) The location of each sensor node is known(2) Each node has at least two neighbor nodes(3) The valid perception radius of sensor node is more

than half the distance between two neighboringnodes

(4) All nodes are synchronized(5) All sensors are deployed in a two-dimensional region(6) There is only one target entity in the region and it

moves uniformly and linearly

32 Evaluation Indicators of the Agent Migration AlgorithmIn order to evaluate the existing agent migration algorithmswe adopt the three indicators proposed by Xu et al [7] energyconsumption network lifetime and migration hops

Each migration step of mobile agent is called a hop Andthe energy consumption of each hop consists of three aspectstransmission consumption 119864

119886119887(119864119886119887

represents the energyconsumption of an agent migrating from node 119886 to node119887) initialization consumption 119864

119894(119864119894is constant) and data

processing consumption The network energy consumptionis primarily concentrated in transmissions

Amigration is primarily composed of four parts sendingan agent receiving an agent data processing and decidingthe next hop And each consumption time of the fourprocesses is fixed Thus evaluating the migration time of anagent is equivalent to calculating the migration hops

33 Nodersquos Sensor Model We adopt the following wirelesstransmission model [13] proposed by Rappaport

119911119896(119905) prop

119864

1003817100381710038171003817119883 (119905) minus 119883119896

1003817100381710038171003817

2 (1)

where 119911119896(119905) is a measure value of sensor 119896 related to the

location and time 119864 is the signal power of the target entity119883(119905) represents the location of the target entity in time

International Journal of Distributed Sensor Networks 3

A A AB B BD D

r1r1

r1

r2 r2r2

D1

D2

C C C

Figure 1 The target localization algorithm

119905 119883119896represents the location of node 119896 119883(119905) and 119883

119896are

two-dimensional vectors and 119883(119905) minus 119883119896 represents the

Euclidean distance between node 119896 and target entity intime 119905

34 Nodersquos Information CollectionModel In order to quantifythe data provided by sensors we need to define the amountof information According to (1) at any time 119911

119896(119905) is only

related to 119883(119905) minus 119883119896 Thus adopting the Gaussian model

the amount of information when the agent migrates to node119896 is the following

119868119896(119905) =

1

radic2120587120590

119890minus(119905)minus119883119896

2

21205902

(2)

where 120590 is the standard deviation of the amount of infor-mation which indicates the decrease of speed of 119868

119896(119905) when

the distance between sensor node and target entity increasesand 119883(119905) represents the location of the target entity which iscalculated by the target localization algorithm in Section 36

35 Beacon Frames We use beacons to termly exchangeinformation of the neighboring nodes A beacon frame con-sists of the following five fields 119883

119896represents the location of

node 119896 119890119896(119905) represents the residual energy of node 119896 in time

119905 119911119896(119905) represents the sensor measurement of node 119896 in time

119905 and the last two fields are sending time 119905119904and receiving time

119905119903 Beacon frames are used to provide information of location

andmeasurement to the target localization algorithm and theagentmigration algorithmMoreover they can be used to testthe survivability of neighbor nodes

36 Target Localization Algorithm As illustrated in (2) thelocation of target entity is calculated by the target localizationalgorithm At present target localization algorithms mostlyadopt the method of centroid weighting And the basic ideais to make nodersquos beacons take control of the centroidrsquoscoordinate and to reflect the general relations of the locationsby using weighting factors which influence the positions ofthe nodesrsquo centroids And according to the RSSIs of the threesensorsrsquo beacons it is easy to obtain the other nodersquos locationby the trilateration

As shown in Figure 1 the locations of 119860(119909119860 119910119860)

119861(119909119861 119910119861) and 119862(119909

119862 119910119862) are known and one of 119863(119909

119863 119910119863)

needs to be tested According to the RSSIs we can calculate

the distance of 119860119863 119861119863 and 119862119863 denoted by 1199031 1199032 and 119903

3

respectively Thus we can obtain the following equation set

(119909119861minus 119909119863)2

+ (119910119861minus 119910119863)2

= 1199032

2

(119909119860minus 119909119863)2

+ (119910119860minus 119910119863)2

= 1199032

1

(3)

Apparently the distance of 119860 and 119861 is less than or equalto 1199031+ 1199032 So (3) has two real solutions that are identical or

not which correspond to the coordinates of the two nodes1198631

and 1198632that are identical or not Select the one that is nearer

to 119862 as the approximate location of 119863 denoted as (1199091 1199101)

In the same way we can obtain the other two approximatelocations of119863 denoted as (119909

2 1199102) and (119909

3 1199103)Therefore the

coordinate of 119863 is (119909119863 119910119863) which can be calculated by the

following

119909119863

=1199091 (1199031+ 1199032) + 1199092 (1199032+ 1199033) + 1199093 (1199033+ 1199031)

1 (1199031+ 1199032) + 1 (119903

2+ 1199033) + 1 (119903

3+ 1199031)

119910119863

=1199101 (1199031+ 1199032) + 1199102 (1199032+ 1199033) + 1199103 (1199033+ 1199031)

1 (1199031+ 1199032) + 1 (119903

2+ 1199033) + 1 (119903

3+ 1199031)

(4)

As a result according to the previous target localizationalgorithm we can obtain 119909(119905)

4 Migration Algorithms Based on SingleMobile Agent

This section would propose SMLA and DMLA algorithmsfrom static and dynamic aspects respectively And it wouldthen propose pid-DMLA based on disadvantages of SMLAand DMLA

41 SMLA In SMLA algorithm we use static informa-tion Before distributing mobile agents the CIS collects bycommunicating with other nodes and it then generates amigration plan In thismigration plan information collectionvalues of all nodes in the network decreaseThe specific stepsof SMLA are as follows

Algorithm 1 (SMLA)(1) At time 119905 = 0 the node that has first detected

information of target entity is denoted as cluster headit generates a mobile agent

(2) At time 119905 = 0 calculate location information oftarget entity 119883(0) according to the target locationalgorithm

4 International Journal of Distributed Sensor Networks

Calculate the amount of information collection 1198680

according to (2) and set 119878119906119898119860119903119903119886119910 = 1198680

Select the next hop according to the predeterminedmigration sequence

(3) At time t when agent reaches node k the location oftarget entity is 119883(119905) according to the target locationalgorithmCalculate the amount of information collection ofnode 119896 119868

119896 and update 119878119906119898119860119903119903119886119910 = 119878119906119898119860119903119903119886119910 + 119868

119896

If 119878119906119898119860119903119903119886119910 is greater than the threshold then go tostep 4 otherwise continue to migrate to the next hopin the migration sequence update time 119905 = 119905 + 1 andthen repeat step 3

(4) Return to cluster head to process data

Though sometimes SMLA can gain the ideal migrationsequence there are many drawbacks in practical applica-tions Firstly since SMLA uses static information sensorrsquostransmission distance must be large enough to ensure thatCIS is within one-hop range So we need to adopt thenodes with greater power in communication capabilitiesAnd using these nodes will inevitably result in relativelyhigher energy consumption Secondly since SMLA is a staticalgorithm the static information which is acquired beforemigration lack timeliness and cannot reflect the changes ofthe network in the process of migrating For example thatsome nodes suddenly lose efficacy can result in the failure ofthe deployment of SMLA [9] Moreover most of WSNs arehighly dynamic so SMLA has some limitations in WSNs

42 DMLA SMLA is a centralized algorithm and it hasalready decided the migration sequence before distributingmobile agents Because WSN is a dynamic and distributednetwork static routing information exposes a lot of problemsin practical applications Thus DMLA is proposed DMLAdynamically selects the next hop in the process of migrationbased on characteristics of the network

421 The Migration Model of Mobile Agent in DMLA InDMLA mobile agents decide the nodes that they need tomigrate to In the surviving neighbor nodes not all of thenodes can provide information affecting the final resultTherefore the purpose ofmobile agentmigration algorithm isto produce a set of optimal subsets to determine the optimummigration sequence

In the target entity tracking we construct rules to choosethe next hop as followsmobile agent which is located in node119896 always looks for the next hop with minimal overheadsAs the selection of the next hop is a method of balancingoverheads and profit we define the overheads function whenmobile agent moves from node 119894 to neighbor node 119895 at time119905 as follows

119862119894119895(119905) = 119886

119864119894119895

119864max+ 119887(1 minus

119868119895(119905)

119868max) + radic1 minus 119886 minus 119887(1 minus

119890119895(119905)

119890max)

(5)

Equation (5) consists of three parts including energy con-sumption information collection and residual energy where119864119894119895represents transmission energy consumption when the

agent migrates directly from node 119894 to node 119895 119868119895(119905) represents

the amount of information collection of node 119895 at time 119905119890119895(119905) represents the residual energy of node 119895119864max represents

themaximumenergy consumption ofmigration between twonodes which is determined by the maximum transmissiondistance 119868max represents the maximum amount of informa-tion collection 119890max represents the initial energy of sensornode and it is a constant value and 119886 and 119887 are the balancecoefficients 0 le 119886 and 119887 le 1

Define the transmission energy consumption of the agentmigrating from node 119894 to node 119895 as follows

119864119894119895= 119875119894119895sdot 119879trans = 120572119875

119903-th sdot10038171003817100381710038171003817119883119894minus 119883119895

10038171003817100381710038171003817

2

sdot 119879trans (6)

where 119875119894119895represents the transmission power of sensor nodes

119883119894minus 119883119895 represents the distance between node 119894 and node

119895 119879trans a constant represents the time of transmitting themobile agent 120572 = 4120587

2

1205822

11986611198662 1198661 1198662are all of the

antenna gain factors (set them as 1) 120582 represents the signalwavelength and 119875

119903-th represents the receiving thresholdFrom (2) (5) and (6) we can gain

119862119894119895(119905) = 119886

10038171003817100381710038171003817119883119894minus 119883119895

10038171003817100381710038171003817

2

1198892max+ 119887

10038171003817100381710038171003817119883 (119905) minus 119883

119895

10038171003817100381710038171003817

2

1198891199052

max

+ radic1 minus 119886 minus 119887(1 minus

119890119895(119905)

119890max)

(7)

where 119889119905max represents the maximum possible distancebetween two nodes

Equation (7) can be rewritten as follows

119895lowast

= argmin

119886

10038171003817100381710038171003817119883119894minus 119883119895

10038171003817100381710038171003817

2

1198892max+ 119887

10038171003817100381710038171003817119883 (119905) minus 119883

119895

10038171003817100381710038171003817

2

1198891199052

max

+radic1 minus 119886 minus 119887(1 minus

119890119895(119905)

119890max)

119895 isin 119873119894

(8)

where 119873119894represents a group of neighbors of node 119894 node

119895lowast

isin 119873119894represents the next hop and 119895

lowast

isin 119873119894can ensure less

energy consumption stronger sensing capabilities and moreresidual energy

In summary the total cost in the migration process is

Cost =HOP119878minus1

sum

119901=0

119862119894119901119894119901+1

st 119878119906119898119860119903119903119886119910 =

HOP119878sum

119901=0

119868119901gt threshold

(9)

where 119901 represents the number of nodes migrated throughHOP119878represents the total number of migration hops In

the practical migration once mobile agent gains enoughinformation and it will terminate migration and return tothe CIS

International Journal of Distributed Sensor Networks 5

422The SpecificMigration Procedure of DMLA Themobileagent migrates to the alive node which has not been reachedand with the minimal migration overheads according to (7)In this way we can gain a table close to the optimalmigrationIf all nodes have been reached mobile agent will return thenode that distributes it And when the mobile agent hasacquired enough information it will return to the CIS

Since each node periodically receives beacon frames fromneighbor nodes DMLA can always use the updated localvariables to determine the next hop As only using localvariables sensor nodes can avoid the nodewhich has a greatertransmission distance so part of the transmission energycan be saved In addition when some nodes suddenly loseefficacy the whole migration process will not make mistakesSoDMLAalgorithm is a fault-tolerant algorithmThe specificmigration steps of DMLA are as follows

Algorithm 2 (DMLA)

(1) At time 119905 = 0 the node that has first detectedinformation of target entity is denoted as cluster headit generates a mobile agent

(2) At time 119905 = 0 calculate location information oftarget entity 119883(0) according to the target locationalgorithmCalculate the amount of information collection 119868

0

according to (2) and set 119878119906119898119860119903119903119886119910 = 1198680

Select the next hop according to (8)(3) At time 119905 when agent reaches node 119896 the location of

target entity is 119883(119905) according to the target locationalgorithmCalculate the amount of information collection ofnode 119896 119868

119896 and update 119878119906119898119860119903119903119886119910 = 119878119906119898119860119903119903119886119910 + 119868

119896

If 119878119906119898119860119903119903119886119910 is greater than the threshold then go tostep 4 otherwise continue to migrate to the next hopaccording to (8) update time 119905 = 119905+1 and then repeatstep 3

(4) Return to cluster head to process data

43 pid-DMLA

431 The Proposition of pid-DMLA In the cooperationprocessing applications the movement of target entity isone of the important factors that affect the final result [14ndash19] DMLA algorithm does not take the direction of themovement of target entity into account So in some cases itcannot select the optimumnext hop For example in Figure 2at time 119905 the mobile agent is located in node 119860 Nodes 119861

and 119862 are neighbors of 119860 and they are equidistant to 119860According to DMLA mobile agent calculates node 119861 as thenext hop This is because node 119861 is nearer to the targetentity than node 119862 Then at time 119905 + 1 the mobile agent islocated in node 119861 But the target entity moves to a positioncloser to node 119862 Thus DMLA on the contrary selectsthe node which is farther from the target and has smalleramount of information collection and measurement value

t t + 1

A A

B B

CC

Figure 2 The movement of the target entity

as the next hop Therefore when designing agent migrationalgorithms we need to focus on predicting the movement ofthe target entity and selecting the node with the maximuminformation And we will propose the pid-DMLA algorithmto solve the problems of the movement of the target entity inthe migration process

432 Agent Migration Model in pid-DMLA Suppose thatthe moving direction and speed of the target entity do notchange Then the change of the target entityrsquos displacementis a constant at each identical time interval as follows

119883 (119905 + 1) minus 119883 (119905) = 119883 (119905) minus 119883 (119905 minus 1) (10)

Equivalently

119883 (119905 + 1) = 2119883 (119905) minus 119883 (119905 minus 1) (11)

As illustrated in (11) the position information119883(119905 + 1) attime 119905 + 1 can be deduced from the positions at time 119905 minus 1 andtime 119905 In this way the location information119883(119905minus1) and119883(119905)

can be calculated by target location algorithm and accordingto (11) the position of target entity119883(119905+1) can be estimatedat time 119905 + 1 as follows

119883 (119905 + 1) = 2119883 (119905) minus 119883 (119905 minus 1) (12)

Update the overhead function in (7) and get

119862119894119895(119905) = 119886

10038171003817100381710038171003817119883119894minus 119883119895

10038171003817100381710038171003817

2

1198892max+ 119887

10038171003817100381710038171003817119883 (119905 + 1) minus 119883

119895

10038171003817100381710038171003817

2

1198891199052max

+ radic1 minus 119886 minus 119887(1 minus

119890119895(119905)

119890max)

(13)

where 119883(119905 + 1) minus 119883119895 represents the distance between node

119895 and the target entity at time 119905 + 1And the next hop is

119895lowast

= argmin

119886

10038171003817100381710038171003817119883119894minus 119883119895

10038171003817100381710038171003817

2

1198892max+ 119887

10038171003817100381710038171003817119883 (119905 + 1) minus 119883

119895

10038171003817100381710038171003817

2

1198891199052max

+radic1 minus 119886 minus 119887(1 minus

119890119895(119905)

119890max)

119895 isin 119873119894

(14)

6 International Journal of Distributed Sensor Networks

t

t

t + 1

t + 1

t + 2

t + 2

t minus 1

t minus 1

AB DC

Figure 3 The migration of pid-DMLA

At time 119905 = 0 since the position of the target entity atthe previous time is unknown the amount of informationcollection can be calculated by the DMLA At time 119905 gt 0mobile agent migrates to the node closer to the estimatedposition of the target entity So it can obtain larger sensingmeasurement and information collection which is the coreof pid-DMLA

433 The Specific Migration Procedure of pid-DMLA Theprocess of mobile agent migration of pid-DMLA is shown inFigure 3 In a period of time the target entity moves alongthe dotted line and mobile agent migrates from node 119860 to119863

through 119861 and 119862 As shown in Figure 3 mobile agent alwaysmigrates to the node nearest to the next position of the targetentity

The specific migration procedure of pid-DMLA is asfollows

Algorithm 3 (pid-DMLA)

(1) At time 119905 = 0 the node that has first detectedinformation of target entity is denoted as cluster head

it generates a mobile agent

(2) At time 119905 = 0 calculate location information oftarget entity 119883(0) according to the target locationalgorithm

Calculate the amount of information collection 1198680

according to (2) and set 119878119906119898119860119903119903119886119910 = 1198680

Select the next hop according to (14)

(3) At time 119905 when agent reaches node 119896 the location oftarget entity is 119883(119905) according to the target locationalgorithm

Calculate the amount of information collection ofnode 119896 119868

119896 and update 119878119906119898119860119903119903119886119910 = 119878119906119898119860119903119903119886119910 + 119868

119896

If 119878119906119898119860119903119903119886119910 is greater than the threshold then goto step 4 otherwise continue to migrate to the nexthop according to (14) update time 119905 = 119905 + 1 and thenrepeat step 3

(4) Return to cluster head to process data

5 Migration Algorithm Based on MultipleMobile Agents

51 The Architecture of Mpid-DMLA Many researchers haveproposed and realized feasible ways to organize themigrationof a single mobile agent such as Agilla program [18] andSensorWare program [19] Consider the functional require-ments of these programs Firstly they have not formed anarchitecture to manage the agent migration mechanismsin WSN Secondly they lack migration algorithms whena tracking application needs the collaboration of multipleagents So when dealing with multiple agentsrsquo migration weneed a specific agent to manage the applications

Therefore we will propose Mpid-DMLA to solve thisproblem

Suppose that multiple target entity applications aredeployed in an area of sensors each application is responsiblefor monitoring a part of the area and composed of oneor more mobile agents Therefore we have the followingdefinitions

Definition 4 (code segment application) A centralized pro-gram which consists of multiple mobile agents deployed inthe WSN to accomplish a specific task is called centralizedcode segment application of multiple agents which is calledcode segment application for short

Code segment applications are associated with sensingdata that need to be evaluated and they need to cooperate andinteract based on the need of programs On the other handaccording to the Deluge amp Mate model [20] code segmentapplications need to redeploy themselves so that they can bedistributed by the CIS

To track targets code segment applications require thefollowing four functions [21]

(1) Sensing Sensor nodes are distributed around thetarget node and they test and collect information ofthe surrounding environment

(2) Processing Information After acquiring sensor infor-mation most of the nodes send it to the CIS forprocessing while cluster heads will directly processinformation

(3) Communication In order tomaintain the applicationsensor nodes need to keep exchanging informationwith other nodes or CIS

(4) Migration Mobile agents migrate based on the needof code segment application

Constructing the architecture of multiple mobile agentsalso requires the following definitions

Definition 5 (main component agent) The main componentagent is responsible for managingmigration information andactivity cycle of all component agents processing sensingdata of other agents and collecting and exchanging theinformation of subcomponent agents Each code segmentapplication has one and only main component agent

International Journal of Distributed Sensor Networks 7

Definition 6 (subcomponent agent) The subcomponentagent is used to obtain the data of nodes and report them tothe main component agent Each code segment applicationhas one or more subcomponent agents

Definition 7 (communication component agent) The com-munication component agent is used to collect and exchangethe entity information of WSN by migrating Each codesegment application has one or more communication com-ponent agents

Definition 8 (middleware software) The middleware soft-ware located between the physical and the operating systemlayers is the software entity running in mobile agents

52 The Migration Model of Mobile Agent in Mpid-DMLAWhen a code segment application changes monitoring posi-tions all of the mobile agents in this application needto migrate together to reconstruct a new code segmentapplication In Agilla the main component agent controlsthe entire process by sending the migration sequence whileother mobile agents can communicate by accessing theindependent tuple space [22] And the engine virtualmachineworks as the kernel to provide services and controls all of theconcurrent execution of all of the agents in a node Howeverwith the increase of the mobile agents the possibilities ofthe loss of packet and the failure of migration will increasetoo Suppose that a code segment application is composedof 119873 mobile agents Define 119875 as the failure coefficient of asingle agent And then we can obtain the success coefficientof migration in the entire application as follows

119875success = (1 minus 119875)119873

(15)

Communication overheads based on pid-DMLA aregiven as follows

119862119894119895(119905) = [

[

119886

10038171003817100381710038171003817119883119894minus 119883119895

10038171003817100381710038171003817

2

1198892max+ 119887

10038171003817100381710038171003817119883 (119905 + 1) minus 119883

119895

10038171003817100381710038171003817

2

1198891199052max

+radic1 minus 119886 minus 119887(1 minus

119890119895(119905)

119890max)] times (119875success)

minus1

(16)

And the next hop is

119895lowast

= argmin

[

[

119886

10038171003817100381710038171003817119883119894minus 119883119895

10038171003817100381710038171003817

2

1198892max+ 119887

10038171003817100381710038171003817119883 (119905 + 1) minus 119883

119895

10038171003817100381710038171003817

2

1198891199052max

+radic1 minus 119886 minus 119887(1 minus

119890119895(119905)

119890max)]

]

times (119875success)minus1

(17)

53 The Specific Migration Procedure of Mpid-DMLA In theconstruction of the target entity tracking we assume thatsubcomponent agent is a small program of lightweight codewhich is suitable for migration And it can independently

Figure 4 Migration based on predicting information drive ofmultiple mobile agents

access the target entityrsquos location As shown in Figure 4hollow circles represent sensor nodes which are distributedin the sensing region code segment is composed of one maincomponent agent (solid circle) four subcomponent agents(solid squares) and one communication component agent(solid triangle) Each agent runs on one node executingcode segment applications The migration of code segmentapplication means the migration of the six mobile agentswhich is the migration of multiple mobile agents [19] Thefollowing algorithm can be used to describe the above-mentioned procedure

Algorithm 9 (Mpid-DMLA)

(1) A main component agent a communication com-ponent agent and several subcomponent agents aredeployed in different sensor nodes

(2) Sensor nodes are monitoring around the target entityThe node that has first detected information of targetentity is denoted as cluster head themain componentagent then migrates to it

(3) At time 119905 = 0 calculate location information oftarget entity 119883(0) according to the target locationalgorithmCalculate the amount of information collection 119868

0

according to (2)Define the threshold of the amount of information as119879ℎ119903119890119904ℎ119900119897119889 119880119875 and let 119879ℎ119903119890119904ℎ119900119897119889 119880119875 = 119868

0

The main component agent generates new com-munication component agent and transfers 119868

0and

119879ℎ119903119890119904ℎ119900119897119889 119880119875 to itThen the communication component agent sets tomigrate

(4) At time 119905 when agent reaches node 119896 the location oftarget entity is 119883(119905) according to the target locationalgorithmCalculate the amount of information collectionof node 119896 119868

119896 and update 119879ℎ119903119890119904ℎ119900119897119889 119880119875 =

119879ℎ119903119890119904ℎ119900119897119889 119880119875 + 119868119896

If 119879ℎ119903119890119904ℎ119900119897119889 119880119875 is greater than the threshold thengo to step 5 otherwise select the next hop formigration according to (17) update time 119905 = 119905 + 1and then repeat step 3

8 International Journal of Distributed Sensor Networks

t

t

t + 1

t + 1

t + 3 t + 4 t + 5 t + 6 t + 7t + 2

t + 3

t + 4

t + 5

t + 6

t + 7

t + 2

Figure 5 The migration process of SMLA

t t + 1 t + 3 t + 4 t + 5 t + 6 t + 7t + 2

tt + 1 t + 3

t + 4

t + 5

t + 6t + 7

t + 2

Figure 6 The migration process of DMLA

(5) Communication component agent stores the infor-mation of the migration path (119868

0(119905 = 0) 119868

1(119905 =

1) 119868119896(119905 = 119896) 119883 (119905 = 0) 119883 (119905 = 1) 119883 (119905 =

119896)) and other relevant information

And it interacts with the main component agent

(6) The main component agent generates the migrationsequence 119873

1 1198732 119873

119896 according to location of

target entity

Send it to each subcomponent agent (this step isoptional for different applications)

(7) The main component agent sends complete instruc-tions to the lower-layered middleware software

Method functions which are predefined in the mid-dleware software recycle the communication compo-nent agent and the subcomponent agents

(8) The main component agent goes to the destinationnode based on the migration sequence

If the migration fails send failure information toCIS to terminate this migration otherwise generatenew communication component agent and subcom-ponent agents and this migration ends

6 Simulation and Analyses

61 The Effect of Single Mobile Agent in Target TrackingBefore the simulation experiments we firstly use targettracking as a sample to demonstrate the migration effect ofa single mobile agent of different migration algorithms Asshown in Figures 5 6 and 7 suppose that a car is runningfrom left to right through the area covered byWSNThe smallcircle represents the sensor node solid small circle representsthe cluster head which distributes agents large dotted circlerepresents the communication range of sensor node and eachnode selects the next hop only within its communicationrange The above-mentioned figures illustrate the specificlocations of the car and agent from time 119905 to time 119905 + 7

The migration process of SMLA is shown in Figure 5 Attime 119905 CIS distributes the mobile agent to the node whichis the nearest to the target entity according to the prede-termined migration sequence When the object is stationaryor its motion can be estimated this static approach is verysuitable However it loses efficacy in some conditions Forexample if the object remains in the location at time 119905+1 thenthe migration sequence will be the positions that are aroundthe target entity But once the target entity starts to move notaccording to the predetermination SMLA will fail

DMLA can in some sense optimize the performanceof SMLA As shown in Figure 6 mobile agent tracks themovement of the target entity However the movement of

International Journal of Distributed Sensor Networks 9

t

t + 1 t + 3

t + 4

t + 5

t + 6t + 7

t + 2

t t + 1 t + 3 t + 4 t + 5 t + 6 t + 7t + 2

Figure 7 The migration process of pid-DMLA

Table 1 Basic simulation parameters of single mobile agent migra-tion algorithms

Network area 500 times 500mNumber of nodes 200 or 800Nodersquos initial energy 50 JNodersquos sensing radius 5mEnergy consumption for data processing 03 JEnergy consumption for sending beacon frames 01 JThe interval of node broadcasting beacon frames 01 sThe speed of the target entity 0sim50msThe threshold of the amount of information 18The size of a mobile agent 800 bitsSimulation time 100 s

the target entity is unpredictable so it is not accurate to decidemigration path only by the amount of information which ishowever related to the distance to the target entity

The pid-DMLA can predict movement information oftarget entity dynamically which is a better solution to suchproblems As shown in Figure 7 at time 119905 + 5 mobile agentselects the next hop according to the predicted informationwhich is different from DMLA In brief pid-DMLA rebuildsmigration sequence based on predicting themovement of thetarget entity to achieve the purpose of tracking

62 Performance Analysis on the Single Mobile Agent Migra-tion Algorithms We use simulation software NS2 to eval-uate the three single mobile agent migration algorithmsrsquoperformances The default simulation parameters are shownin Table 1

Suppose that once the agent migrates one hop all sensornodes in the network consume 02 J For SMLA the transmis-sion distance is set to 30m while for DMLA and pid-DMLAthe transmission distance is set to be 10m Therefore theconsumption of transmission for SMLA is three times morethan that of sendingThe reason for this setting is that SMLAneeds bigger transmission range to collect information Thedirection and size of movement of the target entity areconstant denoted as V (ms) Once the target entity reachesthe edge of the sensing area it reverses direction immediatelywhich ensures its being always within the sensing area

0 5 10 15 20 25 30 35 40 45 500

2

4

6

8

10

12

Target speed (ms)

Ener

gy co

nsum

ptio

n (J

)

SMLADMLApid-DMLA

Figure 8 Energy consumption of the single agent migrationalgorithms in Scenario 1

Scenario 1 Set the number of nodes as 200 the target entityspeed as 0sim50ms and the simulation time as 100 s whichcan guarantee a complete migration The simulation resultsare shown in Figures 8 and 9 It is obvious that DMLA andpid-DMLA consume less energy than SMLA in the wholenetwork This is because the communication cost of SMLAis higher than that of DMLA But with the increasing speedof the target entity the effect of SMLA algorithm in trackingthe target entity becomes less and less obvious thereforemigration may terminate and the energy consumption onthe contrary becomes slower On the other hand when thespeed of target entity is very low there are no significantdifferences among the three migration schemes

Scenario 2 Set the speed of target entity as 5ms and thenumber of nodes as 800 Generally the coverage and thenumber of nodes are positively correlated The simulationresults are shown in Figures 10 and 11 It is clear that pid-DMLA has the lowest energy consumption when the numberof nodes is more than 100 This is because with the increaseof the nodes the choices of predicting the next hop become

10 International Journal of Distributed Sensor Networks

0102030405060708090

100

Life

time (

s)

0 5 10 15 20 25 30 35 40 45 50Target speed (ms)

SMLADMLApid-DMLA

Figure 9 Lifetime of the single agent migration algorithms inScenario 1

Ener

gy co

nsum

ptio

n (J

)

SMLADMLApid-DMLA

0 100 200 300 400 500 600 700 8000

1

2

3

4

5

6

7

8

9

Target speed (ms)

Figure 10 Energy consumption of the single agent migrationalgorithms in Scenario 2

more as a result the overheads become less [23] When thenumber of nodes is less than 50 the performance of SMLAis better than those of the other two This is because withless nodes static and dynamic migration algorithms share themore similar migration path however the migration tablehas been predetermined in SMLA thus it on the contraryhas the least overheads On the other hand when the densityof nodes is high there would be more neighbors and moremigration choices Thus DMLA and pid-DMLA outperformSMLA

63 Multiple Mobile Agents Migration Algorithm PerformanceAnalysis Before the assessment of Mpid-DMLA we need toconsider two problems [24 25] The first is the effect of code

SMLADMLApid-DMLA

10

20

30

40

50

60

70

80

Life

time (

s)

0 100 200 300 400 500 600 700 800Target speed (ms)

Figure 11 Lifetime of the single agent migration algorithms inScenario 2

segment applications on the amount of communication inthe migration process To ensure the comprehensiveness ofthe evaluation set the number of communication componentagents as 1 and the number of subcomponent agents as 5 10or 20 and evaluate Mpid-DMLA based on code segmentswith different scales The second is the effect of differentenvironment and running platform in practical operation onthe amount of information and the success rate of migrationTherefore we only consider the relation of the number ofcommunication and energy instead of the influence of theagentrsquos size

The performance of the migration of mobile agent inAgilla system has been evaluated in [19] First Agilla systemis not designed to support multiple mobile agents specificallywhich means that it is not aware of the number of the agentsin the code segment application And the systemrsquos overheadsare related to the size ofmobile agent So inAgilla all kinds ofagents are considered to be identical and there is no discrim-ination between the main and communication componentagents However this does not affect the simulation resultsbecause all agents are uniformly assumed in the simulationin this paper

The purpose of multiple mobile agents migration isto complete the entire application migration with limitedenergy The default simulation parameters are shown inTable 2

And when main component agent recycles and sendsmobile agents suppose that the energy consumption of thenode where main component agent stays is neglected whilethat of other nodes is 005 J

Figures 12 13 and 14 illustrate the relation between thenumber of application code segments and the success rateof one hop migration when the numbers of subcomponentagents are 5 10 and 15 respectivelyThe success rate ofMpid-DMLA is higher when the number of subcomponent agentsis 5 than those when the numbers of subcomponent agents

International Journal of Distributed Sensor Networks 11

Table 2 Basic simulation algorithm parameters of a single mobileagent migration

Network area 100 times 100mNumber of nodes 7 times 7Number of main component agents per codesegment application 1

Number of subcomponent agents per code segmentapplication 5 10 or 15

Number of communication component agents percode segment application 1

Nodersquos initial energy 2 JEnergy consumption of a node when an agentreaches it 005 J

Interaction consumption when an agent migrates 01 J

20

30

40

50

60

70

80

90

100

Mpid-DMLAAgilla

2 25 3 35 45 554 5 6Number of migration applications

Succ

ess r

ate (

)

Figure 12 The relation between the number of code segmentapplications and the success rate of migration (I)

are 10 and 20 It is obvious that the successful rate of Mpid-DMLA is generally not less than 50 which is much largerthan that of Agilla

Moreover Figure 15 demonstrates the relationship be-tween themigration hops and the amount of communicationwhen the subcomponent agents are 5 As shown in Figure 15with the samemigration hops the amount of communicationof Mpid-DMLA is much less than that of Agilla As the num-ber of data packet reflects the amount of energy consumptionMpid-DMLA consumes less energy than Agilla

Although the simulation results show that Mpid-DMLAis a suitable migration scheme for multiple mobile agents itbases on some specific assumptions In general this schemestill encounters the following three issues

(1) When the main component agent loses efficacy inthe process of migration code segment applicationscannot be rebuilt

(2) When subcomponent agents lose efficacy afterdeployment code segment applications will keepwaiting because the applications cannot be recycled

20

30

40

50

60

70

80

90

100

Mpid-DMLAAgilla

2 25 3 35 45 554 5 6Number of migration applications

Succ

ess r

ate (

)

Figure 13 The relation between the number of code segmentapplications and the success rate of migration (II)

10

20

30

40

50

60

70

80

0

Mpid-DMLAAgilla

2 25 3 35 45 554 5 6Number of migration applications

Succ

ess r

ate (

)

Figure 14 The relation between the number of code segmentapplications and the success rate of migration (III)

1 15 2 25 3 35 420

40

60

80

100

120

140

160

180

200

Hops of main agent

Num

ber o

f dat

a pac

kets

Mpid-DMLAAgilla

Figure 15The relation between themigration hops and the amountof communication

12 International Journal of Distributed Sensor Networks

(3) If the main and the subcomponent agents which loseefficacy in issue 1 or 2 are left in the nodes withoutbeing recycled then othermobile agents cannot reachthese nodes

Therefore in designing middleware software it is nec-essary to add the feedback mechanism of main componentagent and it needs to periodically return ACK informationto the higher layer of the middleware software When themiddleware finds that the main component agent losesefficacy it needs to distribute a new one or a specializedindependent agent to recycle the useless agents in the wholenetwork

In Agilla the method of recycling the useless agentsis by adding the unified interface into all of the agentsRemoving all of the mobile agents will affect the function ofcode segment applications Therefore developing the moreappropriate middleware software is the future work whichis also related to the development of the better migrationstrategies [26]

7 Conclusion

As big data is the intensification of information architec-ture while distributed computing is the intensification ofinfrastructure and resource distributed computing is anappropriate method to deal with big data In this paperwe have firstly illustrated the mobile agent model in thedistributed computing And it led to the two considerablefactors of migration the selection of nodes and of thesequence of them Then we have proposed the assumptionsof the migration algorithms Moreover we have designedthe nodersquos sensor model and information collection modelFurthermore we have formulated the beacon frames ofmobile agent for communication Generally mobile agentmigration algorithms can be transferred as the target trackingissue Thus we have described a simple target localizationalgorithm applicable for migration On this basis we havegeneralized the whole migration process of the mobile agent

Considering the single mobile agent we have proposedthe static SMLA and the dynamic DMLA algorithms Afteranalyzing the drawbacks of SMLA and DMLA we thenproposed the pid-DMLA algorithm Section 4 has demon-strated the modeling procedure of the abovementioned threealgorithms Section 6 has emulated and evaluated their per-formances and studied their advantages and disadvantages indifferent scenarios

Considering the multiple mobile agents applied in prac-tical target tracking we have also put forward the Mpid-DMLA algorithm In the process of mobile agent componentconstructing code segment applications we defined threemobile agents with different functions to cooperate with eachother to accomplish the migration Section 5 illustrated thearchitecture modeling procedure and basic process ofMpid-DMLA Section 6 compared it with Agilla and evaluated itsadvantages and disadvantages

Acknowledgments

The paper is sponsored by the National Natural ScienceFoundation of China (61170065 61171053 61003039 6110019961003236 61202355 61201163 and 61373138) the Natu-ral Science Foundation of Jiangsu Province (BK2011755BK2011072 and BK2012436) the Scientific amp Technolog-ical Support Project of Jiangsu Province (BE2012183 andBE2012755) the Natural Science Key Fund for Collegesand Universities in Jiangsu Province (12KJA520002) theScientific Research amp Industry Promotion Project for HigherEducation Institutions (JHB2012-7) the Postdoctoral Foun-dation (2012M511753 1101011B 1202034C and 2013T60536)the Doctoral Fund of Ministry of Education of China(20113223110002 and 20123223120006) the Fund of NanjingUniversity of Posts and Telecommunications (NY212047)and the project funded by the Priority Academic Pro-gram Development of Jiangsu Higher Education Institutions(yx002001)

References

[1] Research Trends special issue on big data httpwwwresearch-trendscomwp-contentuploads201209Research Trends Is-sue30pdf 2012

[2] L Ismail and D Hagimont ldquoA performance evaluation of themobile agent paradigmrdquoACM SIGPLANNotices vol 34 no 10pp 306ndash313 1999

[3] AMainwaring J Polastre R Szewczyk D Culler and J Ander-son ldquoWireless sensor networks for habitat monitoringrdquo inProceedings of the 1st ACM International Workshop on WirelessSensor Networks and Applications pp 88ndash97 September 2002

[4] IVY-A sensor network infrastructure for the college of engi-neering httpwww-bsaceecsberkeleyeduprojectsivy 2008

[5] I F AkyildizW Su Y Sankarasubramaniam and E Cayirci ldquoAsurvey on sensor networksrdquo IEEE Communications Magazinevol 40 no 8 pp 102ndash105 2002

[6] Y-C Tseng S-P Kuo H-W Lee and C-F Huang ldquoLocationtracking in a wireless sensor network by mobile agents and itsdata fusion strategiesrdquoComputer Journal vol 47 no 4 pp 448ndash460 2004

[7] Y Xu and H Qi ldquoMobile agent migration modeling anddesign for target tracking in wireless sensor networksrdquo Ad HocNetworks vol 6 no 1 pp 1ndash16 2008

[8] Y Xu and H Qi ldquoDistributed computing paradigms for collab-orative signal and information processing in sensor networksrdquoJournal of Parallel and Distributed Computing vol 64 no 8 pp945ndash959 2004

[9] F Zhao J Shin and J Reich ldquoInformation-driven dynamicsensor collaborationrdquo IEEE Signal Processing Magazine vol 19no 2 pp 61ndash72 2002

[10] J Liu J Reich and F Zhao ldquoCollaborative in-network pro-cessing for target trackingrdquo Eurasip Journal on Applied SignalProcessing vol 2003 no 4 pp 378ndash391 2003

[11] J-W Kim J-S In K Hur J-W Kim and D-S Eom ldquoAnintelligent agent-based routing structure for mobile sinks inWSNsrdquo IEEE Transactions on Consumer Electronics vol 56 no4 pp 2310ndash2316 2010

[12] H Wang G Pottie K Yao and D Estrin ldquoEntropy-basedsensor selection heuristic for target localizationrdquo in Proceedings

International Journal of Distributed Sensor Networks 13

of the 3rd International Symposium on Information Processing inSensor Networks (IPSN rsquo04) pp 36ndash45 ACM Press April 2004

[13] T S RappaportWireless Communications Principles and Prac-tice IEEE Press 1996

[14] H Mousannif and I Khalil ldquoCooperative data management inwireless sensor networksrdquo International Journal of Computa-tional Intelligence Systems vol 5 no 3 pp 403ndash412 2012

[15] N Labraoui M Guerroui M Aliouat and T Zia ldquoDataaggregation security challenge in wireless sensor networks asurveyrdquo Ad-Hoc and Sensor Wireless Networks vol 12 no 3-4pp 295ndash324 2011

[16] M Chen L T Yang T Kwon L Zhou and M Jo ldquoItineraryplanning for energy-efficient agent communications in wirelesssensor networksrdquo IEEE Transactions on Vehicular Technologyvol 60 no 7 pp 3290ndash3299 2011

[17] Q Wu N S V Rao J Barhen et al ldquoOn computing mobileagent routes for data fusion in distributed sensor networksrdquoIEEE Transactions on Knowledge and Data Engineering vol 16no 6 pp 740ndash753 2004

[18] C-L Fok G-C Roman and C Lu ldquoMobile agent middlewarefor sensor networks an application case studyrdquo in Proceedingsof the 4th International Symposium on Information Processing inSensor Networks (IPSN rsquo05) pp 382ndash387 April 2005

[19] A Boulis C Han andM B Srivastava ldquoDesign and implemen-tation of a framework for efficient and programmable sensornetworksrdquo in Proceedings of the 1st international conference onMobile systems applications and services ACM San FranciscoCalif USA 2003

[20] P Levis and D Culler ldquoMate a tiny virtual machine for sensornetworksrdquo ACM Sigplan Notices vol 37 no 10 pp 85ndash95 2002

[21] S Suenaga N Yoshioka and S Honiden ldquoGroup migration bymobile agents in wireless sensor networksrdquo Computer Journalvol 54 no 3 pp 345ndash355 2011

[22] P Bellavista Y Cai and T Magedanz ldquoRecent advances inmobile middleware for wireless systems and servicesrdquo MobileNetworks and Applications vol 16 no 3 pp 267ndash269 2011

[23] Z Zhirou and C Wengang ldquoModel and optimization ofmobile agentrsquos migration in gridrdquo in Proceedings of the 2ndInternational Conference on Bio-Inspired Computing Theoriesand Applications (BICTA rsquo07) pp 127ndash130 September 2007

[24] J-P Zhang M Jun J Yang and L-L Cheng ldquoAn improvedmigration strategy of mobile agentrdquo in Proceedings of the4th International Conference on Fuzzy Systems and KnowledgeDiscovery (FSKD rsquo07) pp 577ndash581 August 2007

[25] Y Lee and K Kim ldquoOptimal migration path searching usingpath adjustment and reassignment for mobile agentrdquo in Pro-ceedings of the 4th International Conference on NetworkedComputing and Advanced InformationManagement (NCM rsquo08)pp 564ndash569 Gyeongju South Korea September 2008

[26] M Tentori M Rodrıguez and J Favela Vara ldquoAn agent-basedmiddleware for the design of activity-aware applicationsrdquo IEEEIntelligent Systems vol 26 no 3 pp 15ndash23 2011

International Journal of

AerospaceEngineeringHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

RoboticsJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Active and Passive Electronic Components

Control Scienceand Engineering

Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

International Journal of

RotatingMachinery

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporation httpwwwhindawicom

Journal ofEngineeringVolume 2014

Submit your manuscripts athttpwwwhindawicom

VLSI Design

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Shock and Vibration

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Civil EngineeringAdvances in

Acoustics and VibrationAdvances in

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Electrical and Computer Engineering

Journal of

Advances inOptoElectronics

Hindawi Publishing Corporation httpwwwhindawicom

Volume 2014

The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014

SensorsJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Modelling amp Simulation in EngineeringHindawi Publishing Corporation httpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Chemical EngineeringInternational Journal of Antennas and

Propagation

International Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Navigation and Observation

International Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

DistributedSensor Networks

International Journal of

International Journal of Distributed Sensor Networks 3

A A AB B BD D

r1r1

r1

r2 r2r2

D1

D2

C C C

Figure 1 The target localization algorithm

119905 119883119896represents the location of node 119896 119883(119905) and 119883

119896are

two-dimensional vectors and 119883(119905) minus 119883119896 represents the

Euclidean distance between node 119896 and target entity intime 119905

34 Nodersquos Information CollectionModel In order to quantifythe data provided by sensors we need to define the amountof information According to (1) at any time 119911

119896(119905) is only

related to 119883(119905) minus 119883119896 Thus adopting the Gaussian model

the amount of information when the agent migrates to node119896 is the following

119868119896(119905) =

1

radic2120587120590

119890minus(119905)minus119883119896

2

21205902

(2)

where 120590 is the standard deviation of the amount of infor-mation which indicates the decrease of speed of 119868

119896(119905) when

the distance between sensor node and target entity increasesand 119883(119905) represents the location of the target entity which iscalculated by the target localization algorithm in Section 36

35 Beacon Frames We use beacons to termly exchangeinformation of the neighboring nodes A beacon frame con-sists of the following five fields 119883

119896represents the location of

node 119896 119890119896(119905) represents the residual energy of node 119896 in time

119905 119911119896(119905) represents the sensor measurement of node 119896 in time

119905 and the last two fields are sending time 119905119904and receiving time

119905119903 Beacon frames are used to provide information of location

andmeasurement to the target localization algorithm and theagentmigration algorithmMoreover they can be used to testthe survivability of neighbor nodes

36 Target Localization Algorithm As illustrated in (2) thelocation of target entity is calculated by the target localizationalgorithm At present target localization algorithms mostlyadopt the method of centroid weighting And the basic ideais to make nodersquos beacons take control of the centroidrsquoscoordinate and to reflect the general relations of the locationsby using weighting factors which influence the positions ofthe nodesrsquo centroids And according to the RSSIs of the threesensorsrsquo beacons it is easy to obtain the other nodersquos locationby the trilateration

As shown in Figure 1 the locations of 119860(119909119860 119910119860)

119861(119909119861 119910119861) and 119862(119909

119862 119910119862) are known and one of 119863(119909

119863 119910119863)

needs to be tested According to the RSSIs we can calculate

the distance of 119860119863 119861119863 and 119862119863 denoted by 1199031 1199032 and 119903

3

respectively Thus we can obtain the following equation set

(119909119861minus 119909119863)2

+ (119910119861minus 119910119863)2

= 1199032

2

(119909119860minus 119909119863)2

+ (119910119860minus 119910119863)2

= 1199032

1

(3)

Apparently the distance of 119860 and 119861 is less than or equalto 1199031+ 1199032 So (3) has two real solutions that are identical or

not which correspond to the coordinates of the two nodes1198631

and 1198632that are identical or not Select the one that is nearer

to 119862 as the approximate location of 119863 denoted as (1199091 1199101)

In the same way we can obtain the other two approximatelocations of119863 denoted as (119909

2 1199102) and (119909

3 1199103)Therefore the

coordinate of 119863 is (119909119863 119910119863) which can be calculated by the

following

119909119863

=1199091 (1199031+ 1199032) + 1199092 (1199032+ 1199033) + 1199093 (1199033+ 1199031)

1 (1199031+ 1199032) + 1 (119903

2+ 1199033) + 1 (119903

3+ 1199031)

119910119863

=1199101 (1199031+ 1199032) + 1199102 (1199032+ 1199033) + 1199103 (1199033+ 1199031)

1 (1199031+ 1199032) + 1 (119903

2+ 1199033) + 1 (119903

3+ 1199031)

(4)

As a result according to the previous target localizationalgorithm we can obtain 119909(119905)

4 Migration Algorithms Based on SingleMobile Agent

This section would propose SMLA and DMLA algorithmsfrom static and dynamic aspects respectively And it wouldthen propose pid-DMLA based on disadvantages of SMLAand DMLA

41 SMLA In SMLA algorithm we use static informa-tion Before distributing mobile agents the CIS collects bycommunicating with other nodes and it then generates amigration plan In thismigration plan information collectionvalues of all nodes in the network decreaseThe specific stepsof SMLA are as follows

Algorithm 1 (SMLA)(1) At time 119905 = 0 the node that has first detected

information of target entity is denoted as cluster headit generates a mobile agent

(2) At time 119905 = 0 calculate location information oftarget entity 119883(0) according to the target locationalgorithm

4 International Journal of Distributed Sensor Networks

Calculate the amount of information collection 1198680

according to (2) and set 119878119906119898119860119903119903119886119910 = 1198680

Select the next hop according to the predeterminedmigration sequence

(3) At time t when agent reaches node k the location oftarget entity is 119883(119905) according to the target locationalgorithmCalculate the amount of information collection ofnode 119896 119868

119896 and update 119878119906119898119860119903119903119886119910 = 119878119906119898119860119903119903119886119910 + 119868

119896

If 119878119906119898119860119903119903119886119910 is greater than the threshold then go tostep 4 otherwise continue to migrate to the next hopin the migration sequence update time 119905 = 119905 + 1 andthen repeat step 3

(4) Return to cluster head to process data

Though sometimes SMLA can gain the ideal migrationsequence there are many drawbacks in practical applica-tions Firstly since SMLA uses static information sensorrsquostransmission distance must be large enough to ensure thatCIS is within one-hop range So we need to adopt thenodes with greater power in communication capabilitiesAnd using these nodes will inevitably result in relativelyhigher energy consumption Secondly since SMLA is a staticalgorithm the static information which is acquired beforemigration lack timeliness and cannot reflect the changes ofthe network in the process of migrating For example thatsome nodes suddenly lose efficacy can result in the failure ofthe deployment of SMLA [9] Moreover most of WSNs arehighly dynamic so SMLA has some limitations in WSNs

42 DMLA SMLA is a centralized algorithm and it hasalready decided the migration sequence before distributingmobile agents Because WSN is a dynamic and distributednetwork static routing information exposes a lot of problemsin practical applications Thus DMLA is proposed DMLAdynamically selects the next hop in the process of migrationbased on characteristics of the network

421 The Migration Model of Mobile Agent in DMLA InDMLA mobile agents decide the nodes that they need tomigrate to In the surviving neighbor nodes not all of thenodes can provide information affecting the final resultTherefore the purpose ofmobile agentmigration algorithm isto produce a set of optimal subsets to determine the optimummigration sequence

In the target entity tracking we construct rules to choosethe next hop as followsmobile agent which is located in node119896 always looks for the next hop with minimal overheadsAs the selection of the next hop is a method of balancingoverheads and profit we define the overheads function whenmobile agent moves from node 119894 to neighbor node 119895 at time119905 as follows

119862119894119895(119905) = 119886

119864119894119895

119864max+ 119887(1 minus

119868119895(119905)

119868max) + radic1 minus 119886 minus 119887(1 minus

119890119895(119905)

119890max)

(5)

Equation (5) consists of three parts including energy con-sumption information collection and residual energy where119864119894119895represents transmission energy consumption when the

agent migrates directly from node 119894 to node 119895 119868119895(119905) represents

the amount of information collection of node 119895 at time 119905119890119895(119905) represents the residual energy of node 119895119864max represents

themaximumenergy consumption ofmigration between twonodes which is determined by the maximum transmissiondistance 119868max represents the maximum amount of informa-tion collection 119890max represents the initial energy of sensornode and it is a constant value and 119886 and 119887 are the balancecoefficients 0 le 119886 and 119887 le 1

Define the transmission energy consumption of the agentmigrating from node 119894 to node 119895 as follows

119864119894119895= 119875119894119895sdot 119879trans = 120572119875

119903-th sdot10038171003817100381710038171003817119883119894minus 119883119895

10038171003817100381710038171003817

2

sdot 119879trans (6)

where 119875119894119895represents the transmission power of sensor nodes

119883119894minus 119883119895 represents the distance between node 119894 and node

119895 119879trans a constant represents the time of transmitting themobile agent 120572 = 4120587

2

1205822

11986611198662 1198661 1198662are all of the

antenna gain factors (set them as 1) 120582 represents the signalwavelength and 119875

119903-th represents the receiving thresholdFrom (2) (5) and (6) we can gain

119862119894119895(119905) = 119886

10038171003817100381710038171003817119883119894minus 119883119895

10038171003817100381710038171003817

2

1198892max+ 119887

10038171003817100381710038171003817119883 (119905) minus 119883

119895

10038171003817100381710038171003817

2

1198891199052

max

+ radic1 minus 119886 minus 119887(1 minus

119890119895(119905)

119890max)

(7)

where 119889119905max represents the maximum possible distancebetween two nodes

Equation (7) can be rewritten as follows

119895lowast

= argmin

119886

10038171003817100381710038171003817119883119894minus 119883119895

10038171003817100381710038171003817

2

1198892max+ 119887

10038171003817100381710038171003817119883 (119905) minus 119883

119895

10038171003817100381710038171003817

2

1198891199052

max

+radic1 minus 119886 minus 119887(1 minus

119890119895(119905)

119890max)

119895 isin 119873119894

(8)

where 119873119894represents a group of neighbors of node 119894 node

119895lowast

isin 119873119894represents the next hop and 119895

lowast

isin 119873119894can ensure less

energy consumption stronger sensing capabilities and moreresidual energy

In summary the total cost in the migration process is

Cost =HOP119878minus1

sum

119901=0

119862119894119901119894119901+1

st 119878119906119898119860119903119903119886119910 =

HOP119878sum

119901=0

119868119901gt threshold

(9)

where 119901 represents the number of nodes migrated throughHOP119878represents the total number of migration hops In

the practical migration once mobile agent gains enoughinformation and it will terminate migration and return tothe CIS

International Journal of Distributed Sensor Networks 5

422The SpecificMigration Procedure of DMLA Themobileagent migrates to the alive node which has not been reachedand with the minimal migration overheads according to (7)In this way we can gain a table close to the optimalmigrationIf all nodes have been reached mobile agent will return thenode that distributes it And when the mobile agent hasacquired enough information it will return to the CIS

Since each node periodically receives beacon frames fromneighbor nodes DMLA can always use the updated localvariables to determine the next hop As only using localvariables sensor nodes can avoid the nodewhich has a greatertransmission distance so part of the transmission energycan be saved In addition when some nodes suddenly loseefficacy the whole migration process will not make mistakesSoDMLAalgorithm is a fault-tolerant algorithmThe specificmigration steps of DMLA are as follows

Algorithm 2 (DMLA)

(1) At time 119905 = 0 the node that has first detectedinformation of target entity is denoted as cluster headit generates a mobile agent

(2) At time 119905 = 0 calculate location information oftarget entity 119883(0) according to the target locationalgorithmCalculate the amount of information collection 119868

0

according to (2) and set 119878119906119898119860119903119903119886119910 = 1198680

Select the next hop according to (8)(3) At time 119905 when agent reaches node 119896 the location of

target entity is 119883(119905) according to the target locationalgorithmCalculate the amount of information collection ofnode 119896 119868

119896 and update 119878119906119898119860119903119903119886119910 = 119878119906119898119860119903119903119886119910 + 119868

119896

If 119878119906119898119860119903119903119886119910 is greater than the threshold then go tostep 4 otherwise continue to migrate to the next hopaccording to (8) update time 119905 = 119905+1 and then repeatstep 3

(4) Return to cluster head to process data

43 pid-DMLA

431 The Proposition of pid-DMLA In the cooperationprocessing applications the movement of target entity isone of the important factors that affect the final result [14ndash19] DMLA algorithm does not take the direction of themovement of target entity into account So in some cases itcannot select the optimumnext hop For example in Figure 2at time 119905 the mobile agent is located in node 119860 Nodes 119861

and 119862 are neighbors of 119860 and they are equidistant to 119860According to DMLA mobile agent calculates node 119861 as thenext hop This is because node 119861 is nearer to the targetentity than node 119862 Then at time 119905 + 1 the mobile agent islocated in node 119861 But the target entity moves to a positioncloser to node 119862 Thus DMLA on the contrary selectsthe node which is farther from the target and has smalleramount of information collection and measurement value

t t + 1

A A

B B

CC

Figure 2 The movement of the target entity

as the next hop Therefore when designing agent migrationalgorithms we need to focus on predicting the movement ofthe target entity and selecting the node with the maximuminformation And we will propose the pid-DMLA algorithmto solve the problems of the movement of the target entity inthe migration process

432 Agent Migration Model in pid-DMLA Suppose thatthe moving direction and speed of the target entity do notchange Then the change of the target entityrsquos displacementis a constant at each identical time interval as follows

119883 (119905 + 1) minus 119883 (119905) = 119883 (119905) minus 119883 (119905 minus 1) (10)

Equivalently

119883 (119905 + 1) = 2119883 (119905) minus 119883 (119905 minus 1) (11)

As illustrated in (11) the position information119883(119905 + 1) attime 119905 + 1 can be deduced from the positions at time 119905 minus 1 andtime 119905 In this way the location information119883(119905minus1) and119883(119905)

can be calculated by target location algorithm and accordingto (11) the position of target entity119883(119905+1) can be estimatedat time 119905 + 1 as follows

119883 (119905 + 1) = 2119883 (119905) minus 119883 (119905 minus 1) (12)

Update the overhead function in (7) and get

119862119894119895(119905) = 119886

10038171003817100381710038171003817119883119894minus 119883119895

10038171003817100381710038171003817

2

1198892max+ 119887

10038171003817100381710038171003817119883 (119905 + 1) minus 119883

119895

10038171003817100381710038171003817

2

1198891199052max

+ radic1 minus 119886 minus 119887(1 minus

119890119895(119905)

119890max)

(13)

where 119883(119905 + 1) minus 119883119895 represents the distance between node

119895 and the target entity at time 119905 + 1And the next hop is

119895lowast

= argmin

119886

10038171003817100381710038171003817119883119894minus 119883119895

10038171003817100381710038171003817

2

1198892max+ 119887

10038171003817100381710038171003817119883 (119905 + 1) minus 119883

119895

10038171003817100381710038171003817

2

1198891199052max

+radic1 minus 119886 minus 119887(1 minus

119890119895(119905)

119890max)

119895 isin 119873119894

(14)

6 International Journal of Distributed Sensor Networks

t

t

t + 1

t + 1

t + 2

t + 2

t minus 1

t minus 1

AB DC

Figure 3 The migration of pid-DMLA

At time 119905 = 0 since the position of the target entity atthe previous time is unknown the amount of informationcollection can be calculated by the DMLA At time 119905 gt 0mobile agent migrates to the node closer to the estimatedposition of the target entity So it can obtain larger sensingmeasurement and information collection which is the coreof pid-DMLA

433 The Specific Migration Procedure of pid-DMLA Theprocess of mobile agent migration of pid-DMLA is shown inFigure 3 In a period of time the target entity moves alongthe dotted line and mobile agent migrates from node 119860 to119863

through 119861 and 119862 As shown in Figure 3 mobile agent alwaysmigrates to the node nearest to the next position of the targetentity

The specific migration procedure of pid-DMLA is asfollows

Algorithm 3 (pid-DMLA)

(1) At time 119905 = 0 the node that has first detectedinformation of target entity is denoted as cluster head

it generates a mobile agent

(2) At time 119905 = 0 calculate location information oftarget entity 119883(0) according to the target locationalgorithm

Calculate the amount of information collection 1198680

according to (2) and set 119878119906119898119860119903119903119886119910 = 1198680

Select the next hop according to (14)

(3) At time 119905 when agent reaches node 119896 the location oftarget entity is 119883(119905) according to the target locationalgorithm

Calculate the amount of information collection ofnode 119896 119868

119896 and update 119878119906119898119860119903119903119886119910 = 119878119906119898119860119903119903119886119910 + 119868

119896

If 119878119906119898119860119903119903119886119910 is greater than the threshold then goto step 4 otherwise continue to migrate to the nexthop according to (14) update time 119905 = 119905 + 1 and thenrepeat step 3

(4) Return to cluster head to process data

5 Migration Algorithm Based on MultipleMobile Agents

51 The Architecture of Mpid-DMLA Many researchers haveproposed and realized feasible ways to organize themigrationof a single mobile agent such as Agilla program [18] andSensorWare program [19] Consider the functional require-ments of these programs Firstly they have not formed anarchitecture to manage the agent migration mechanismsin WSN Secondly they lack migration algorithms whena tracking application needs the collaboration of multipleagents So when dealing with multiple agentsrsquo migration weneed a specific agent to manage the applications

Therefore we will propose Mpid-DMLA to solve thisproblem

Suppose that multiple target entity applications aredeployed in an area of sensors each application is responsiblefor monitoring a part of the area and composed of oneor more mobile agents Therefore we have the followingdefinitions

Definition 4 (code segment application) A centralized pro-gram which consists of multiple mobile agents deployed inthe WSN to accomplish a specific task is called centralizedcode segment application of multiple agents which is calledcode segment application for short

Code segment applications are associated with sensingdata that need to be evaluated and they need to cooperate andinteract based on the need of programs On the other handaccording to the Deluge amp Mate model [20] code segmentapplications need to redeploy themselves so that they can bedistributed by the CIS

To track targets code segment applications require thefollowing four functions [21]

(1) Sensing Sensor nodes are distributed around thetarget node and they test and collect information ofthe surrounding environment

(2) Processing Information After acquiring sensor infor-mation most of the nodes send it to the CIS forprocessing while cluster heads will directly processinformation

(3) Communication In order tomaintain the applicationsensor nodes need to keep exchanging informationwith other nodes or CIS

(4) Migration Mobile agents migrate based on the needof code segment application

Constructing the architecture of multiple mobile agentsalso requires the following definitions

Definition 5 (main component agent) The main componentagent is responsible for managingmigration information andactivity cycle of all component agents processing sensingdata of other agents and collecting and exchanging theinformation of subcomponent agents Each code segmentapplication has one and only main component agent

International Journal of Distributed Sensor Networks 7

Definition 6 (subcomponent agent) The subcomponentagent is used to obtain the data of nodes and report them tothe main component agent Each code segment applicationhas one or more subcomponent agents

Definition 7 (communication component agent) The com-munication component agent is used to collect and exchangethe entity information of WSN by migrating Each codesegment application has one or more communication com-ponent agents

Definition 8 (middleware software) The middleware soft-ware located between the physical and the operating systemlayers is the software entity running in mobile agents

52 The Migration Model of Mobile Agent in Mpid-DMLAWhen a code segment application changes monitoring posi-tions all of the mobile agents in this application needto migrate together to reconstruct a new code segmentapplication In Agilla the main component agent controlsthe entire process by sending the migration sequence whileother mobile agents can communicate by accessing theindependent tuple space [22] And the engine virtualmachineworks as the kernel to provide services and controls all of theconcurrent execution of all of the agents in a node Howeverwith the increase of the mobile agents the possibilities ofthe loss of packet and the failure of migration will increasetoo Suppose that a code segment application is composedof 119873 mobile agents Define 119875 as the failure coefficient of asingle agent And then we can obtain the success coefficientof migration in the entire application as follows

119875success = (1 minus 119875)119873

(15)

Communication overheads based on pid-DMLA aregiven as follows

119862119894119895(119905) = [

[

119886

10038171003817100381710038171003817119883119894minus 119883119895

10038171003817100381710038171003817

2

1198892max+ 119887

10038171003817100381710038171003817119883 (119905 + 1) minus 119883

119895

10038171003817100381710038171003817

2

1198891199052max

+radic1 minus 119886 minus 119887(1 minus

119890119895(119905)

119890max)] times (119875success)

minus1

(16)

And the next hop is

119895lowast

= argmin

[

[

119886

10038171003817100381710038171003817119883119894minus 119883119895

10038171003817100381710038171003817

2

1198892max+ 119887

10038171003817100381710038171003817119883 (119905 + 1) minus 119883

119895

10038171003817100381710038171003817

2

1198891199052max

+radic1 minus 119886 minus 119887(1 minus

119890119895(119905)

119890max)]

]

times (119875success)minus1

(17)

53 The Specific Migration Procedure of Mpid-DMLA In theconstruction of the target entity tracking we assume thatsubcomponent agent is a small program of lightweight codewhich is suitable for migration And it can independently

Figure 4 Migration based on predicting information drive ofmultiple mobile agents

access the target entityrsquos location As shown in Figure 4hollow circles represent sensor nodes which are distributedin the sensing region code segment is composed of one maincomponent agent (solid circle) four subcomponent agents(solid squares) and one communication component agent(solid triangle) Each agent runs on one node executingcode segment applications The migration of code segmentapplication means the migration of the six mobile agentswhich is the migration of multiple mobile agents [19] Thefollowing algorithm can be used to describe the above-mentioned procedure

Algorithm 9 (Mpid-DMLA)

(1) A main component agent a communication com-ponent agent and several subcomponent agents aredeployed in different sensor nodes

(2) Sensor nodes are monitoring around the target entityThe node that has first detected information of targetentity is denoted as cluster head themain componentagent then migrates to it

(3) At time 119905 = 0 calculate location information oftarget entity 119883(0) according to the target locationalgorithmCalculate the amount of information collection 119868

0

according to (2)Define the threshold of the amount of information as119879ℎ119903119890119904ℎ119900119897119889 119880119875 and let 119879ℎ119903119890119904ℎ119900119897119889 119880119875 = 119868

0

The main component agent generates new com-munication component agent and transfers 119868

0and

119879ℎ119903119890119904ℎ119900119897119889 119880119875 to itThen the communication component agent sets tomigrate

(4) At time 119905 when agent reaches node 119896 the location oftarget entity is 119883(119905) according to the target locationalgorithmCalculate the amount of information collectionof node 119896 119868

119896 and update 119879ℎ119903119890119904ℎ119900119897119889 119880119875 =

119879ℎ119903119890119904ℎ119900119897119889 119880119875 + 119868119896

If 119879ℎ119903119890119904ℎ119900119897119889 119880119875 is greater than the threshold thengo to step 5 otherwise select the next hop formigration according to (17) update time 119905 = 119905 + 1and then repeat step 3

8 International Journal of Distributed Sensor Networks

t

t

t + 1

t + 1

t + 3 t + 4 t + 5 t + 6 t + 7t + 2

t + 3

t + 4

t + 5

t + 6

t + 7

t + 2

Figure 5 The migration process of SMLA

t t + 1 t + 3 t + 4 t + 5 t + 6 t + 7t + 2

tt + 1 t + 3

t + 4

t + 5

t + 6t + 7

t + 2

Figure 6 The migration process of DMLA

(5) Communication component agent stores the infor-mation of the migration path (119868

0(119905 = 0) 119868

1(119905 =

1) 119868119896(119905 = 119896) 119883 (119905 = 0) 119883 (119905 = 1) 119883 (119905 =

119896)) and other relevant information

And it interacts with the main component agent

(6) The main component agent generates the migrationsequence 119873

1 1198732 119873

119896 according to location of

target entity

Send it to each subcomponent agent (this step isoptional for different applications)

(7) The main component agent sends complete instruc-tions to the lower-layered middleware software

Method functions which are predefined in the mid-dleware software recycle the communication compo-nent agent and the subcomponent agents

(8) The main component agent goes to the destinationnode based on the migration sequence

If the migration fails send failure information toCIS to terminate this migration otherwise generatenew communication component agent and subcom-ponent agents and this migration ends

6 Simulation and Analyses

61 The Effect of Single Mobile Agent in Target TrackingBefore the simulation experiments we firstly use targettracking as a sample to demonstrate the migration effect ofa single mobile agent of different migration algorithms Asshown in Figures 5 6 and 7 suppose that a car is runningfrom left to right through the area covered byWSNThe smallcircle represents the sensor node solid small circle representsthe cluster head which distributes agents large dotted circlerepresents the communication range of sensor node and eachnode selects the next hop only within its communicationrange The above-mentioned figures illustrate the specificlocations of the car and agent from time 119905 to time 119905 + 7

The migration process of SMLA is shown in Figure 5 Attime 119905 CIS distributes the mobile agent to the node whichis the nearest to the target entity according to the prede-termined migration sequence When the object is stationaryor its motion can be estimated this static approach is verysuitable However it loses efficacy in some conditions Forexample if the object remains in the location at time 119905+1 thenthe migration sequence will be the positions that are aroundthe target entity But once the target entity starts to move notaccording to the predetermination SMLA will fail

DMLA can in some sense optimize the performanceof SMLA As shown in Figure 6 mobile agent tracks themovement of the target entity However the movement of

International Journal of Distributed Sensor Networks 9

t

t + 1 t + 3

t + 4

t + 5

t + 6t + 7

t + 2

t t + 1 t + 3 t + 4 t + 5 t + 6 t + 7t + 2

Figure 7 The migration process of pid-DMLA

Table 1 Basic simulation parameters of single mobile agent migra-tion algorithms

Network area 500 times 500mNumber of nodes 200 or 800Nodersquos initial energy 50 JNodersquos sensing radius 5mEnergy consumption for data processing 03 JEnergy consumption for sending beacon frames 01 JThe interval of node broadcasting beacon frames 01 sThe speed of the target entity 0sim50msThe threshold of the amount of information 18The size of a mobile agent 800 bitsSimulation time 100 s

the target entity is unpredictable so it is not accurate to decidemigration path only by the amount of information which ishowever related to the distance to the target entity

The pid-DMLA can predict movement information oftarget entity dynamically which is a better solution to suchproblems As shown in Figure 7 at time 119905 + 5 mobile agentselects the next hop according to the predicted informationwhich is different from DMLA In brief pid-DMLA rebuildsmigration sequence based on predicting themovement of thetarget entity to achieve the purpose of tracking

62 Performance Analysis on the Single Mobile Agent Migra-tion Algorithms We use simulation software NS2 to eval-uate the three single mobile agent migration algorithmsrsquoperformances The default simulation parameters are shownin Table 1

Suppose that once the agent migrates one hop all sensornodes in the network consume 02 J For SMLA the transmis-sion distance is set to 30m while for DMLA and pid-DMLAthe transmission distance is set to be 10m Therefore theconsumption of transmission for SMLA is three times morethan that of sendingThe reason for this setting is that SMLAneeds bigger transmission range to collect information Thedirection and size of movement of the target entity areconstant denoted as V (ms) Once the target entity reachesthe edge of the sensing area it reverses direction immediatelywhich ensures its being always within the sensing area

0 5 10 15 20 25 30 35 40 45 500

2

4

6

8

10

12

Target speed (ms)

Ener

gy co

nsum

ptio

n (J

)

SMLADMLApid-DMLA

Figure 8 Energy consumption of the single agent migrationalgorithms in Scenario 1

Scenario 1 Set the number of nodes as 200 the target entityspeed as 0sim50ms and the simulation time as 100 s whichcan guarantee a complete migration The simulation resultsare shown in Figures 8 and 9 It is obvious that DMLA andpid-DMLA consume less energy than SMLA in the wholenetwork This is because the communication cost of SMLAis higher than that of DMLA But with the increasing speedof the target entity the effect of SMLA algorithm in trackingthe target entity becomes less and less obvious thereforemigration may terminate and the energy consumption onthe contrary becomes slower On the other hand when thespeed of target entity is very low there are no significantdifferences among the three migration schemes

Scenario 2 Set the speed of target entity as 5ms and thenumber of nodes as 800 Generally the coverage and thenumber of nodes are positively correlated The simulationresults are shown in Figures 10 and 11 It is clear that pid-DMLA has the lowest energy consumption when the numberof nodes is more than 100 This is because with the increaseof the nodes the choices of predicting the next hop become

10 International Journal of Distributed Sensor Networks

0102030405060708090

100

Life

time (

s)

0 5 10 15 20 25 30 35 40 45 50Target speed (ms)

SMLADMLApid-DMLA

Figure 9 Lifetime of the single agent migration algorithms inScenario 1

Ener

gy co

nsum

ptio

n (J

)

SMLADMLApid-DMLA

0 100 200 300 400 500 600 700 8000

1

2

3

4

5

6

7

8

9

Target speed (ms)

Figure 10 Energy consumption of the single agent migrationalgorithms in Scenario 2

more as a result the overheads become less [23] When thenumber of nodes is less than 50 the performance of SMLAis better than those of the other two This is because withless nodes static and dynamic migration algorithms share themore similar migration path however the migration tablehas been predetermined in SMLA thus it on the contraryhas the least overheads On the other hand when the densityof nodes is high there would be more neighbors and moremigration choices Thus DMLA and pid-DMLA outperformSMLA

63 Multiple Mobile Agents Migration Algorithm PerformanceAnalysis Before the assessment of Mpid-DMLA we need toconsider two problems [24 25] The first is the effect of code

SMLADMLApid-DMLA

10

20

30

40

50

60

70

80

Life

time (

s)

0 100 200 300 400 500 600 700 800Target speed (ms)

Figure 11 Lifetime of the single agent migration algorithms inScenario 2

segment applications on the amount of communication inthe migration process To ensure the comprehensiveness ofthe evaluation set the number of communication componentagents as 1 and the number of subcomponent agents as 5 10or 20 and evaluate Mpid-DMLA based on code segmentswith different scales The second is the effect of differentenvironment and running platform in practical operation onthe amount of information and the success rate of migrationTherefore we only consider the relation of the number ofcommunication and energy instead of the influence of theagentrsquos size

The performance of the migration of mobile agent inAgilla system has been evaluated in [19] First Agilla systemis not designed to support multiple mobile agents specificallywhich means that it is not aware of the number of the agentsin the code segment application And the systemrsquos overheadsare related to the size ofmobile agent So inAgilla all kinds ofagents are considered to be identical and there is no discrim-ination between the main and communication componentagents However this does not affect the simulation resultsbecause all agents are uniformly assumed in the simulationin this paper

The purpose of multiple mobile agents migration isto complete the entire application migration with limitedenergy The default simulation parameters are shown inTable 2

And when main component agent recycles and sendsmobile agents suppose that the energy consumption of thenode where main component agent stays is neglected whilethat of other nodes is 005 J

Figures 12 13 and 14 illustrate the relation between thenumber of application code segments and the success rateof one hop migration when the numbers of subcomponentagents are 5 10 and 15 respectivelyThe success rate ofMpid-DMLA is higher when the number of subcomponent agentsis 5 than those when the numbers of subcomponent agents

International Journal of Distributed Sensor Networks 11

Table 2 Basic simulation algorithm parameters of a single mobileagent migration

Network area 100 times 100mNumber of nodes 7 times 7Number of main component agents per codesegment application 1

Number of subcomponent agents per code segmentapplication 5 10 or 15

Number of communication component agents percode segment application 1

Nodersquos initial energy 2 JEnergy consumption of a node when an agentreaches it 005 J

Interaction consumption when an agent migrates 01 J

20

30

40

50

60

70

80

90

100

Mpid-DMLAAgilla

2 25 3 35 45 554 5 6Number of migration applications

Succ

ess r

ate (

)

Figure 12 The relation between the number of code segmentapplications and the success rate of migration (I)

are 10 and 20 It is obvious that the successful rate of Mpid-DMLA is generally not less than 50 which is much largerthan that of Agilla

Moreover Figure 15 demonstrates the relationship be-tween themigration hops and the amount of communicationwhen the subcomponent agents are 5 As shown in Figure 15with the samemigration hops the amount of communicationof Mpid-DMLA is much less than that of Agilla As the num-ber of data packet reflects the amount of energy consumptionMpid-DMLA consumes less energy than Agilla

Although the simulation results show that Mpid-DMLAis a suitable migration scheme for multiple mobile agents itbases on some specific assumptions In general this schemestill encounters the following three issues

(1) When the main component agent loses efficacy inthe process of migration code segment applicationscannot be rebuilt

(2) When subcomponent agents lose efficacy afterdeployment code segment applications will keepwaiting because the applications cannot be recycled

20

30

40

50

60

70

80

90

100

Mpid-DMLAAgilla

2 25 3 35 45 554 5 6Number of migration applications

Succ

ess r

ate (

)

Figure 13 The relation between the number of code segmentapplications and the success rate of migration (II)

10

20

30

40

50

60

70

80

0

Mpid-DMLAAgilla

2 25 3 35 45 554 5 6Number of migration applications

Succ

ess r

ate (

)

Figure 14 The relation between the number of code segmentapplications and the success rate of migration (III)

1 15 2 25 3 35 420

40

60

80

100

120

140

160

180

200

Hops of main agent

Num

ber o

f dat

a pac

kets

Mpid-DMLAAgilla

Figure 15The relation between themigration hops and the amountof communication

12 International Journal of Distributed Sensor Networks

(3) If the main and the subcomponent agents which loseefficacy in issue 1 or 2 are left in the nodes withoutbeing recycled then othermobile agents cannot reachthese nodes

Therefore in designing middleware software it is nec-essary to add the feedback mechanism of main componentagent and it needs to periodically return ACK informationto the higher layer of the middleware software When themiddleware finds that the main component agent losesefficacy it needs to distribute a new one or a specializedindependent agent to recycle the useless agents in the wholenetwork

In Agilla the method of recycling the useless agentsis by adding the unified interface into all of the agentsRemoving all of the mobile agents will affect the function ofcode segment applications Therefore developing the moreappropriate middleware software is the future work whichis also related to the development of the better migrationstrategies [26]

7 Conclusion

As big data is the intensification of information architec-ture while distributed computing is the intensification ofinfrastructure and resource distributed computing is anappropriate method to deal with big data In this paperwe have firstly illustrated the mobile agent model in thedistributed computing And it led to the two considerablefactors of migration the selection of nodes and of thesequence of them Then we have proposed the assumptionsof the migration algorithms Moreover we have designedthe nodersquos sensor model and information collection modelFurthermore we have formulated the beacon frames ofmobile agent for communication Generally mobile agentmigration algorithms can be transferred as the target trackingissue Thus we have described a simple target localizationalgorithm applicable for migration On this basis we havegeneralized the whole migration process of the mobile agent

Considering the single mobile agent we have proposedthe static SMLA and the dynamic DMLA algorithms Afteranalyzing the drawbacks of SMLA and DMLA we thenproposed the pid-DMLA algorithm Section 4 has demon-strated the modeling procedure of the abovementioned threealgorithms Section 6 has emulated and evaluated their per-formances and studied their advantages and disadvantages indifferent scenarios

Considering the multiple mobile agents applied in prac-tical target tracking we have also put forward the Mpid-DMLA algorithm In the process of mobile agent componentconstructing code segment applications we defined threemobile agents with different functions to cooperate with eachother to accomplish the migration Section 5 illustrated thearchitecture modeling procedure and basic process ofMpid-DMLA Section 6 compared it with Agilla and evaluated itsadvantages and disadvantages

Acknowledgments

The paper is sponsored by the National Natural ScienceFoundation of China (61170065 61171053 61003039 6110019961003236 61202355 61201163 and 61373138) the Natu-ral Science Foundation of Jiangsu Province (BK2011755BK2011072 and BK2012436) the Scientific amp Technolog-ical Support Project of Jiangsu Province (BE2012183 andBE2012755) the Natural Science Key Fund for Collegesand Universities in Jiangsu Province (12KJA520002) theScientific Research amp Industry Promotion Project for HigherEducation Institutions (JHB2012-7) the Postdoctoral Foun-dation (2012M511753 1101011B 1202034C and 2013T60536)the Doctoral Fund of Ministry of Education of China(20113223110002 and 20123223120006) the Fund of NanjingUniversity of Posts and Telecommunications (NY212047)and the project funded by the Priority Academic Pro-gram Development of Jiangsu Higher Education Institutions(yx002001)

References

[1] Research Trends special issue on big data httpwwwresearch-trendscomwp-contentuploads201209Research Trends Is-sue30pdf 2012

[2] L Ismail and D Hagimont ldquoA performance evaluation of themobile agent paradigmrdquoACM SIGPLANNotices vol 34 no 10pp 306ndash313 1999

[3] AMainwaring J Polastre R Szewczyk D Culler and J Ander-son ldquoWireless sensor networks for habitat monitoringrdquo inProceedings of the 1st ACM International Workshop on WirelessSensor Networks and Applications pp 88ndash97 September 2002

[4] IVY-A sensor network infrastructure for the college of engi-neering httpwww-bsaceecsberkeleyeduprojectsivy 2008

[5] I F AkyildizW Su Y Sankarasubramaniam and E Cayirci ldquoAsurvey on sensor networksrdquo IEEE Communications Magazinevol 40 no 8 pp 102ndash105 2002

[6] Y-C Tseng S-P Kuo H-W Lee and C-F Huang ldquoLocationtracking in a wireless sensor network by mobile agents and itsdata fusion strategiesrdquoComputer Journal vol 47 no 4 pp 448ndash460 2004

[7] Y Xu and H Qi ldquoMobile agent migration modeling anddesign for target tracking in wireless sensor networksrdquo Ad HocNetworks vol 6 no 1 pp 1ndash16 2008

[8] Y Xu and H Qi ldquoDistributed computing paradigms for collab-orative signal and information processing in sensor networksrdquoJournal of Parallel and Distributed Computing vol 64 no 8 pp945ndash959 2004

[9] F Zhao J Shin and J Reich ldquoInformation-driven dynamicsensor collaborationrdquo IEEE Signal Processing Magazine vol 19no 2 pp 61ndash72 2002

[10] J Liu J Reich and F Zhao ldquoCollaborative in-network pro-cessing for target trackingrdquo Eurasip Journal on Applied SignalProcessing vol 2003 no 4 pp 378ndash391 2003

[11] J-W Kim J-S In K Hur J-W Kim and D-S Eom ldquoAnintelligent agent-based routing structure for mobile sinks inWSNsrdquo IEEE Transactions on Consumer Electronics vol 56 no4 pp 2310ndash2316 2010

[12] H Wang G Pottie K Yao and D Estrin ldquoEntropy-basedsensor selection heuristic for target localizationrdquo in Proceedings

International Journal of Distributed Sensor Networks 13

of the 3rd International Symposium on Information Processing inSensor Networks (IPSN rsquo04) pp 36ndash45 ACM Press April 2004

[13] T S RappaportWireless Communications Principles and Prac-tice IEEE Press 1996

[14] H Mousannif and I Khalil ldquoCooperative data management inwireless sensor networksrdquo International Journal of Computa-tional Intelligence Systems vol 5 no 3 pp 403ndash412 2012

[15] N Labraoui M Guerroui M Aliouat and T Zia ldquoDataaggregation security challenge in wireless sensor networks asurveyrdquo Ad-Hoc and Sensor Wireless Networks vol 12 no 3-4pp 295ndash324 2011

[16] M Chen L T Yang T Kwon L Zhou and M Jo ldquoItineraryplanning for energy-efficient agent communications in wirelesssensor networksrdquo IEEE Transactions on Vehicular Technologyvol 60 no 7 pp 3290ndash3299 2011

[17] Q Wu N S V Rao J Barhen et al ldquoOn computing mobileagent routes for data fusion in distributed sensor networksrdquoIEEE Transactions on Knowledge and Data Engineering vol 16no 6 pp 740ndash753 2004

[18] C-L Fok G-C Roman and C Lu ldquoMobile agent middlewarefor sensor networks an application case studyrdquo in Proceedingsof the 4th International Symposium on Information Processing inSensor Networks (IPSN rsquo05) pp 382ndash387 April 2005

[19] A Boulis C Han andM B Srivastava ldquoDesign and implemen-tation of a framework for efficient and programmable sensornetworksrdquo in Proceedings of the 1st international conference onMobile systems applications and services ACM San FranciscoCalif USA 2003

[20] P Levis and D Culler ldquoMate a tiny virtual machine for sensornetworksrdquo ACM Sigplan Notices vol 37 no 10 pp 85ndash95 2002

[21] S Suenaga N Yoshioka and S Honiden ldquoGroup migration bymobile agents in wireless sensor networksrdquo Computer Journalvol 54 no 3 pp 345ndash355 2011

[22] P Bellavista Y Cai and T Magedanz ldquoRecent advances inmobile middleware for wireless systems and servicesrdquo MobileNetworks and Applications vol 16 no 3 pp 267ndash269 2011

[23] Z Zhirou and C Wengang ldquoModel and optimization ofmobile agentrsquos migration in gridrdquo in Proceedings of the 2ndInternational Conference on Bio-Inspired Computing Theoriesand Applications (BICTA rsquo07) pp 127ndash130 September 2007

[24] J-P Zhang M Jun J Yang and L-L Cheng ldquoAn improvedmigration strategy of mobile agentrdquo in Proceedings of the4th International Conference on Fuzzy Systems and KnowledgeDiscovery (FSKD rsquo07) pp 577ndash581 August 2007

[25] Y Lee and K Kim ldquoOptimal migration path searching usingpath adjustment and reassignment for mobile agentrdquo in Pro-ceedings of the 4th International Conference on NetworkedComputing and Advanced InformationManagement (NCM rsquo08)pp 564ndash569 Gyeongju South Korea September 2008

[26] M Tentori M Rodrıguez and J Favela Vara ldquoAn agent-basedmiddleware for the design of activity-aware applicationsrdquo IEEEIntelligent Systems vol 26 no 3 pp 15ndash23 2011

International Journal of

AerospaceEngineeringHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

RoboticsJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Active and Passive Electronic Components

Control Scienceand Engineering

Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

International Journal of

RotatingMachinery

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporation httpwwwhindawicom

Journal ofEngineeringVolume 2014

Submit your manuscripts athttpwwwhindawicom

VLSI Design

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Shock and Vibration

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Civil EngineeringAdvances in

Acoustics and VibrationAdvances in

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Electrical and Computer Engineering

Journal of

Advances inOptoElectronics

Hindawi Publishing Corporation httpwwwhindawicom

Volume 2014

The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014

SensorsJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Modelling amp Simulation in EngineeringHindawi Publishing Corporation httpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Chemical EngineeringInternational Journal of Antennas and

Propagation

International Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Navigation and Observation

International Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

DistributedSensor Networks

International Journal of

4 International Journal of Distributed Sensor Networks

Calculate the amount of information collection 1198680

according to (2) and set 119878119906119898119860119903119903119886119910 = 1198680

Select the next hop according to the predeterminedmigration sequence

(3) At time t when agent reaches node k the location oftarget entity is 119883(119905) according to the target locationalgorithmCalculate the amount of information collection ofnode 119896 119868

119896 and update 119878119906119898119860119903119903119886119910 = 119878119906119898119860119903119903119886119910 + 119868

119896

If 119878119906119898119860119903119903119886119910 is greater than the threshold then go tostep 4 otherwise continue to migrate to the next hopin the migration sequence update time 119905 = 119905 + 1 andthen repeat step 3

(4) Return to cluster head to process data

Though sometimes SMLA can gain the ideal migrationsequence there are many drawbacks in practical applica-tions Firstly since SMLA uses static information sensorrsquostransmission distance must be large enough to ensure thatCIS is within one-hop range So we need to adopt thenodes with greater power in communication capabilitiesAnd using these nodes will inevitably result in relativelyhigher energy consumption Secondly since SMLA is a staticalgorithm the static information which is acquired beforemigration lack timeliness and cannot reflect the changes ofthe network in the process of migrating For example thatsome nodes suddenly lose efficacy can result in the failure ofthe deployment of SMLA [9] Moreover most of WSNs arehighly dynamic so SMLA has some limitations in WSNs

42 DMLA SMLA is a centralized algorithm and it hasalready decided the migration sequence before distributingmobile agents Because WSN is a dynamic and distributednetwork static routing information exposes a lot of problemsin practical applications Thus DMLA is proposed DMLAdynamically selects the next hop in the process of migrationbased on characteristics of the network

421 The Migration Model of Mobile Agent in DMLA InDMLA mobile agents decide the nodes that they need tomigrate to In the surviving neighbor nodes not all of thenodes can provide information affecting the final resultTherefore the purpose ofmobile agentmigration algorithm isto produce a set of optimal subsets to determine the optimummigration sequence

In the target entity tracking we construct rules to choosethe next hop as followsmobile agent which is located in node119896 always looks for the next hop with minimal overheadsAs the selection of the next hop is a method of balancingoverheads and profit we define the overheads function whenmobile agent moves from node 119894 to neighbor node 119895 at time119905 as follows

119862119894119895(119905) = 119886

119864119894119895

119864max+ 119887(1 minus

119868119895(119905)

119868max) + radic1 minus 119886 minus 119887(1 minus

119890119895(119905)

119890max)

(5)

Equation (5) consists of three parts including energy con-sumption information collection and residual energy where119864119894119895represents transmission energy consumption when the

agent migrates directly from node 119894 to node 119895 119868119895(119905) represents

the amount of information collection of node 119895 at time 119905119890119895(119905) represents the residual energy of node 119895119864max represents

themaximumenergy consumption ofmigration between twonodes which is determined by the maximum transmissiondistance 119868max represents the maximum amount of informa-tion collection 119890max represents the initial energy of sensornode and it is a constant value and 119886 and 119887 are the balancecoefficients 0 le 119886 and 119887 le 1

Define the transmission energy consumption of the agentmigrating from node 119894 to node 119895 as follows

119864119894119895= 119875119894119895sdot 119879trans = 120572119875

119903-th sdot10038171003817100381710038171003817119883119894minus 119883119895

10038171003817100381710038171003817

2

sdot 119879trans (6)

where 119875119894119895represents the transmission power of sensor nodes

119883119894minus 119883119895 represents the distance between node 119894 and node

119895 119879trans a constant represents the time of transmitting themobile agent 120572 = 4120587

2

1205822

11986611198662 1198661 1198662are all of the

antenna gain factors (set them as 1) 120582 represents the signalwavelength and 119875

119903-th represents the receiving thresholdFrom (2) (5) and (6) we can gain

119862119894119895(119905) = 119886

10038171003817100381710038171003817119883119894minus 119883119895

10038171003817100381710038171003817

2

1198892max+ 119887

10038171003817100381710038171003817119883 (119905) minus 119883

119895

10038171003817100381710038171003817

2

1198891199052

max

+ radic1 minus 119886 minus 119887(1 minus

119890119895(119905)

119890max)

(7)

where 119889119905max represents the maximum possible distancebetween two nodes

Equation (7) can be rewritten as follows

119895lowast

= argmin

119886

10038171003817100381710038171003817119883119894minus 119883119895

10038171003817100381710038171003817

2

1198892max+ 119887

10038171003817100381710038171003817119883 (119905) minus 119883

119895

10038171003817100381710038171003817

2

1198891199052

max

+radic1 minus 119886 minus 119887(1 minus

119890119895(119905)

119890max)

119895 isin 119873119894

(8)

where 119873119894represents a group of neighbors of node 119894 node

119895lowast

isin 119873119894represents the next hop and 119895

lowast

isin 119873119894can ensure less

energy consumption stronger sensing capabilities and moreresidual energy

In summary the total cost in the migration process is

Cost =HOP119878minus1

sum

119901=0

119862119894119901119894119901+1

st 119878119906119898119860119903119903119886119910 =

HOP119878sum

119901=0

119868119901gt threshold

(9)

where 119901 represents the number of nodes migrated throughHOP119878represents the total number of migration hops In

the practical migration once mobile agent gains enoughinformation and it will terminate migration and return tothe CIS

International Journal of Distributed Sensor Networks 5

422The SpecificMigration Procedure of DMLA Themobileagent migrates to the alive node which has not been reachedand with the minimal migration overheads according to (7)In this way we can gain a table close to the optimalmigrationIf all nodes have been reached mobile agent will return thenode that distributes it And when the mobile agent hasacquired enough information it will return to the CIS

Since each node periodically receives beacon frames fromneighbor nodes DMLA can always use the updated localvariables to determine the next hop As only using localvariables sensor nodes can avoid the nodewhich has a greatertransmission distance so part of the transmission energycan be saved In addition when some nodes suddenly loseefficacy the whole migration process will not make mistakesSoDMLAalgorithm is a fault-tolerant algorithmThe specificmigration steps of DMLA are as follows

Algorithm 2 (DMLA)

(1) At time 119905 = 0 the node that has first detectedinformation of target entity is denoted as cluster headit generates a mobile agent

(2) At time 119905 = 0 calculate location information oftarget entity 119883(0) according to the target locationalgorithmCalculate the amount of information collection 119868

0

according to (2) and set 119878119906119898119860119903119903119886119910 = 1198680

Select the next hop according to (8)(3) At time 119905 when agent reaches node 119896 the location of

target entity is 119883(119905) according to the target locationalgorithmCalculate the amount of information collection ofnode 119896 119868

119896 and update 119878119906119898119860119903119903119886119910 = 119878119906119898119860119903119903119886119910 + 119868

119896

If 119878119906119898119860119903119903119886119910 is greater than the threshold then go tostep 4 otherwise continue to migrate to the next hopaccording to (8) update time 119905 = 119905+1 and then repeatstep 3

(4) Return to cluster head to process data

43 pid-DMLA

431 The Proposition of pid-DMLA In the cooperationprocessing applications the movement of target entity isone of the important factors that affect the final result [14ndash19] DMLA algorithm does not take the direction of themovement of target entity into account So in some cases itcannot select the optimumnext hop For example in Figure 2at time 119905 the mobile agent is located in node 119860 Nodes 119861

and 119862 are neighbors of 119860 and they are equidistant to 119860According to DMLA mobile agent calculates node 119861 as thenext hop This is because node 119861 is nearer to the targetentity than node 119862 Then at time 119905 + 1 the mobile agent islocated in node 119861 But the target entity moves to a positioncloser to node 119862 Thus DMLA on the contrary selectsthe node which is farther from the target and has smalleramount of information collection and measurement value

t t + 1

A A

B B

CC

Figure 2 The movement of the target entity

as the next hop Therefore when designing agent migrationalgorithms we need to focus on predicting the movement ofthe target entity and selecting the node with the maximuminformation And we will propose the pid-DMLA algorithmto solve the problems of the movement of the target entity inthe migration process

432 Agent Migration Model in pid-DMLA Suppose thatthe moving direction and speed of the target entity do notchange Then the change of the target entityrsquos displacementis a constant at each identical time interval as follows

119883 (119905 + 1) minus 119883 (119905) = 119883 (119905) minus 119883 (119905 minus 1) (10)

Equivalently

119883 (119905 + 1) = 2119883 (119905) minus 119883 (119905 minus 1) (11)

As illustrated in (11) the position information119883(119905 + 1) attime 119905 + 1 can be deduced from the positions at time 119905 minus 1 andtime 119905 In this way the location information119883(119905minus1) and119883(119905)

can be calculated by target location algorithm and accordingto (11) the position of target entity119883(119905+1) can be estimatedat time 119905 + 1 as follows

119883 (119905 + 1) = 2119883 (119905) minus 119883 (119905 minus 1) (12)

Update the overhead function in (7) and get

119862119894119895(119905) = 119886

10038171003817100381710038171003817119883119894minus 119883119895

10038171003817100381710038171003817

2

1198892max+ 119887

10038171003817100381710038171003817119883 (119905 + 1) minus 119883

119895

10038171003817100381710038171003817

2

1198891199052max

+ radic1 minus 119886 minus 119887(1 minus

119890119895(119905)

119890max)

(13)

where 119883(119905 + 1) minus 119883119895 represents the distance between node

119895 and the target entity at time 119905 + 1And the next hop is

119895lowast

= argmin

119886

10038171003817100381710038171003817119883119894minus 119883119895

10038171003817100381710038171003817

2

1198892max+ 119887

10038171003817100381710038171003817119883 (119905 + 1) minus 119883

119895

10038171003817100381710038171003817

2

1198891199052max

+radic1 minus 119886 minus 119887(1 minus

119890119895(119905)

119890max)

119895 isin 119873119894

(14)

6 International Journal of Distributed Sensor Networks

t

t

t + 1

t + 1

t + 2

t + 2

t minus 1

t minus 1

AB DC

Figure 3 The migration of pid-DMLA

At time 119905 = 0 since the position of the target entity atthe previous time is unknown the amount of informationcollection can be calculated by the DMLA At time 119905 gt 0mobile agent migrates to the node closer to the estimatedposition of the target entity So it can obtain larger sensingmeasurement and information collection which is the coreof pid-DMLA

433 The Specific Migration Procedure of pid-DMLA Theprocess of mobile agent migration of pid-DMLA is shown inFigure 3 In a period of time the target entity moves alongthe dotted line and mobile agent migrates from node 119860 to119863

through 119861 and 119862 As shown in Figure 3 mobile agent alwaysmigrates to the node nearest to the next position of the targetentity

The specific migration procedure of pid-DMLA is asfollows

Algorithm 3 (pid-DMLA)

(1) At time 119905 = 0 the node that has first detectedinformation of target entity is denoted as cluster head

it generates a mobile agent

(2) At time 119905 = 0 calculate location information oftarget entity 119883(0) according to the target locationalgorithm

Calculate the amount of information collection 1198680

according to (2) and set 119878119906119898119860119903119903119886119910 = 1198680

Select the next hop according to (14)

(3) At time 119905 when agent reaches node 119896 the location oftarget entity is 119883(119905) according to the target locationalgorithm

Calculate the amount of information collection ofnode 119896 119868

119896 and update 119878119906119898119860119903119903119886119910 = 119878119906119898119860119903119903119886119910 + 119868

119896

If 119878119906119898119860119903119903119886119910 is greater than the threshold then goto step 4 otherwise continue to migrate to the nexthop according to (14) update time 119905 = 119905 + 1 and thenrepeat step 3

(4) Return to cluster head to process data

5 Migration Algorithm Based on MultipleMobile Agents

51 The Architecture of Mpid-DMLA Many researchers haveproposed and realized feasible ways to organize themigrationof a single mobile agent such as Agilla program [18] andSensorWare program [19] Consider the functional require-ments of these programs Firstly they have not formed anarchitecture to manage the agent migration mechanismsin WSN Secondly they lack migration algorithms whena tracking application needs the collaboration of multipleagents So when dealing with multiple agentsrsquo migration weneed a specific agent to manage the applications

Therefore we will propose Mpid-DMLA to solve thisproblem

Suppose that multiple target entity applications aredeployed in an area of sensors each application is responsiblefor monitoring a part of the area and composed of oneor more mobile agents Therefore we have the followingdefinitions

Definition 4 (code segment application) A centralized pro-gram which consists of multiple mobile agents deployed inthe WSN to accomplish a specific task is called centralizedcode segment application of multiple agents which is calledcode segment application for short

Code segment applications are associated with sensingdata that need to be evaluated and they need to cooperate andinteract based on the need of programs On the other handaccording to the Deluge amp Mate model [20] code segmentapplications need to redeploy themselves so that they can bedistributed by the CIS

To track targets code segment applications require thefollowing four functions [21]

(1) Sensing Sensor nodes are distributed around thetarget node and they test and collect information ofthe surrounding environment

(2) Processing Information After acquiring sensor infor-mation most of the nodes send it to the CIS forprocessing while cluster heads will directly processinformation

(3) Communication In order tomaintain the applicationsensor nodes need to keep exchanging informationwith other nodes or CIS

(4) Migration Mobile agents migrate based on the needof code segment application

Constructing the architecture of multiple mobile agentsalso requires the following definitions

Definition 5 (main component agent) The main componentagent is responsible for managingmigration information andactivity cycle of all component agents processing sensingdata of other agents and collecting and exchanging theinformation of subcomponent agents Each code segmentapplication has one and only main component agent

International Journal of Distributed Sensor Networks 7

Definition 6 (subcomponent agent) The subcomponentagent is used to obtain the data of nodes and report them tothe main component agent Each code segment applicationhas one or more subcomponent agents

Definition 7 (communication component agent) The com-munication component agent is used to collect and exchangethe entity information of WSN by migrating Each codesegment application has one or more communication com-ponent agents

Definition 8 (middleware software) The middleware soft-ware located between the physical and the operating systemlayers is the software entity running in mobile agents

52 The Migration Model of Mobile Agent in Mpid-DMLAWhen a code segment application changes monitoring posi-tions all of the mobile agents in this application needto migrate together to reconstruct a new code segmentapplication In Agilla the main component agent controlsthe entire process by sending the migration sequence whileother mobile agents can communicate by accessing theindependent tuple space [22] And the engine virtualmachineworks as the kernel to provide services and controls all of theconcurrent execution of all of the agents in a node Howeverwith the increase of the mobile agents the possibilities ofthe loss of packet and the failure of migration will increasetoo Suppose that a code segment application is composedof 119873 mobile agents Define 119875 as the failure coefficient of asingle agent And then we can obtain the success coefficientof migration in the entire application as follows

119875success = (1 minus 119875)119873

(15)

Communication overheads based on pid-DMLA aregiven as follows

119862119894119895(119905) = [

[

119886

10038171003817100381710038171003817119883119894minus 119883119895

10038171003817100381710038171003817

2

1198892max+ 119887

10038171003817100381710038171003817119883 (119905 + 1) minus 119883

119895

10038171003817100381710038171003817

2

1198891199052max

+radic1 minus 119886 minus 119887(1 minus

119890119895(119905)

119890max)] times (119875success)

minus1

(16)

And the next hop is

119895lowast

= argmin

[

[

119886

10038171003817100381710038171003817119883119894minus 119883119895

10038171003817100381710038171003817

2

1198892max+ 119887

10038171003817100381710038171003817119883 (119905 + 1) minus 119883

119895

10038171003817100381710038171003817

2

1198891199052max

+radic1 minus 119886 minus 119887(1 minus

119890119895(119905)

119890max)]

]

times (119875success)minus1

(17)

53 The Specific Migration Procedure of Mpid-DMLA In theconstruction of the target entity tracking we assume thatsubcomponent agent is a small program of lightweight codewhich is suitable for migration And it can independently

Figure 4 Migration based on predicting information drive ofmultiple mobile agents

access the target entityrsquos location As shown in Figure 4hollow circles represent sensor nodes which are distributedin the sensing region code segment is composed of one maincomponent agent (solid circle) four subcomponent agents(solid squares) and one communication component agent(solid triangle) Each agent runs on one node executingcode segment applications The migration of code segmentapplication means the migration of the six mobile agentswhich is the migration of multiple mobile agents [19] Thefollowing algorithm can be used to describe the above-mentioned procedure

Algorithm 9 (Mpid-DMLA)

(1) A main component agent a communication com-ponent agent and several subcomponent agents aredeployed in different sensor nodes

(2) Sensor nodes are monitoring around the target entityThe node that has first detected information of targetentity is denoted as cluster head themain componentagent then migrates to it

(3) At time 119905 = 0 calculate location information oftarget entity 119883(0) according to the target locationalgorithmCalculate the amount of information collection 119868

0

according to (2)Define the threshold of the amount of information as119879ℎ119903119890119904ℎ119900119897119889 119880119875 and let 119879ℎ119903119890119904ℎ119900119897119889 119880119875 = 119868

0

The main component agent generates new com-munication component agent and transfers 119868

0and

119879ℎ119903119890119904ℎ119900119897119889 119880119875 to itThen the communication component agent sets tomigrate

(4) At time 119905 when agent reaches node 119896 the location oftarget entity is 119883(119905) according to the target locationalgorithmCalculate the amount of information collectionof node 119896 119868

119896 and update 119879ℎ119903119890119904ℎ119900119897119889 119880119875 =

119879ℎ119903119890119904ℎ119900119897119889 119880119875 + 119868119896

If 119879ℎ119903119890119904ℎ119900119897119889 119880119875 is greater than the threshold thengo to step 5 otherwise select the next hop formigration according to (17) update time 119905 = 119905 + 1and then repeat step 3

8 International Journal of Distributed Sensor Networks

t

t

t + 1

t + 1

t + 3 t + 4 t + 5 t + 6 t + 7t + 2

t + 3

t + 4

t + 5

t + 6

t + 7

t + 2

Figure 5 The migration process of SMLA

t t + 1 t + 3 t + 4 t + 5 t + 6 t + 7t + 2

tt + 1 t + 3

t + 4

t + 5

t + 6t + 7

t + 2

Figure 6 The migration process of DMLA

(5) Communication component agent stores the infor-mation of the migration path (119868

0(119905 = 0) 119868

1(119905 =

1) 119868119896(119905 = 119896) 119883 (119905 = 0) 119883 (119905 = 1) 119883 (119905 =

119896)) and other relevant information

And it interacts with the main component agent

(6) The main component agent generates the migrationsequence 119873

1 1198732 119873

119896 according to location of

target entity

Send it to each subcomponent agent (this step isoptional for different applications)

(7) The main component agent sends complete instruc-tions to the lower-layered middleware software

Method functions which are predefined in the mid-dleware software recycle the communication compo-nent agent and the subcomponent agents

(8) The main component agent goes to the destinationnode based on the migration sequence

If the migration fails send failure information toCIS to terminate this migration otherwise generatenew communication component agent and subcom-ponent agents and this migration ends

6 Simulation and Analyses

61 The Effect of Single Mobile Agent in Target TrackingBefore the simulation experiments we firstly use targettracking as a sample to demonstrate the migration effect ofa single mobile agent of different migration algorithms Asshown in Figures 5 6 and 7 suppose that a car is runningfrom left to right through the area covered byWSNThe smallcircle represents the sensor node solid small circle representsthe cluster head which distributes agents large dotted circlerepresents the communication range of sensor node and eachnode selects the next hop only within its communicationrange The above-mentioned figures illustrate the specificlocations of the car and agent from time 119905 to time 119905 + 7

The migration process of SMLA is shown in Figure 5 Attime 119905 CIS distributes the mobile agent to the node whichis the nearest to the target entity according to the prede-termined migration sequence When the object is stationaryor its motion can be estimated this static approach is verysuitable However it loses efficacy in some conditions Forexample if the object remains in the location at time 119905+1 thenthe migration sequence will be the positions that are aroundthe target entity But once the target entity starts to move notaccording to the predetermination SMLA will fail

DMLA can in some sense optimize the performanceof SMLA As shown in Figure 6 mobile agent tracks themovement of the target entity However the movement of

International Journal of Distributed Sensor Networks 9

t

t + 1 t + 3

t + 4

t + 5

t + 6t + 7

t + 2

t t + 1 t + 3 t + 4 t + 5 t + 6 t + 7t + 2

Figure 7 The migration process of pid-DMLA

Table 1 Basic simulation parameters of single mobile agent migra-tion algorithms

Network area 500 times 500mNumber of nodes 200 or 800Nodersquos initial energy 50 JNodersquos sensing radius 5mEnergy consumption for data processing 03 JEnergy consumption for sending beacon frames 01 JThe interval of node broadcasting beacon frames 01 sThe speed of the target entity 0sim50msThe threshold of the amount of information 18The size of a mobile agent 800 bitsSimulation time 100 s

the target entity is unpredictable so it is not accurate to decidemigration path only by the amount of information which ishowever related to the distance to the target entity

The pid-DMLA can predict movement information oftarget entity dynamically which is a better solution to suchproblems As shown in Figure 7 at time 119905 + 5 mobile agentselects the next hop according to the predicted informationwhich is different from DMLA In brief pid-DMLA rebuildsmigration sequence based on predicting themovement of thetarget entity to achieve the purpose of tracking

62 Performance Analysis on the Single Mobile Agent Migra-tion Algorithms We use simulation software NS2 to eval-uate the three single mobile agent migration algorithmsrsquoperformances The default simulation parameters are shownin Table 1

Suppose that once the agent migrates one hop all sensornodes in the network consume 02 J For SMLA the transmis-sion distance is set to 30m while for DMLA and pid-DMLAthe transmission distance is set to be 10m Therefore theconsumption of transmission for SMLA is three times morethan that of sendingThe reason for this setting is that SMLAneeds bigger transmission range to collect information Thedirection and size of movement of the target entity areconstant denoted as V (ms) Once the target entity reachesthe edge of the sensing area it reverses direction immediatelywhich ensures its being always within the sensing area

0 5 10 15 20 25 30 35 40 45 500

2

4

6

8

10

12

Target speed (ms)

Ener

gy co

nsum

ptio

n (J

)

SMLADMLApid-DMLA

Figure 8 Energy consumption of the single agent migrationalgorithms in Scenario 1

Scenario 1 Set the number of nodes as 200 the target entityspeed as 0sim50ms and the simulation time as 100 s whichcan guarantee a complete migration The simulation resultsare shown in Figures 8 and 9 It is obvious that DMLA andpid-DMLA consume less energy than SMLA in the wholenetwork This is because the communication cost of SMLAis higher than that of DMLA But with the increasing speedof the target entity the effect of SMLA algorithm in trackingthe target entity becomes less and less obvious thereforemigration may terminate and the energy consumption onthe contrary becomes slower On the other hand when thespeed of target entity is very low there are no significantdifferences among the three migration schemes

Scenario 2 Set the speed of target entity as 5ms and thenumber of nodes as 800 Generally the coverage and thenumber of nodes are positively correlated The simulationresults are shown in Figures 10 and 11 It is clear that pid-DMLA has the lowest energy consumption when the numberof nodes is more than 100 This is because with the increaseof the nodes the choices of predicting the next hop become

10 International Journal of Distributed Sensor Networks

0102030405060708090

100

Life

time (

s)

0 5 10 15 20 25 30 35 40 45 50Target speed (ms)

SMLADMLApid-DMLA

Figure 9 Lifetime of the single agent migration algorithms inScenario 1

Ener

gy co

nsum

ptio

n (J

)

SMLADMLApid-DMLA

0 100 200 300 400 500 600 700 8000

1

2

3

4

5

6

7

8

9

Target speed (ms)

Figure 10 Energy consumption of the single agent migrationalgorithms in Scenario 2

more as a result the overheads become less [23] When thenumber of nodes is less than 50 the performance of SMLAis better than those of the other two This is because withless nodes static and dynamic migration algorithms share themore similar migration path however the migration tablehas been predetermined in SMLA thus it on the contraryhas the least overheads On the other hand when the densityof nodes is high there would be more neighbors and moremigration choices Thus DMLA and pid-DMLA outperformSMLA

63 Multiple Mobile Agents Migration Algorithm PerformanceAnalysis Before the assessment of Mpid-DMLA we need toconsider two problems [24 25] The first is the effect of code

SMLADMLApid-DMLA

10

20

30

40

50

60

70

80

Life

time (

s)

0 100 200 300 400 500 600 700 800Target speed (ms)

Figure 11 Lifetime of the single agent migration algorithms inScenario 2

segment applications on the amount of communication inthe migration process To ensure the comprehensiveness ofthe evaluation set the number of communication componentagents as 1 and the number of subcomponent agents as 5 10or 20 and evaluate Mpid-DMLA based on code segmentswith different scales The second is the effect of differentenvironment and running platform in practical operation onthe amount of information and the success rate of migrationTherefore we only consider the relation of the number ofcommunication and energy instead of the influence of theagentrsquos size

The performance of the migration of mobile agent inAgilla system has been evaluated in [19] First Agilla systemis not designed to support multiple mobile agents specificallywhich means that it is not aware of the number of the agentsin the code segment application And the systemrsquos overheadsare related to the size ofmobile agent So inAgilla all kinds ofagents are considered to be identical and there is no discrim-ination between the main and communication componentagents However this does not affect the simulation resultsbecause all agents are uniformly assumed in the simulationin this paper

The purpose of multiple mobile agents migration isto complete the entire application migration with limitedenergy The default simulation parameters are shown inTable 2

And when main component agent recycles and sendsmobile agents suppose that the energy consumption of thenode where main component agent stays is neglected whilethat of other nodes is 005 J

Figures 12 13 and 14 illustrate the relation between thenumber of application code segments and the success rateof one hop migration when the numbers of subcomponentagents are 5 10 and 15 respectivelyThe success rate ofMpid-DMLA is higher when the number of subcomponent agentsis 5 than those when the numbers of subcomponent agents

International Journal of Distributed Sensor Networks 11

Table 2 Basic simulation algorithm parameters of a single mobileagent migration

Network area 100 times 100mNumber of nodes 7 times 7Number of main component agents per codesegment application 1

Number of subcomponent agents per code segmentapplication 5 10 or 15

Number of communication component agents percode segment application 1

Nodersquos initial energy 2 JEnergy consumption of a node when an agentreaches it 005 J

Interaction consumption when an agent migrates 01 J

20

30

40

50

60

70

80

90

100

Mpid-DMLAAgilla

2 25 3 35 45 554 5 6Number of migration applications

Succ

ess r

ate (

)

Figure 12 The relation between the number of code segmentapplications and the success rate of migration (I)

are 10 and 20 It is obvious that the successful rate of Mpid-DMLA is generally not less than 50 which is much largerthan that of Agilla

Moreover Figure 15 demonstrates the relationship be-tween themigration hops and the amount of communicationwhen the subcomponent agents are 5 As shown in Figure 15with the samemigration hops the amount of communicationof Mpid-DMLA is much less than that of Agilla As the num-ber of data packet reflects the amount of energy consumptionMpid-DMLA consumes less energy than Agilla

Although the simulation results show that Mpid-DMLAis a suitable migration scheme for multiple mobile agents itbases on some specific assumptions In general this schemestill encounters the following three issues

(1) When the main component agent loses efficacy inthe process of migration code segment applicationscannot be rebuilt

(2) When subcomponent agents lose efficacy afterdeployment code segment applications will keepwaiting because the applications cannot be recycled

20

30

40

50

60

70

80

90

100

Mpid-DMLAAgilla

2 25 3 35 45 554 5 6Number of migration applications

Succ

ess r

ate (

)

Figure 13 The relation between the number of code segmentapplications and the success rate of migration (II)

10

20

30

40

50

60

70

80

0

Mpid-DMLAAgilla

2 25 3 35 45 554 5 6Number of migration applications

Succ

ess r

ate (

)

Figure 14 The relation between the number of code segmentapplications and the success rate of migration (III)

1 15 2 25 3 35 420

40

60

80

100

120

140

160

180

200

Hops of main agent

Num

ber o

f dat

a pac

kets

Mpid-DMLAAgilla

Figure 15The relation between themigration hops and the amountof communication

12 International Journal of Distributed Sensor Networks

(3) If the main and the subcomponent agents which loseefficacy in issue 1 or 2 are left in the nodes withoutbeing recycled then othermobile agents cannot reachthese nodes

Therefore in designing middleware software it is nec-essary to add the feedback mechanism of main componentagent and it needs to periodically return ACK informationto the higher layer of the middleware software When themiddleware finds that the main component agent losesefficacy it needs to distribute a new one or a specializedindependent agent to recycle the useless agents in the wholenetwork

In Agilla the method of recycling the useless agentsis by adding the unified interface into all of the agentsRemoving all of the mobile agents will affect the function ofcode segment applications Therefore developing the moreappropriate middleware software is the future work whichis also related to the development of the better migrationstrategies [26]

7 Conclusion

As big data is the intensification of information architec-ture while distributed computing is the intensification ofinfrastructure and resource distributed computing is anappropriate method to deal with big data In this paperwe have firstly illustrated the mobile agent model in thedistributed computing And it led to the two considerablefactors of migration the selection of nodes and of thesequence of them Then we have proposed the assumptionsof the migration algorithms Moreover we have designedthe nodersquos sensor model and information collection modelFurthermore we have formulated the beacon frames ofmobile agent for communication Generally mobile agentmigration algorithms can be transferred as the target trackingissue Thus we have described a simple target localizationalgorithm applicable for migration On this basis we havegeneralized the whole migration process of the mobile agent

Considering the single mobile agent we have proposedthe static SMLA and the dynamic DMLA algorithms Afteranalyzing the drawbacks of SMLA and DMLA we thenproposed the pid-DMLA algorithm Section 4 has demon-strated the modeling procedure of the abovementioned threealgorithms Section 6 has emulated and evaluated their per-formances and studied their advantages and disadvantages indifferent scenarios

Considering the multiple mobile agents applied in prac-tical target tracking we have also put forward the Mpid-DMLA algorithm In the process of mobile agent componentconstructing code segment applications we defined threemobile agents with different functions to cooperate with eachother to accomplish the migration Section 5 illustrated thearchitecture modeling procedure and basic process ofMpid-DMLA Section 6 compared it with Agilla and evaluated itsadvantages and disadvantages

Acknowledgments

The paper is sponsored by the National Natural ScienceFoundation of China (61170065 61171053 61003039 6110019961003236 61202355 61201163 and 61373138) the Natu-ral Science Foundation of Jiangsu Province (BK2011755BK2011072 and BK2012436) the Scientific amp Technolog-ical Support Project of Jiangsu Province (BE2012183 andBE2012755) the Natural Science Key Fund for Collegesand Universities in Jiangsu Province (12KJA520002) theScientific Research amp Industry Promotion Project for HigherEducation Institutions (JHB2012-7) the Postdoctoral Foun-dation (2012M511753 1101011B 1202034C and 2013T60536)the Doctoral Fund of Ministry of Education of China(20113223110002 and 20123223120006) the Fund of NanjingUniversity of Posts and Telecommunications (NY212047)and the project funded by the Priority Academic Pro-gram Development of Jiangsu Higher Education Institutions(yx002001)

References

[1] Research Trends special issue on big data httpwwwresearch-trendscomwp-contentuploads201209Research Trends Is-sue30pdf 2012

[2] L Ismail and D Hagimont ldquoA performance evaluation of themobile agent paradigmrdquoACM SIGPLANNotices vol 34 no 10pp 306ndash313 1999

[3] AMainwaring J Polastre R Szewczyk D Culler and J Ander-son ldquoWireless sensor networks for habitat monitoringrdquo inProceedings of the 1st ACM International Workshop on WirelessSensor Networks and Applications pp 88ndash97 September 2002

[4] IVY-A sensor network infrastructure for the college of engi-neering httpwww-bsaceecsberkeleyeduprojectsivy 2008

[5] I F AkyildizW Su Y Sankarasubramaniam and E Cayirci ldquoAsurvey on sensor networksrdquo IEEE Communications Magazinevol 40 no 8 pp 102ndash105 2002

[6] Y-C Tseng S-P Kuo H-W Lee and C-F Huang ldquoLocationtracking in a wireless sensor network by mobile agents and itsdata fusion strategiesrdquoComputer Journal vol 47 no 4 pp 448ndash460 2004

[7] Y Xu and H Qi ldquoMobile agent migration modeling anddesign for target tracking in wireless sensor networksrdquo Ad HocNetworks vol 6 no 1 pp 1ndash16 2008

[8] Y Xu and H Qi ldquoDistributed computing paradigms for collab-orative signal and information processing in sensor networksrdquoJournal of Parallel and Distributed Computing vol 64 no 8 pp945ndash959 2004

[9] F Zhao J Shin and J Reich ldquoInformation-driven dynamicsensor collaborationrdquo IEEE Signal Processing Magazine vol 19no 2 pp 61ndash72 2002

[10] J Liu J Reich and F Zhao ldquoCollaborative in-network pro-cessing for target trackingrdquo Eurasip Journal on Applied SignalProcessing vol 2003 no 4 pp 378ndash391 2003

[11] J-W Kim J-S In K Hur J-W Kim and D-S Eom ldquoAnintelligent agent-based routing structure for mobile sinks inWSNsrdquo IEEE Transactions on Consumer Electronics vol 56 no4 pp 2310ndash2316 2010

[12] H Wang G Pottie K Yao and D Estrin ldquoEntropy-basedsensor selection heuristic for target localizationrdquo in Proceedings

International Journal of Distributed Sensor Networks 13

of the 3rd International Symposium on Information Processing inSensor Networks (IPSN rsquo04) pp 36ndash45 ACM Press April 2004

[13] T S RappaportWireless Communications Principles and Prac-tice IEEE Press 1996

[14] H Mousannif and I Khalil ldquoCooperative data management inwireless sensor networksrdquo International Journal of Computa-tional Intelligence Systems vol 5 no 3 pp 403ndash412 2012

[15] N Labraoui M Guerroui M Aliouat and T Zia ldquoDataaggregation security challenge in wireless sensor networks asurveyrdquo Ad-Hoc and Sensor Wireless Networks vol 12 no 3-4pp 295ndash324 2011

[16] M Chen L T Yang T Kwon L Zhou and M Jo ldquoItineraryplanning for energy-efficient agent communications in wirelesssensor networksrdquo IEEE Transactions on Vehicular Technologyvol 60 no 7 pp 3290ndash3299 2011

[17] Q Wu N S V Rao J Barhen et al ldquoOn computing mobileagent routes for data fusion in distributed sensor networksrdquoIEEE Transactions on Knowledge and Data Engineering vol 16no 6 pp 740ndash753 2004

[18] C-L Fok G-C Roman and C Lu ldquoMobile agent middlewarefor sensor networks an application case studyrdquo in Proceedingsof the 4th International Symposium on Information Processing inSensor Networks (IPSN rsquo05) pp 382ndash387 April 2005

[19] A Boulis C Han andM B Srivastava ldquoDesign and implemen-tation of a framework for efficient and programmable sensornetworksrdquo in Proceedings of the 1st international conference onMobile systems applications and services ACM San FranciscoCalif USA 2003

[20] P Levis and D Culler ldquoMate a tiny virtual machine for sensornetworksrdquo ACM Sigplan Notices vol 37 no 10 pp 85ndash95 2002

[21] S Suenaga N Yoshioka and S Honiden ldquoGroup migration bymobile agents in wireless sensor networksrdquo Computer Journalvol 54 no 3 pp 345ndash355 2011

[22] P Bellavista Y Cai and T Magedanz ldquoRecent advances inmobile middleware for wireless systems and servicesrdquo MobileNetworks and Applications vol 16 no 3 pp 267ndash269 2011

[23] Z Zhirou and C Wengang ldquoModel and optimization ofmobile agentrsquos migration in gridrdquo in Proceedings of the 2ndInternational Conference on Bio-Inspired Computing Theoriesand Applications (BICTA rsquo07) pp 127ndash130 September 2007

[24] J-P Zhang M Jun J Yang and L-L Cheng ldquoAn improvedmigration strategy of mobile agentrdquo in Proceedings of the4th International Conference on Fuzzy Systems and KnowledgeDiscovery (FSKD rsquo07) pp 577ndash581 August 2007

[25] Y Lee and K Kim ldquoOptimal migration path searching usingpath adjustment and reassignment for mobile agentrdquo in Pro-ceedings of the 4th International Conference on NetworkedComputing and Advanced InformationManagement (NCM rsquo08)pp 564ndash569 Gyeongju South Korea September 2008

[26] M Tentori M Rodrıguez and J Favela Vara ldquoAn agent-basedmiddleware for the design of activity-aware applicationsrdquo IEEEIntelligent Systems vol 26 no 3 pp 15ndash23 2011

International Journal of

AerospaceEngineeringHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

RoboticsJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Active and Passive Electronic Components

Control Scienceand Engineering

Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

International Journal of

RotatingMachinery

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporation httpwwwhindawicom

Journal ofEngineeringVolume 2014

Submit your manuscripts athttpwwwhindawicom

VLSI Design

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Shock and Vibration

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Civil EngineeringAdvances in

Acoustics and VibrationAdvances in

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Electrical and Computer Engineering

Journal of

Advances inOptoElectronics

Hindawi Publishing Corporation httpwwwhindawicom

Volume 2014

The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014

SensorsJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Modelling amp Simulation in EngineeringHindawi Publishing Corporation httpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Chemical EngineeringInternational Journal of Antennas and

Propagation

International Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Navigation and Observation

International Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

DistributedSensor Networks

International Journal of

International Journal of Distributed Sensor Networks 5

422The SpecificMigration Procedure of DMLA Themobileagent migrates to the alive node which has not been reachedand with the minimal migration overheads according to (7)In this way we can gain a table close to the optimalmigrationIf all nodes have been reached mobile agent will return thenode that distributes it And when the mobile agent hasacquired enough information it will return to the CIS

Since each node periodically receives beacon frames fromneighbor nodes DMLA can always use the updated localvariables to determine the next hop As only using localvariables sensor nodes can avoid the nodewhich has a greatertransmission distance so part of the transmission energycan be saved In addition when some nodes suddenly loseefficacy the whole migration process will not make mistakesSoDMLAalgorithm is a fault-tolerant algorithmThe specificmigration steps of DMLA are as follows

Algorithm 2 (DMLA)

(1) At time 119905 = 0 the node that has first detectedinformation of target entity is denoted as cluster headit generates a mobile agent

(2) At time 119905 = 0 calculate location information oftarget entity 119883(0) according to the target locationalgorithmCalculate the amount of information collection 119868

0

according to (2) and set 119878119906119898119860119903119903119886119910 = 1198680

Select the next hop according to (8)(3) At time 119905 when agent reaches node 119896 the location of

target entity is 119883(119905) according to the target locationalgorithmCalculate the amount of information collection ofnode 119896 119868

119896 and update 119878119906119898119860119903119903119886119910 = 119878119906119898119860119903119903119886119910 + 119868

119896

If 119878119906119898119860119903119903119886119910 is greater than the threshold then go tostep 4 otherwise continue to migrate to the next hopaccording to (8) update time 119905 = 119905+1 and then repeatstep 3

(4) Return to cluster head to process data

43 pid-DMLA

431 The Proposition of pid-DMLA In the cooperationprocessing applications the movement of target entity isone of the important factors that affect the final result [14ndash19] DMLA algorithm does not take the direction of themovement of target entity into account So in some cases itcannot select the optimumnext hop For example in Figure 2at time 119905 the mobile agent is located in node 119860 Nodes 119861

and 119862 are neighbors of 119860 and they are equidistant to 119860According to DMLA mobile agent calculates node 119861 as thenext hop This is because node 119861 is nearer to the targetentity than node 119862 Then at time 119905 + 1 the mobile agent islocated in node 119861 But the target entity moves to a positioncloser to node 119862 Thus DMLA on the contrary selectsthe node which is farther from the target and has smalleramount of information collection and measurement value

t t + 1

A A

B B

CC

Figure 2 The movement of the target entity

as the next hop Therefore when designing agent migrationalgorithms we need to focus on predicting the movement ofthe target entity and selecting the node with the maximuminformation And we will propose the pid-DMLA algorithmto solve the problems of the movement of the target entity inthe migration process

432 Agent Migration Model in pid-DMLA Suppose thatthe moving direction and speed of the target entity do notchange Then the change of the target entityrsquos displacementis a constant at each identical time interval as follows

119883 (119905 + 1) minus 119883 (119905) = 119883 (119905) minus 119883 (119905 minus 1) (10)

Equivalently

119883 (119905 + 1) = 2119883 (119905) minus 119883 (119905 minus 1) (11)

As illustrated in (11) the position information119883(119905 + 1) attime 119905 + 1 can be deduced from the positions at time 119905 minus 1 andtime 119905 In this way the location information119883(119905minus1) and119883(119905)

can be calculated by target location algorithm and accordingto (11) the position of target entity119883(119905+1) can be estimatedat time 119905 + 1 as follows

119883 (119905 + 1) = 2119883 (119905) minus 119883 (119905 minus 1) (12)

Update the overhead function in (7) and get

119862119894119895(119905) = 119886

10038171003817100381710038171003817119883119894minus 119883119895

10038171003817100381710038171003817

2

1198892max+ 119887

10038171003817100381710038171003817119883 (119905 + 1) minus 119883

119895

10038171003817100381710038171003817

2

1198891199052max

+ radic1 minus 119886 minus 119887(1 minus

119890119895(119905)

119890max)

(13)

where 119883(119905 + 1) minus 119883119895 represents the distance between node

119895 and the target entity at time 119905 + 1And the next hop is

119895lowast

= argmin

119886

10038171003817100381710038171003817119883119894minus 119883119895

10038171003817100381710038171003817

2

1198892max+ 119887

10038171003817100381710038171003817119883 (119905 + 1) minus 119883

119895

10038171003817100381710038171003817

2

1198891199052max

+radic1 minus 119886 minus 119887(1 minus

119890119895(119905)

119890max)

119895 isin 119873119894

(14)

6 International Journal of Distributed Sensor Networks

t

t

t + 1

t + 1

t + 2

t + 2

t minus 1

t minus 1

AB DC

Figure 3 The migration of pid-DMLA

At time 119905 = 0 since the position of the target entity atthe previous time is unknown the amount of informationcollection can be calculated by the DMLA At time 119905 gt 0mobile agent migrates to the node closer to the estimatedposition of the target entity So it can obtain larger sensingmeasurement and information collection which is the coreof pid-DMLA

433 The Specific Migration Procedure of pid-DMLA Theprocess of mobile agent migration of pid-DMLA is shown inFigure 3 In a period of time the target entity moves alongthe dotted line and mobile agent migrates from node 119860 to119863

through 119861 and 119862 As shown in Figure 3 mobile agent alwaysmigrates to the node nearest to the next position of the targetentity

The specific migration procedure of pid-DMLA is asfollows

Algorithm 3 (pid-DMLA)

(1) At time 119905 = 0 the node that has first detectedinformation of target entity is denoted as cluster head

it generates a mobile agent

(2) At time 119905 = 0 calculate location information oftarget entity 119883(0) according to the target locationalgorithm

Calculate the amount of information collection 1198680

according to (2) and set 119878119906119898119860119903119903119886119910 = 1198680

Select the next hop according to (14)

(3) At time 119905 when agent reaches node 119896 the location oftarget entity is 119883(119905) according to the target locationalgorithm

Calculate the amount of information collection ofnode 119896 119868

119896 and update 119878119906119898119860119903119903119886119910 = 119878119906119898119860119903119903119886119910 + 119868

119896

If 119878119906119898119860119903119903119886119910 is greater than the threshold then goto step 4 otherwise continue to migrate to the nexthop according to (14) update time 119905 = 119905 + 1 and thenrepeat step 3

(4) Return to cluster head to process data

5 Migration Algorithm Based on MultipleMobile Agents

51 The Architecture of Mpid-DMLA Many researchers haveproposed and realized feasible ways to organize themigrationof a single mobile agent such as Agilla program [18] andSensorWare program [19] Consider the functional require-ments of these programs Firstly they have not formed anarchitecture to manage the agent migration mechanismsin WSN Secondly they lack migration algorithms whena tracking application needs the collaboration of multipleagents So when dealing with multiple agentsrsquo migration weneed a specific agent to manage the applications

Therefore we will propose Mpid-DMLA to solve thisproblem

Suppose that multiple target entity applications aredeployed in an area of sensors each application is responsiblefor monitoring a part of the area and composed of oneor more mobile agents Therefore we have the followingdefinitions

Definition 4 (code segment application) A centralized pro-gram which consists of multiple mobile agents deployed inthe WSN to accomplish a specific task is called centralizedcode segment application of multiple agents which is calledcode segment application for short

Code segment applications are associated with sensingdata that need to be evaluated and they need to cooperate andinteract based on the need of programs On the other handaccording to the Deluge amp Mate model [20] code segmentapplications need to redeploy themselves so that they can bedistributed by the CIS

To track targets code segment applications require thefollowing four functions [21]

(1) Sensing Sensor nodes are distributed around thetarget node and they test and collect information ofthe surrounding environment

(2) Processing Information After acquiring sensor infor-mation most of the nodes send it to the CIS forprocessing while cluster heads will directly processinformation

(3) Communication In order tomaintain the applicationsensor nodes need to keep exchanging informationwith other nodes or CIS

(4) Migration Mobile agents migrate based on the needof code segment application

Constructing the architecture of multiple mobile agentsalso requires the following definitions

Definition 5 (main component agent) The main componentagent is responsible for managingmigration information andactivity cycle of all component agents processing sensingdata of other agents and collecting and exchanging theinformation of subcomponent agents Each code segmentapplication has one and only main component agent

International Journal of Distributed Sensor Networks 7

Definition 6 (subcomponent agent) The subcomponentagent is used to obtain the data of nodes and report them tothe main component agent Each code segment applicationhas one or more subcomponent agents

Definition 7 (communication component agent) The com-munication component agent is used to collect and exchangethe entity information of WSN by migrating Each codesegment application has one or more communication com-ponent agents

Definition 8 (middleware software) The middleware soft-ware located between the physical and the operating systemlayers is the software entity running in mobile agents

52 The Migration Model of Mobile Agent in Mpid-DMLAWhen a code segment application changes monitoring posi-tions all of the mobile agents in this application needto migrate together to reconstruct a new code segmentapplication In Agilla the main component agent controlsthe entire process by sending the migration sequence whileother mobile agents can communicate by accessing theindependent tuple space [22] And the engine virtualmachineworks as the kernel to provide services and controls all of theconcurrent execution of all of the agents in a node Howeverwith the increase of the mobile agents the possibilities ofthe loss of packet and the failure of migration will increasetoo Suppose that a code segment application is composedof 119873 mobile agents Define 119875 as the failure coefficient of asingle agent And then we can obtain the success coefficientof migration in the entire application as follows

119875success = (1 minus 119875)119873

(15)

Communication overheads based on pid-DMLA aregiven as follows

119862119894119895(119905) = [

[

119886

10038171003817100381710038171003817119883119894minus 119883119895

10038171003817100381710038171003817

2

1198892max+ 119887

10038171003817100381710038171003817119883 (119905 + 1) minus 119883

119895

10038171003817100381710038171003817

2

1198891199052max

+radic1 minus 119886 minus 119887(1 minus

119890119895(119905)

119890max)] times (119875success)

minus1

(16)

And the next hop is

119895lowast

= argmin

[

[

119886

10038171003817100381710038171003817119883119894minus 119883119895

10038171003817100381710038171003817

2

1198892max+ 119887

10038171003817100381710038171003817119883 (119905 + 1) minus 119883

119895

10038171003817100381710038171003817

2

1198891199052max

+radic1 minus 119886 minus 119887(1 minus

119890119895(119905)

119890max)]

]

times (119875success)minus1

(17)

53 The Specific Migration Procedure of Mpid-DMLA In theconstruction of the target entity tracking we assume thatsubcomponent agent is a small program of lightweight codewhich is suitable for migration And it can independently

Figure 4 Migration based on predicting information drive ofmultiple mobile agents

access the target entityrsquos location As shown in Figure 4hollow circles represent sensor nodes which are distributedin the sensing region code segment is composed of one maincomponent agent (solid circle) four subcomponent agents(solid squares) and one communication component agent(solid triangle) Each agent runs on one node executingcode segment applications The migration of code segmentapplication means the migration of the six mobile agentswhich is the migration of multiple mobile agents [19] Thefollowing algorithm can be used to describe the above-mentioned procedure

Algorithm 9 (Mpid-DMLA)

(1) A main component agent a communication com-ponent agent and several subcomponent agents aredeployed in different sensor nodes

(2) Sensor nodes are monitoring around the target entityThe node that has first detected information of targetentity is denoted as cluster head themain componentagent then migrates to it

(3) At time 119905 = 0 calculate location information oftarget entity 119883(0) according to the target locationalgorithmCalculate the amount of information collection 119868

0

according to (2)Define the threshold of the amount of information as119879ℎ119903119890119904ℎ119900119897119889 119880119875 and let 119879ℎ119903119890119904ℎ119900119897119889 119880119875 = 119868

0

The main component agent generates new com-munication component agent and transfers 119868

0and

119879ℎ119903119890119904ℎ119900119897119889 119880119875 to itThen the communication component agent sets tomigrate

(4) At time 119905 when agent reaches node 119896 the location oftarget entity is 119883(119905) according to the target locationalgorithmCalculate the amount of information collectionof node 119896 119868

119896 and update 119879ℎ119903119890119904ℎ119900119897119889 119880119875 =

119879ℎ119903119890119904ℎ119900119897119889 119880119875 + 119868119896

If 119879ℎ119903119890119904ℎ119900119897119889 119880119875 is greater than the threshold thengo to step 5 otherwise select the next hop formigration according to (17) update time 119905 = 119905 + 1and then repeat step 3

8 International Journal of Distributed Sensor Networks

t

t

t + 1

t + 1

t + 3 t + 4 t + 5 t + 6 t + 7t + 2

t + 3

t + 4

t + 5

t + 6

t + 7

t + 2

Figure 5 The migration process of SMLA

t t + 1 t + 3 t + 4 t + 5 t + 6 t + 7t + 2

tt + 1 t + 3

t + 4

t + 5

t + 6t + 7

t + 2

Figure 6 The migration process of DMLA

(5) Communication component agent stores the infor-mation of the migration path (119868

0(119905 = 0) 119868

1(119905 =

1) 119868119896(119905 = 119896) 119883 (119905 = 0) 119883 (119905 = 1) 119883 (119905 =

119896)) and other relevant information

And it interacts with the main component agent

(6) The main component agent generates the migrationsequence 119873

1 1198732 119873

119896 according to location of

target entity

Send it to each subcomponent agent (this step isoptional for different applications)

(7) The main component agent sends complete instruc-tions to the lower-layered middleware software

Method functions which are predefined in the mid-dleware software recycle the communication compo-nent agent and the subcomponent agents

(8) The main component agent goes to the destinationnode based on the migration sequence

If the migration fails send failure information toCIS to terminate this migration otherwise generatenew communication component agent and subcom-ponent agents and this migration ends

6 Simulation and Analyses

61 The Effect of Single Mobile Agent in Target TrackingBefore the simulation experiments we firstly use targettracking as a sample to demonstrate the migration effect ofa single mobile agent of different migration algorithms Asshown in Figures 5 6 and 7 suppose that a car is runningfrom left to right through the area covered byWSNThe smallcircle represents the sensor node solid small circle representsthe cluster head which distributes agents large dotted circlerepresents the communication range of sensor node and eachnode selects the next hop only within its communicationrange The above-mentioned figures illustrate the specificlocations of the car and agent from time 119905 to time 119905 + 7

The migration process of SMLA is shown in Figure 5 Attime 119905 CIS distributes the mobile agent to the node whichis the nearest to the target entity according to the prede-termined migration sequence When the object is stationaryor its motion can be estimated this static approach is verysuitable However it loses efficacy in some conditions Forexample if the object remains in the location at time 119905+1 thenthe migration sequence will be the positions that are aroundthe target entity But once the target entity starts to move notaccording to the predetermination SMLA will fail

DMLA can in some sense optimize the performanceof SMLA As shown in Figure 6 mobile agent tracks themovement of the target entity However the movement of

International Journal of Distributed Sensor Networks 9

t

t + 1 t + 3

t + 4

t + 5

t + 6t + 7

t + 2

t t + 1 t + 3 t + 4 t + 5 t + 6 t + 7t + 2

Figure 7 The migration process of pid-DMLA

Table 1 Basic simulation parameters of single mobile agent migra-tion algorithms

Network area 500 times 500mNumber of nodes 200 or 800Nodersquos initial energy 50 JNodersquos sensing radius 5mEnergy consumption for data processing 03 JEnergy consumption for sending beacon frames 01 JThe interval of node broadcasting beacon frames 01 sThe speed of the target entity 0sim50msThe threshold of the amount of information 18The size of a mobile agent 800 bitsSimulation time 100 s

the target entity is unpredictable so it is not accurate to decidemigration path only by the amount of information which ishowever related to the distance to the target entity

The pid-DMLA can predict movement information oftarget entity dynamically which is a better solution to suchproblems As shown in Figure 7 at time 119905 + 5 mobile agentselects the next hop according to the predicted informationwhich is different from DMLA In brief pid-DMLA rebuildsmigration sequence based on predicting themovement of thetarget entity to achieve the purpose of tracking

62 Performance Analysis on the Single Mobile Agent Migra-tion Algorithms We use simulation software NS2 to eval-uate the three single mobile agent migration algorithmsrsquoperformances The default simulation parameters are shownin Table 1

Suppose that once the agent migrates one hop all sensornodes in the network consume 02 J For SMLA the transmis-sion distance is set to 30m while for DMLA and pid-DMLAthe transmission distance is set to be 10m Therefore theconsumption of transmission for SMLA is three times morethan that of sendingThe reason for this setting is that SMLAneeds bigger transmission range to collect information Thedirection and size of movement of the target entity areconstant denoted as V (ms) Once the target entity reachesthe edge of the sensing area it reverses direction immediatelywhich ensures its being always within the sensing area

0 5 10 15 20 25 30 35 40 45 500

2

4

6

8

10

12

Target speed (ms)

Ener

gy co

nsum

ptio

n (J

)

SMLADMLApid-DMLA

Figure 8 Energy consumption of the single agent migrationalgorithms in Scenario 1

Scenario 1 Set the number of nodes as 200 the target entityspeed as 0sim50ms and the simulation time as 100 s whichcan guarantee a complete migration The simulation resultsare shown in Figures 8 and 9 It is obvious that DMLA andpid-DMLA consume less energy than SMLA in the wholenetwork This is because the communication cost of SMLAis higher than that of DMLA But with the increasing speedof the target entity the effect of SMLA algorithm in trackingthe target entity becomes less and less obvious thereforemigration may terminate and the energy consumption onthe contrary becomes slower On the other hand when thespeed of target entity is very low there are no significantdifferences among the three migration schemes

Scenario 2 Set the speed of target entity as 5ms and thenumber of nodes as 800 Generally the coverage and thenumber of nodes are positively correlated The simulationresults are shown in Figures 10 and 11 It is clear that pid-DMLA has the lowest energy consumption when the numberof nodes is more than 100 This is because with the increaseof the nodes the choices of predicting the next hop become

10 International Journal of Distributed Sensor Networks

0102030405060708090

100

Life

time (

s)

0 5 10 15 20 25 30 35 40 45 50Target speed (ms)

SMLADMLApid-DMLA

Figure 9 Lifetime of the single agent migration algorithms inScenario 1

Ener

gy co

nsum

ptio

n (J

)

SMLADMLApid-DMLA

0 100 200 300 400 500 600 700 8000

1

2

3

4

5

6

7

8

9

Target speed (ms)

Figure 10 Energy consumption of the single agent migrationalgorithms in Scenario 2

more as a result the overheads become less [23] When thenumber of nodes is less than 50 the performance of SMLAis better than those of the other two This is because withless nodes static and dynamic migration algorithms share themore similar migration path however the migration tablehas been predetermined in SMLA thus it on the contraryhas the least overheads On the other hand when the densityof nodes is high there would be more neighbors and moremigration choices Thus DMLA and pid-DMLA outperformSMLA

63 Multiple Mobile Agents Migration Algorithm PerformanceAnalysis Before the assessment of Mpid-DMLA we need toconsider two problems [24 25] The first is the effect of code

SMLADMLApid-DMLA

10

20

30

40

50

60

70

80

Life

time (

s)

0 100 200 300 400 500 600 700 800Target speed (ms)

Figure 11 Lifetime of the single agent migration algorithms inScenario 2

segment applications on the amount of communication inthe migration process To ensure the comprehensiveness ofthe evaluation set the number of communication componentagents as 1 and the number of subcomponent agents as 5 10or 20 and evaluate Mpid-DMLA based on code segmentswith different scales The second is the effect of differentenvironment and running platform in practical operation onthe amount of information and the success rate of migrationTherefore we only consider the relation of the number ofcommunication and energy instead of the influence of theagentrsquos size

The performance of the migration of mobile agent inAgilla system has been evaluated in [19] First Agilla systemis not designed to support multiple mobile agents specificallywhich means that it is not aware of the number of the agentsin the code segment application And the systemrsquos overheadsare related to the size ofmobile agent So inAgilla all kinds ofagents are considered to be identical and there is no discrim-ination between the main and communication componentagents However this does not affect the simulation resultsbecause all agents are uniformly assumed in the simulationin this paper

The purpose of multiple mobile agents migration isto complete the entire application migration with limitedenergy The default simulation parameters are shown inTable 2

And when main component agent recycles and sendsmobile agents suppose that the energy consumption of thenode where main component agent stays is neglected whilethat of other nodes is 005 J

Figures 12 13 and 14 illustrate the relation between thenumber of application code segments and the success rateof one hop migration when the numbers of subcomponentagents are 5 10 and 15 respectivelyThe success rate ofMpid-DMLA is higher when the number of subcomponent agentsis 5 than those when the numbers of subcomponent agents

International Journal of Distributed Sensor Networks 11

Table 2 Basic simulation algorithm parameters of a single mobileagent migration

Network area 100 times 100mNumber of nodes 7 times 7Number of main component agents per codesegment application 1

Number of subcomponent agents per code segmentapplication 5 10 or 15

Number of communication component agents percode segment application 1

Nodersquos initial energy 2 JEnergy consumption of a node when an agentreaches it 005 J

Interaction consumption when an agent migrates 01 J

20

30

40

50

60

70

80

90

100

Mpid-DMLAAgilla

2 25 3 35 45 554 5 6Number of migration applications

Succ

ess r

ate (

)

Figure 12 The relation between the number of code segmentapplications and the success rate of migration (I)

are 10 and 20 It is obvious that the successful rate of Mpid-DMLA is generally not less than 50 which is much largerthan that of Agilla

Moreover Figure 15 demonstrates the relationship be-tween themigration hops and the amount of communicationwhen the subcomponent agents are 5 As shown in Figure 15with the samemigration hops the amount of communicationof Mpid-DMLA is much less than that of Agilla As the num-ber of data packet reflects the amount of energy consumptionMpid-DMLA consumes less energy than Agilla

Although the simulation results show that Mpid-DMLAis a suitable migration scheme for multiple mobile agents itbases on some specific assumptions In general this schemestill encounters the following three issues

(1) When the main component agent loses efficacy inthe process of migration code segment applicationscannot be rebuilt

(2) When subcomponent agents lose efficacy afterdeployment code segment applications will keepwaiting because the applications cannot be recycled

20

30

40

50

60

70

80

90

100

Mpid-DMLAAgilla

2 25 3 35 45 554 5 6Number of migration applications

Succ

ess r

ate (

)

Figure 13 The relation between the number of code segmentapplications and the success rate of migration (II)

10

20

30

40

50

60

70

80

0

Mpid-DMLAAgilla

2 25 3 35 45 554 5 6Number of migration applications

Succ

ess r

ate (

)

Figure 14 The relation between the number of code segmentapplications and the success rate of migration (III)

1 15 2 25 3 35 420

40

60

80

100

120

140

160

180

200

Hops of main agent

Num

ber o

f dat

a pac

kets

Mpid-DMLAAgilla

Figure 15The relation between themigration hops and the amountof communication

12 International Journal of Distributed Sensor Networks

(3) If the main and the subcomponent agents which loseefficacy in issue 1 or 2 are left in the nodes withoutbeing recycled then othermobile agents cannot reachthese nodes

Therefore in designing middleware software it is nec-essary to add the feedback mechanism of main componentagent and it needs to periodically return ACK informationto the higher layer of the middleware software When themiddleware finds that the main component agent losesefficacy it needs to distribute a new one or a specializedindependent agent to recycle the useless agents in the wholenetwork

In Agilla the method of recycling the useless agentsis by adding the unified interface into all of the agentsRemoving all of the mobile agents will affect the function ofcode segment applications Therefore developing the moreappropriate middleware software is the future work whichis also related to the development of the better migrationstrategies [26]

7 Conclusion

As big data is the intensification of information architec-ture while distributed computing is the intensification ofinfrastructure and resource distributed computing is anappropriate method to deal with big data In this paperwe have firstly illustrated the mobile agent model in thedistributed computing And it led to the two considerablefactors of migration the selection of nodes and of thesequence of them Then we have proposed the assumptionsof the migration algorithms Moreover we have designedthe nodersquos sensor model and information collection modelFurthermore we have formulated the beacon frames ofmobile agent for communication Generally mobile agentmigration algorithms can be transferred as the target trackingissue Thus we have described a simple target localizationalgorithm applicable for migration On this basis we havegeneralized the whole migration process of the mobile agent

Considering the single mobile agent we have proposedthe static SMLA and the dynamic DMLA algorithms Afteranalyzing the drawbacks of SMLA and DMLA we thenproposed the pid-DMLA algorithm Section 4 has demon-strated the modeling procedure of the abovementioned threealgorithms Section 6 has emulated and evaluated their per-formances and studied their advantages and disadvantages indifferent scenarios

Considering the multiple mobile agents applied in prac-tical target tracking we have also put forward the Mpid-DMLA algorithm In the process of mobile agent componentconstructing code segment applications we defined threemobile agents with different functions to cooperate with eachother to accomplish the migration Section 5 illustrated thearchitecture modeling procedure and basic process ofMpid-DMLA Section 6 compared it with Agilla and evaluated itsadvantages and disadvantages

Acknowledgments

The paper is sponsored by the National Natural ScienceFoundation of China (61170065 61171053 61003039 6110019961003236 61202355 61201163 and 61373138) the Natu-ral Science Foundation of Jiangsu Province (BK2011755BK2011072 and BK2012436) the Scientific amp Technolog-ical Support Project of Jiangsu Province (BE2012183 andBE2012755) the Natural Science Key Fund for Collegesand Universities in Jiangsu Province (12KJA520002) theScientific Research amp Industry Promotion Project for HigherEducation Institutions (JHB2012-7) the Postdoctoral Foun-dation (2012M511753 1101011B 1202034C and 2013T60536)the Doctoral Fund of Ministry of Education of China(20113223110002 and 20123223120006) the Fund of NanjingUniversity of Posts and Telecommunications (NY212047)and the project funded by the Priority Academic Pro-gram Development of Jiangsu Higher Education Institutions(yx002001)

References

[1] Research Trends special issue on big data httpwwwresearch-trendscomwp-contentuploads201209Research Trends Is-sue30pdf 2012

[2] L Ismail and D Hagimont ldquoA performance evaluation of themobile agent paradigmrdquoACM SIGPLANNotices vol 34 no 10pp 306ndash313 1999

[3] AMainwaring J Polastre R Szewczyk D Culler and J Ander-son ldquoWireless sensor networks for habitat monitoringrdquo inProceedings of the 1st ACM International Workshop on WirelessSensor Networks and Applications pp 88ndash97 September 2002

[4] IVY-A sensor network infrastructure for the college of engi-neering httpwww-bsaceecsberkeleyeduprojectsivy 2008

[5] I F AkyildizW Su Y Sankarasubramaniam and E Cayirci ldquoAsurvey on sensor networksrdquo IEEE Communications Magazinevol 40 no 8 pp 102ndash105 2002

[6] Y-C Tseng S-P Kuo H-W Lee and C-F Huang ldquoLocationtracking in a wireless sensor network by mobile agents and itsdata fusion strategiesrdquoComputer Journal vol 47 no 4 pp 448ndash460 2004

[7] Y Xu and H Qi ldquoMobile agent migration modeling anddesign for target tracking in wireless sensor networksrdquo Ad HocNetworks vol 6 no 1 pp 1ndash16 2008

[8] Y Xu and H Qi ldquoDistributed computing paradigms for collab-orative signal and information processing in sensor networksrdquoJournal of Parallel and Distributed Computing vol 64 no 8 pp945ndash959 2004

[9] F Zhao J Shin and J Reich ldquoInformation-driven dynamicsensor collaborationrdquo IEEE Signal Processing Magazine vol 19no 2 pp 61ndash72 2002

[10] J Liu J Reich and F Zhao ldquoCollaborative in-network pro-cessing for target trackingrdquo Eurasip Journal on Applied SignalProcessing vol 2003 no 4 pp 378ndash391 2003

[11] J-W Kim J-S In K Hur J-W Kim and D-S Eom ldquoAnintelligent agent-based routing structure for mobile sinks inWSNsrdquo IEEE Transactions on Consumer Electronics vol 56 no4 pp 2310ndash2316 2010

[12] H Wang G Pottie K Yao and D Estrin ldquoEntropy-basedsensor selection heuristic for target localizationrdquo in Proceedings

International Journal of Distributed Sensor Networks 13

of the 3rd International Symposium on Information Processing inSensor Networks (IPSN rsquo04) pp 36ndash45 ACM Press April 2004

[13] T S RappaportWireless Communications Principles and Prac-tice IEEE Press 1996

[14] H Mousannif and I Khalil ldquoCooperative data management inwireless sensor networksrdquo International Journal of Computa-tional Intelligence Systems vol 5 no 3 pp 403ndash412 2012

[15] N Labraoui M Guerroui M Aliouat and T Zia ldquoDataaggregation security challenge in wireless sensor networks asurveyrdquo Ad-Hoc and Sensor Wireless Networks vol 12 no 3-4pp 295ndash324 2011

[16] M Chen L T Yang T Kwon L Zhou and M Jo ldquoItineraryplanning for energy-efficient agent communications in wirelesssensor networksrdquo IEEE Transactions on Vehicular Technologyvol 60 no 7 pp 3290ndash3299 2011

[17] Q Wu N S V Rao J Barhen et al ldquoOn computing mobileagent routes for data fusion in distributed sensor networksrdquoIEEE Transactions on Knowledge and Data Engineering vol 16no 6 pp 740ndash753 2004

[18] C-L Fok G-C Roman and C Lu ldquoMobile agent middlewarefor sensor networks an application case studyrdquo in Proceedingsof the 4th International Symposium on Information Processing inSensor Networks (IPSN rsquo05) pp 382ndash387 April 2005

[19] A Boulis C Han andM B Srivastava ldquoDesign and implemen-tation of a framework for efficient and programmable sensornetworksrdquo in Proceedings of the 1st international conference onMobile systems applications and services ACM San FranciscoCalif USA 2003

[20] P Levis and D Culler ldquoMate a tiny virtual machine for sensornetworksrdquo ACM Sigplan Notices vol 37 no 10 pp 85ndash95 2002

[21] S Suenaga N Yoshioka and S Honiden ldquoGroup migration bymobile agents in wireless sensor networksrdquo Computer Journalvol 54 no 3 pp 345ndash355 2011

[22] P Bellavista Y Cai and T Magedanz ldquoRecent advances inmobile middleware for wireless systems and servicesrdquo MobileNetworks and Applications vol 16 no 3 pp 267ndash269 2011

[23] Z Zhirou and C Wengang ldquoModel and optimization ofmobile agentrsquos migration in gridrdquo in Proceedings of the 2ndInternational Conference on Bio-Inspired Computing Theoriesand Applications (BICTA rsquo07) pp 127ndash130 September 2007

[24] J-P Zhang M Jun J Yang and L-L Cheng ldquoAn improvedmigration strategy of mobile agentrdquo in Proceedings of the4th International Conference on Fuzzy Systems and KnowledgeDiscovery (FSKD rsquo07) pp 577ndash581 August 2007

[25] Y Lee and K Kim ldquoOptimal migration path searching usingpath adjustment and reassignment for mobile agentrdquo in Pro-ceedings of the 4th International Conference on NetworkedComputing and Advanced InformationManagement (NCM rsquo08)pp 564ndash569 Gyeongju South Korea September 2008

[26] M Tentori M Rodrıguez and J Favela Vara ldquoAn agent-basedmiddleware for the design of activity-aware applicationsrdquo IEEEIntelligent Systems vol 26 no 3 pp 15ndash23 2011

International Journal of

AerospaceEngineeringHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

RoboticsJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Active and Passive Electronic Components

Control Scienceand Engineering

Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

International Journal of

RotatingMachinery

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporation httpwwwhindawicom

Journal ofEngineeringVolume 2014

Submit your manuscripts athttpwwwhindawicom

VLSI Design

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Shock and Vibration

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Civil EngineeringAdvances in

Acoustics and VibrationAdvances in

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Electrical and Computer Engineering

Journal of

Advances inOptoElectronics

Hindawi Publishing Corporation httpwwwhindawicom

Volume 2014

The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014

SensorsJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Modelling amp Simulation in EngineeringHindawi Publishing Corporation httpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Chemical EngineeringInternational Journal of Antennas and

Propagation

International Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Navigation and Observation

International Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

DistributedSensor Networks

International Journal of

6 International Journal of Distributed Sensor Networks

t

t

t + 1

t + 1

t + 2

t + 2

t minus 1

t minus 1

AB DC

Figure 3 The migration of pid-DMLA

At time 119905 = 0 since the position of the target entity atthe previous time is unknown the amount of informationcollection can be calculated by the DMLA At time 119905 gt 0mobile agent migrates to the node closer to the estimatedposition of the target entity So it can obtain larger sensingmeasurement and information collection which is the coreof pid-DMLA

433 The Specific Migration Procedure of pid-DMLA Theprocess of mobile agent migration of pid-DMLA is shown inFigure 3 In a period of time the target entity moves alongthe dotted line and mobile agent migrates from node 119860 to119863

through 119861 and 119862 As shown in Figure 3 mobile agent alwaysmigrates to the node nearest to the next position of the targetentity

The specific migration procedure of pid-DMLA is asfollows

Algorithm 3 (pid-DMLA)

(1) At time 119905 = 0 the node that has first detectedinformation of target entity is denoted as cluster head

it generates a mobile agent

(2) At time 119905 = 0 calculate location information oftarget entity 119883(0) according to the target locationalgorithm

Calculate the amount of information collection 1198680

according to (2) and set 119878119906119898119860119903119903119886119910 = 1198680

Select the next hop according to (14)

(3) At time 119905 when agent reaches node 119896 the location oftarget entity is 119883(119905) according to the target locationalgorithm

Calculate the amount of information collection ofnode 119896 119868

119896 and update 119878119906119898119860119903119903119886119910 = 119878119906119898119860119903119903119886119910 + 119868

119896

If 119878119906119898119860119903119903119886119910 is greater than the threshold then goto step 4 otherwise continue to migrate to the nexthop according to (14) update time 119905 = 119905 + 1 and thenrepeat step 3

(4) Return to cluster head to process data

5 Migration Algorithm Based on MultipleMobile Agents

51 The Architecture of Mpid-DMLA Many researchers haveproposed and realized feasible ways to organize themigrationof a single mobile agent such as Agilla program [18] andSensorWare program [19] Consider the functional require-ments of these programs Firstly they have not formed anarchitecture to manage the agent migration mechanismsin WSN Secondly they lack migration algorithms whena tracking application needs the collaboration of multipleagents So when dealing with multiple agentsrsquo migration weneed a specific agent to manage the applications

Therefore we will propose Mpid-DMLA to solve thisproblem

Suppose that multiple target entity applications aredeployed in an area of sensors each application is responsiblefor monitoring a part of the area and composed of oneor more mobile agents Therefore we have the followingdefinitions

Definition 4 (code segment application) A centralized pro-gram which consists of multiple mobile agents deployed inthe WSN to accomplish a specific task is called centralizedcode segment application of multiple agents which is calledcode segment application for short

Code segment applications are associated with sensingdata that need to be evaluated and they need to cooperate andinteract based on the need of programs On the other handaccording to the Deluge amp Mate model [20] code segmentapplications need to redeploy themselves so that they can bedistributed by the CIS

To track targets code segment applications require thefollowing four functions [21]

(1) Sensing Sensor nodes are distributed around thetarget node and they test and collect information ofthe surrounding environment

(2) Processing Information After acquiring sensor infor-mation most of the nodes send it to the CIS forprocessing while cluster heads will directly processinformation

(3) Communication In order tomaintain the applicationsensor nodes need to keep exchanging informationwith other nodes or CIS

(4) Migration Mobile agents migrate based on the needof code segment application

Constructing the architecture of multiple mobile agentsalso requires the following definitions

Definition 5 (main component agent) The main componentagent is responsible for managingmigration information andactivity cycle of all component agents processing sensingdata of other agents and collecting and exchanging theinformation of subcomponent agents Each code segmentapplication has one and only main component agent

International Journal of Distributed Sensor Networks 7

Definition 6 (subcomponent agent) The subcomponentagent is used to obtain the data of nodes and report them tothe main component agent Each code segment applicationhas one or more subcomponent agents

Definition 7 (communication component agent) The com-munication component agent is used to collect and exchangethe entity information of WSN by migrating Each codesegment application has one or more communication com-ponent agents

Definition 8 (middleware software) The middleware soft-ware located between the physical and the operating systemlayers is the software entity running in mobile agents

52 The Migration Model of Mobile Agent in Mpid-DMLAWhen a code segment application changes monitoring posi-tions all of the mobile agents in this application needto migrate together to reconstruct a new code segmentapplication In Agilla the main component agent controlsthe entire process by sending the migration sequence whileother mobile agents can communicate by accessing theindependent tuple space [22] And the engine virtualmachineworks as the kernel to provide services and controls all of theconcurrent execution of all of the agents in a node Howeverwith the increase of the mobile agents the possibilities ofthe loss of packet and the failure of migration will increasetoo Suppose that a code segment application is composedof 119873 mobile agents Define 119875 as the failure coefficient of asingle agent And then we can obtain the success coefficientof migration in the entire application as follows

119875success = (1 minus 119875)119873

(15)

Communication overheads based on pid-DMLA aregiven as follows

119862119894119895(119905) = [

[

119886

10038171003817100381710038171003817119883119894minus 119883119895

10038171003817100381710038171003817

2

1198892max+ 119887

10038171003817100381710038171003817119883 (119905 + 1) minus 119883

119895

10038171003817100381710038171003817

2

1198891199052max

+radic1 minus 119886 minus 119887(1 minus

119890119895(119905)

119890max)] times (119875success)

minus1

(16)

And the next hop is

119895lowast

= argmin

[

[

119886

10038171003817100381710038171003817119883119894minus 119883119895

10038171003817100381710038171003817

2

1198892max+ 119887

10038171003817100381710038171003817119883 (119905 + 1) minus 119883

119895

10038171003817100381710038171003817

2

1198891199052max

+radic1 minus 119886 minus 119887(1 minus

119890119895(119905)

119890max)]

]

times (119875success)minus1

(17)

53 The Specific Migration Procedure of Mpid-DMLA In theconstruction of the target entity tracking we assume thatsubcomponent agent is a small program of lightweight codewhich is suitable for migration And it can independently

Figure 4 Migration based on predicting information drive ofmultiple mobile agents

access the target entityrsquos location As shown in Figure 4hollow circles represent sensor nodes which are distributedin the sensing region code segment is composed of one maincomponent agent (solid circle) four subcomponent agents(solid squares) and one communication component agent(solid triangle) Each agent runs on one node executingcode segment applications The migration of code segmentapplication means the migration of the six mobile agentswhich is the migration of multiple mobile agents [19] Thefollowing algorithm can be used to describe the above-mentioned procedure

Algorithm 9 (Mpid-DMLA)

(1) A main component agent a communication com-ponent agent and several subcomponent agents aredeployed in different sensor nodes

(2) Sensor nodes are monitoring around the target entityThe node that has first detected information of targetentity is denoted as cluster head themain componentagent then migrates to it

(3) At time 119905 = 0 calculate location information oftarget entity 119883(0) according to the target locationalgorithmCalculate the amount of information collection 119868

0

according to (2)Define the threshold of the amount of information as119879ℎ119903119890119904ℎ119900119897119889 119880119875 and let 119879ℎ119903119890119904ℎ119900119897119889 119880119875 = 119868

0

The main component agent generates new com-munication component agent and transfers 119868

0and

119879ℎ119903119890119904ℎ119900119897119889 119880119875 to itThen the communication component agent sets tomigrate

(4) At time 119905 when agent reaches node 119896 the location oftarget entity is 119883(119905) according to the target locationalgorithmCalculate the amount of information collectionof node 119896 119868

119896 and update 119879ℎ119903119890119904ℎ119900119897119889 119880119875 =

119879ℎ119903119890119904ℎ119900119897119889 119880119875 + 119868119896

If 119879ℎ119903119890119904ℎ119900119897119889 119880119875 is greater than the threshold thengo to step 5 otherwise select the next hop formigration according to (17) update time 119905 = 119905 + 1and then repeat step 3

8 International Journal of Distributed Sensor Networks

t

t

t + 1

t + 1

t + 3 t + 4 t + 5 t + 6 t + 7t + 2

t + 3

t + 4

t + 5

t + 6

t + 7

t + 2

Figure 5 The migration process of SMLA

t t + 1 t + 3 t + 4 t + 5 t + 6 t + 7t + 2

tt + 1 t + 3

t + 4

t + 5

t + 6t + 7

t + 2

Figure 6 The migration process of DMLA

(5) Communication component agent stores the infor-mation of the migration path (119868

0(119905 = 0) 119868

1(119905 =

1) 119868119896(119905 = 119896) 119883 (119905 = 0) 119883 (119905 = 1) 119883 (119905 =

119896)) and other relevant information

And it interacts with the main component agent

(6) The main component agent generates the migrationsequence 119873

1 1198732 119873

119896 according to location of

target entity

Send it to each subcomponent agent (this step isoptional for different applications)

(7) The main component agent sends complete instruc-tions to the lower-layered middleware software

Method functions which are predefined in the mid-dleware software recycle the communication compo-nent agent and the subcomponent agents

(8) The main component agent goes to the destinationnode based on the migration sequence

If the migration fails send failure information toCIS to terminate this migration otherwise generatenew communication component agent and subcom-ponent agents and this migration ends

6 Simulation and Analyses

61 The Effect of Single Mobile Agent in Target TrackingBefore the simulation experiments we firstly use targettracking as a sample to demonstrate the migration effect ofa single mobile agent of different migration algorithms Asshown in Figures 5 6 and 7 suppose that a car is runningfrom left to right through the area covered byWSNThe smallcircle represents the sensor node solid small circle representsthe cluster head which distributes agents large dotted circlerepresents the communication range of sensor node and eachnode selects the next hop only within its communicationrange The above-mentioned figures illustrate the specificlocations of the car and agent from time 119905 to time 119905 + 7

The migration process of SMLA is shown in Figure 5 Attime 119905 CIS distributes the mobile agent to the node whichis the nearest to the target entity according to the prede-termined migration sequence When the object is stationaryor its motion can be estimated this static approach is verysuitable However it loses efficacy in some conditions Forexample if the object remains in the location at time 119905+1 thenthe migration sequence will be the positions that are aroundthe target entity But once the target entity starts to move notaccording to the predetermination SMLA will fail

DMLA can in some sense optimize the performanceof SMLA As shown in Figure 6 mobile agent tracks themovement of the target entity However the movement of

International Journal of Distributed Sensor Networks 9

t

t + 1 t + 3

t + 4

t + 5

t + 6t + 7

t + 2

t t + 1 t + 3 t + 4 t + 5 t + 6 t + 7t + 2

Figure 7 The migration process of pid-DMLA

Table 1 Basic simulation parameters of single mobile agent migra-tion algorithms

Network area 500 times 500mNumber of nodes 200 or 800Nodersquos initial energy 50 JNodersquos sensing radius 5mEnergy consumption for data processing 03 JEnergy consumption for sending beacon frames 01 JThe interval of node broadcasting beacon frames 01 sThe speed of the target entity 0sim50msThe threshold of the amount of information 18The size of a mobile agent 800 bitsSimulation time 100 s

the target entity is unpredictable so it is not accurate to decidemigration path only by the amount of information which ishowever related to the distance to the target entity

The pid-DMLA can predict movement information oftarget entity dynamically which is a better solution to suchproblems As shown in Figure 7 at time 119905 + 5 mobile agentselects the next hop according to the predicted informationwhich is different from DMLA In brief pid-DMLA rebuildsmigration sequence based on predicting themovement of thetarget entity to achieve the purpose of tracking

62 Performance Analysis on the Single Mobile Agent Migra-tion Algorithms We use simulation software NS2 to eval-uate the three single mobile agent migration algorithmsrsquoperformances The default simulation parameters are shownin Table 1

Suppose that once the agent migrates one hop all sensornodes in the network consume 02 J For SMLA the transmis-sion distance is set to 30m while for DMLA and pid-DMLAthe transmission distance is set to be 10m Therefore theconsumption of transmission for SMLA is three times morethan that of sendingThe reason for this setting is that SMLAneeds bigger transmission range to collect information Thedirection and size of movement of the target entity areconstant denoted as V (ms) Once the target entity reachesthe edge of the sensing area it reverses direction immediatelywhich ensures its being always within the sensing area

0 5 10 15 20 25 30 35 40 45 500

2

4

6

8

10

12

Target speed (ms)

Ener

gy co

nsum

ptio

n (J

)

SMLADMLApid-DMLA

Figure 8 Energy consumption of the single agent migrationalgorithms in Scenario 1

Scenario 1 Set the number of nodes as 200 the target entityspeed as 0sim50ms and the simulation time as 100 s whichcan guarantee a complete migration The simulation resultsare shown in Figures 8 and 9 It is obvious that DMLA andpid-DMLA consume less energy than SMLA in the wholenetwork This is because the communication cost of SMLAis higher than that of DMLA But with the increasing speedof the target entity the effect of SMLA algorithm in trackingthe target entity becomes less and less obvious thereforemigration may terminate and the energy consumption onthe contrary becomes slower On the other hand when thespeed of target entity is very low there are no significantdifferences among the three migration schemes

Scenario 2 Set the speed of target entity as 5ms and thenumber of nodes as 800 Generally the coverage and thenumber of nodes are positively correlated The simulationresults are shown in Figures 10 and 11 It is clear that pid-DMLA has the lowest energy consumption when the numberof nodes is more than 100 This is because with the increaseof the nodes the choices of predicting the next hop become

10 International Journal of Distributed Sensor Networks

0102030405060708090

100

Life

time (

s)

0 5 10 15 20 25 30 35 40 45 50Target speed (ms)

SMLADMLApid-DMLA

Figure 9 Lifetime of the single agent migration algorithms inScenario 1

Ener

gy co

nsum

ptio

n (J

)

SMLADMLApid-DMLA

0 100 200 300 400 500 600 700 8000

1

2

3

4

5

6

7

8

9

Target speed (ms)

Figure 10 Energy consumption of the single agent migrationalgorithms in Scenario 2

more as a result the overheads become less [23] When thenumber of nodes is less than 50 the performance of SMLAis better than those of the other two This is because withless nodes static and dynamic migration algorithms share themore similar migration path however the migration tablehas been predetermined in SMLA thus it on the contraryhas the least overheads On the other hand when the densityof nodes is high there would be more neighbors and moremigration choices Thus DMLA and pid-DMLA outperformSMLA

63 Multiple Mobile Agents Migration Algorithm PerformanceAnalysis Before the assessment of Mpid-DMLA we need toconsider two problems [24 25] The first is the effect of code

SMLADMLApid-DMLA

10

20

30

40

50

60

70

80

Life

time (

s)

0 100 200 300 400 500 600 700 800Target speed (ms)

Figure 11 Lifetime of the single agent migration algorithms inScenario 2

segment applications on the amount of communication inthe migration process To ensure the comprehensiveness ofthe evaluation set the number of communication componentagents as 1 and the number of subcomponent agents as 5 10or 20 and evaluate Mpid-DMLA based on code segmentswith different scales The second is the effect of differentenvironment and running platform in practical operation onthe amount of information and the success rate of migrationTherefore we only consider the relation of the number ofcommunication and energy instead of the influence of theagentrsquos size

The performance of the migration of mobile agent inAgilla system has been evaluated in [19] First Agilla systemis not designed to support multiple mobile agents specificallywhich means that it is not aware of the number of the agentsin the code segment application And the systemrsquos overheadsare related to the size ofmobile agent So inAgilla all kinds ofagents are considered to be identical and there is no discrim-ination between the main and communication componentagents However this does not affect the simulation resultsbecause all agents are uniformly assumed in the simulationin this paper

The purpose of multiple mobile agents migration isto complete the entire application migration with limitedenergy The default simulation parameters are shown inTable 2

And when main component agent recycles and sendsmobile agents suppose that the energy consumption of thenode where main component agent stays is neglected whilethat of other nodes is 005 J

Figures 12 13 and 14 illustrate the relation between thenumber of application code segments and the success rateof one hop migration when the numbers of subcomponentagents are 5 10 and 15 respectivelyThe success rate ofMpid-DMLA is higher when the number of subcomponent agentsis 5 than those when the numbers of subcomponent agents

International Journal of Distributed Sensor Networks 11

Table 2 Basic simulation algorithm parameters of a single mobileagent migration

Network area 100 times 100mNumber of nodes 7 times 7Number of main component agents per codesegment application 1

Number of subcomponent agents per code segmentapplication 5 10 or 15

Number of communication component agents percode segment application 1

Nodersquos initial energy 2 JEnergy consumption of a node when an agentreaches it 005 J

Interaction consumption when an agent migrates 01 J

20

30

40

50

60

70

80

90

100

Mpid-DMLAAgilla

2 25 3 35 45 554 5 6Number of migration applications

Succ

ess r

ate (

)

Figure 12 The relation between the number of code segmentapplications and the success rate of migration (I)

are 10 and 20 It is obvious that the successful rate of Mpid-DMLA is generally not less than 50 which is much largerthan that of Agilla

Moreover Figure 15 demonstrates the relationship be-tween themigration hops and the amount of communicationwhen the subcomponent agents are 5 As shown in Figure 15with the samemigration hops the amount of communicationof Mpid-DMLA is much less than that of Agilla As the num-ber of data packet reflects the amount of energy consumptionMpid-DMLA consumes less energy than Agilla

Although the simulation results show that Mpid-DMLAis a suitable migration scheme for multiple mobile agents itbases on some specific assumptions In general this schemestill encounters the following three issues

(1) When the main component agent loses efficacy inthe process of migration code segment applicationscannot be rebuilt

(2) When subcomponent agents lose efficacy afterdeployment code segment applications will keepwaiting because the applications cannot be recycled

20

30

40

50

60

70

80

90

100

Mpid-DMLAAgilla

2 25 3 35 45 554 5 6Number of migration applications

Succ

ess r

ate (

)

Figure 13 The relation between the number of code segmentapplications and the success rate of migration (II)

10

20

30

40

50

60

70

80

0

Mpid-DMLAAgilla

2 25 3 35 45 554 5 6Number of migration applications

Succ

ess r

ate (

)

Figure 14 The relation between the number of code segmentapplications and the success rate of migration (III)

1 15 2 25 3 35 420

40

60

80

100

120

140

160

180

200

Hops of main agent

Num

ber o

f dat

a pac

kets

Mpid-DMLAAgilla

Figure 15The relation between themigration hops and the amountof communication

12 International Journal of Distributed Sensor Networks

(3) If the main and the subcomponent agents which loseefficacy in issue 1 or 2 are left in the nodes withoutbeing recycled then othermobile agents cannot reachthese nodes

Therefore in designing middleware software it is nec-essary to add the feedback mechanism of main componentagent and it needs to periodically return ACK informationto the higher layer of the middleware software When themiddleware finds that the main component agent losesefficacy it needs to distribute a new one or a specializedindependent agent to recycle the useless agents in the wholenetwork

In Agilla the method of recycling the useless agentsis by adding the unified interface into all of the agentsRemoving all of the mobile agents will affect the function ofcode segment applications Therefore developing the moreappropriate middleware software is the future work whichis also related to the development of the better migrationstrategies [26]

7 Conclusion

As big data is the intensification of information architec-ture while distributed computing is the intensification ofinfrastructure and resource distributed computing is anappropriate method to deal with big data In this paperwe have firstly illustrated the mobile agent model in thedistributed computing And it led to the two considerablefactors of migration the selection of nodes and of thesequence of them Then we have proposed the assumptionsof the migration algorithms Moreover we have designedthe nodersquos sensor model and information collection modelFurthermore we have formulated the beacon frames ofmobile agent for communication Generally mobile agentmigration algorithms can be transferred as the target trackingissue Thus we have described a simple target localizationalgorithm applicable for migration On this basis we havegeneralized the whole migration process of the mobile agent

Considering the single mobile agent we have proposedthe static SMLA and the dynamic DMLA algorithms Afteranalyzing the drawbacks of SMLA and DMLA we thenproposed the pid-DMLA algorithm Section 4 has demon-strated the modeling procedure of the abovementioned threealgorithms Section 6 has emulated and evaluated their per-formances and studied their advantages and disadvantages indifferent scenarios

Considering the multiple mobile agents applied in prac-tical target tracking we have also put forward the Mpid-DMLA algorithm In the process of mobile agent componentconstructing code segment applications we defined threemobile agents with different functions to cooperate with eachother to accomplish the migration Section 5 illustrated thearchitecture modeling procedure and basic process ofMpid-DMLA Section 6 compared it with Agilla and evaluated itsadvantages and disadvantages

Acknowledgments

The paper is sponsored by the National Natural ScienceFoundation of China (61170065 61171053 61003039 6110019961003236 61202355 61201163 and 61373138) the Natu-ral Science Foundation of Jiangsu Province (BK2011755BK2011072 and BK2012436) the Scientific amp Technolog-ical Support Project of Jiangsu Province (BE2012183 andBE2012755) the Natural Science Key Fund for Collegesand Universities in Jiangsu Province (12KJA520002) theScientific Research amp Industry Promotion Project for HigherEducation Institutions (JHB2012-7) the Postdoctoral Foun-dation (2012M511753 1101011B 1202034C and 2013T60536)the Doctoral Fund of Ministry of Education of China(20113223110002 and 20123223120006) the Fund of NanjingUniversity of Posts and Telecommunications (NY212047)and the project funded by the Priority Academic Pro-gram Development of Jiangsu Higher Education Institutions(yx002001)

References

[1] Research Trends special issue on big data httpwwwresearch-trendscomwp-contentuploads201209Research Trends Is-sue30pdf 2012

[2] L Ismail and D Hagimont ldquoA performance evaluation of themobile agent paradigmrdquoACM SIGPLANNotices vol 34 no 10pp 306ndash313 1999

[3] AMainwaring J Polastre R Szewczyk D Culler and J Ander-son ldquoWireless sensor networks for habitat monitoringrdquo inProceedings of the 1st ACM International Workshop on WirelessSensor Networks and Applications pp 88ndash97 September 2002

[4] IVY-A sensor network infrastructure for the college of engi-neering httpwww-bsaceecsberkeleyeduprojectsivy 2008

[5] I F AkyildizW Su Y Sankarasubramaniam and E Cayirci ldquoAsurvey on sensor networksrdquo IEEE Communications Magazinevol 40 no 8 pp 102ndash105 2002

[6] Y-C Tseng S-P Kuo H-W Lee and C-F Huang ldquoLocationtracking in a wireless sensor network by mobile agents and itsdata fusion strategiesrdquoComputer Journal vol 47 no 4 pp 448ndash460 2004

[7] Y Xu and H Qi ldquoMobile agent migration modeling anddesign for target tracking in wireless sensor networksrdquo Ad HocNetworks vol 6 no 1 pp 1ndash16 2008

[8] Y Xu and H Qi ldquoDistributed computing paradigms for collab-orative signal and information processing in sensor networksrdquoJournal of Parallel and Distributed Computing vol 64 no 8 pp945ndash959 2004

[9] F Zhao J Shin and J Reich ldquoInformation-driven dynamicsensor collaborationrdquo IEEE Signal Processing Magazine vol 19no 2 pp 61ndash72 2002

[10] J Liu J Reich and F Zhao ldquoCollaborative in-network pro-cessing for target trackingrdquo Eurasip Journal on Applied SignalProcessing vol 2003 no 4 pp 378ndash391 2003

[11] J-W Kim J-S In K Hur J-W Kim and D-S Eom ldquoAnintelligent agent-based routing structure for mobile sinks inWSNsrdquo IEEE Transactions on Consumer Electronics vol 56 no4 pp 2310ndash2316 2010

[12] H Wang G Pottie K Yao and D Estrin ldquoEntropy-basedsensor selection heuristic for target localizationrdquo in Proceedings

International Journal of Distributed Sensor Networks 13

of the 3rd International Symposium on Information Processing inSensor Networks (IPSN rsquo04) pp 36ndash45 ACM Press April 2004

[13] T S RappaportWireless Communications Principles and Prac-tice IEEE Press 1996

[14] H Mousannif and I Khalil ldquoCooperative data management inwireless sensor networksrdquo International Journal of Computa-tional Intelligence Systems vol 5 no 3 pp 403ndash412 2012

[15] N Labraoui M Guerroui M Aliouat and T Zia ldquoDataaggregation security challenge in wireless sensor networks asurveyrdquo Ad-Hoc and Sensor Wireless Networks vol 12 no 3-4pp 295ndash324 2011

[16] M Chen L T Yang T Kwon L Zhou and M Jo ldquoItineraryplanning for energy-efficient agent communications in wirelesssensor networksrdquo IEEE Transactions on Vehicular Technologyvol 60 no 7 pp 3290ndash3299 2011

[17] Q Wu N S V Rao J Barhen et al ldquoOn computing mobileagent routes for data fusion in distributed sensor networksrdquoIEEE Transactions on Knowledge and Data Engineering vol 16no 6 pp 740ndash753 2004

[18] C-L Fok G-C Roman and C Lu ldquoMobile agent middlewarefor sensor networks an application case studyrdquo in Proceedingsof the 4th International Symposium on Information Processing inSensor Networks (IPSN rsquo05) pp 382ndash387 April 2005

[19] A Boulis C Han andM B Srivastava ldquoDesign and implemen-tation of a framework for efficient and programmable sensornetworksrdquo in Proceedings of the 1st international conference onMobile systems applications and services ACM San FranciscoCalif USA 2003

[20] P Levis and D Culler ldquoMate a tiny virtual machine for sensornetworksrdquo ACM Sigplan Notices vol 37 no 10 pp 85ndash95 2002

[21] S Suenaga N Yoshioka and S Honiden ldquoGroup migration bymobile agents in wireless sensor networksrdquo Computer Journalvol 54 no 3 pp 345ndash355 2011

[22] P Bellavista Y Cai and T Magedanz ldquoRecent advances inmobile middleware for wireless systems and servicesrdquo MobileNetworks and Applications vol 16 no 3 pp 267ndash269 2011

[23] Z Zhirou and C Wengang ldquoModel and optimization ofmobile agentrsquos migration in gridrdquo in Proceedings of the 2ndInternational Conference on Bio-Inspired Computing Theoriesand Applications (BICTA rsquo07) pp 127ndash130 September 2007

[24] J-P Zhang M Jun J Yang and L-L Cheng ldquoAn improvedmigration strategy of mobile agentrdquo in Proceedings of the4th International Conference on Fuzzy Systems and KnowledgeDiscovery (FSKD rsquo07) pp 577ndash581 August 2007

[25] Y Lee and K Kim ldquoOptimal migration path searching usingpath adjustment and reassignment for mobile agentrdquo in Pro-ceedings of the 4th International Conference on NetworkedComputing and Advanced InformationManagement (NCM rsquo08)pp 564ndash569 Gyeongju South Korea September 2008

[26] M Tentori M Rodrıguez and J Favela Vara ldquoAn agent-basedmiddleware for the design of activity-aware applicationsrdquo IEEEIntelligent Systems vol 26 no 3 pp 15ndash23 2011

International Journal of

AerospaceEngineeringHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

RoboticsJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Active and Passive Electronic Components

Control Scienceand Engineering

Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

International Journal of

RotatingMachinery

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporation httpwwwhindawicom

Journal ofEngineeringVolume 2014

Submit your manuscripts athttpwwwhindawicom

VLSI Design

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Shock and Vibration

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Civil EngineeringAdvances in

Acoustics and VibrationAdvances in

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Electrical and Computer Engineering

Journal of

Advances inOptoElectronics

Hindawi Publishing Corporation httpwwwhindawicom

Volume 2014

The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014

SensorsJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Modelling amp Simulation in EngineeringHindawi Publishing Corporation httpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Chemical EngineeringInternational Journal of Antennas and

Propagation

International Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Navigation and Observation

International Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

DistributedSensor Networks

International Journal of

International Journal of Distributed Sensor Networks 7

Definition 6 (subcomponent agent) The subcomponentagent is used to obtain the data of nodes and report them tothe main component agent Each code segment applicationhas one or more subcomponent agents

Definition 7 (communication component agent) The com-munication component agent is used to collect and exchangethe entity information of WSN by migrating Each codesegment application has one or more communication com-ponent agents

Definition 8 (middleware software) The middleware soft-ware located between the physical and the operating systemlayers is the software entity running in mobile agents

52 The Migration Model of Mobile Agent in Mpid-DMLAWhen a code segment application changes monitoring posi-tions all of the mobile agents in this application needto migrate together to reconstruct a new code segmentapplication In Agilla the main component agent controlsthe entire process by sending the migration sequence whileother mobile agents can communicate by accessing theindependent tuple space [22] And the engine virtualmachineworks as the kernel to provide services and controls all of theconcurrent execution of all of the agents in a node Howeverwith the increase of the mobile agents the possibilities ofthe loss of packet and the failure of migration will increasetoo Suppose that a code segment application is composedof 119873 mobile agents Define 119875 as the failure coefficient of asingle agent And then we can obtain the success coefficientof migration in the entire application as follows

119875success = (1 minus 119875)119873

(15)

Communication overheads based on pid-DMLA aregiven as follows

119862119894119895(119905) = [

[

119886

10038171003817100381710038171003817119883119894minus 119883119895

10038171003817100381710038171003817

2

1198892max+ 119887

10038171003817100381710038171003817119883 (119905 + 1) minus 119883

119895

10038171003817100381710038171003817

2

1198891199052max

+radic1 minus 119886 minus 119887(1 minus

119890119895(119905)

119890max)] times (119875success)

minus1

(16)

And the next hop is

119895lowast

= argmin

[

[

119886

10038171003817100381710038171003817119883119894minus 119883119895

10038171003817100381710038171003817

2

1198892max+ 119887

10038171003817100381710038171003817119883 (119905 + 1) minus 119883

119895

10038171003817100381710038171003817

2

1198891199052max

+radic1 minus 119886 minus 119887(1 minus

119890119895(119905)

119890max)]

]

times (119875success)minus1

(17)

53 The Specific Migration Procedure of Mpid-DMLA In theconstruction of the target entity tracking we assume thatsubcomponent agent is a small program of lightweight codewhich is suitable for migration And it can independently

Figure 4 Migration based on predicting information drive ofmultiple mobile agents

access the target entityrsquos location As shown in Figure 4hollow circles represent sensor nodes which are distributedin the sensing region code segment is composed of one maincomponent agent (solid circle) four subcomponent agents(solid squares) and one communication component agent(solid triangle) Each agent runs on one node executingcode segment applications The migration of code segmentapplication means the migration of the six mobile agentswhich is the migration of multiple mobile agents [19] Thefollowing algorithm can be used to describe the above-mentioned procedure

Algorithm 9 (Mpid-DMLA)

(1) A main component agent a communication com-ponent agent and several subcomponent agents aredeployed in different sensor nodes

(2) Sensor nodes are monitoring around the target entityThe node that has first detected information of targetentity is denoted as cluster head themain componentagent then migrates to it

(3) At time 119905 = 0 calculate location information oftarget entity 119883(0) according to the target locationalgorithmCalculate the amount of information collection 119868

0

according to (2)Define the threshold of the amount of information as119879ℎ119903119890119904ℎ119900119897119889 119880119875 and let 119879ℎ119903119890119904ℎ119900119897119889 119880119875 = 119868

0

The main component agent generates new com-munication component agent and transfers 119868

0and

119879ℎ119903119890119904ℎ119900119897119889 119880119875 to itThen the communication component agent sets tomigrate

(4) At time 119905 when agent reaches node 119896 the location oftarget entity is 119883(119905) according to the target locationalgorithmCalculate the amount of information collectionof node 119896 119868

119896 and update 119879ℎ119903119890119904ℎ119900119897119889 119880119875 =

119879ℎ119903119890119904ℎ119900119897119889 119880119875 + 119868119896

If 119879ℎ119903119890119904ℎ119900119897119889 119880119875 is greater than the threshold thengo to step 5 otherwise select the next hop formigration according to (17) update time 119905 = 119905 + 1and then repeat step 3

8 International Journal of Distributed Sensor Networks

t

t

t + 1

t + 1

t + 3 t + 4 t + 5 t + 6 t + 7t + 2

t + 3

t + 4

t + 5

t + 6

t + 7

t + 2

Figure 5 The migration process of SMLA

t t + 1 t + 3 t + 4 t + 5 t + 6 t + 7t + 2

tt + 1 t + 3

t + 4

t + 5

t + 6t + 7

t + 2

Figure 6 The migration process of DMLA

(5) Communication component agent stores the infor-mation of the migration path (119868

0(119905 = 0) 119868

1(119905 =

1) 119868119896(119905 = 119896) 119883 (119905 = 0) 119883 (119905 = 1) 119883 (119905 =

119896)) and other relevant information

And it interacts with the main component agent

(6) The main component agent generates the migrationsequence 119873

1 1198732 119873

119896 according to location of

target entity

Send it to each subcomponent agent (this step isoptional for different applications)

(7) The main component agent sends complete instruc-tions to the lower-layered middleware software

Method functions which are predefined in the mid-dleware software recycle the communication compo-nent agent and the subcomponent agents

(8) The main component agent goes to the destinationnode based on the migration sequence

If the migration fails send failure information toCIS to terminate this migration otherwise generatenew communication component agent and subcom-ponent agents and this migration ends

6 Simulation and Analyses

61 The Effect of Single Mobile Agent in Target TrackingBefore the simulation experiments we firstly use targettracking as a sample to demonstrate the migration effect ofa single mobile agent of different migration algorithms Asshown in Figures 5 6 and 7 suppose that a car is runningfrom left to right through the area covered byWSNThe smallcircle represents the sensor node solid small circle representsthe cluster head which distributes agents large dotted circlerepresents the communication range of sensor node and eachnode selects the next hop only within its communicationrange The above-mentioned figures illustrate the specificlocations of the car and agent from time 119905 to time 119905 + 7

The migration process of SMLA is shown in Figure 5 Attime 119905 CIS distributes the mobile agent to the node whichis the nearest to the target entity according to the prede-termined migration sequence When the object is stationaryor its motion can be estimated this static approach is verysuitable However it loses efficacy in some conditions Forexample if the object remains in the location at time 119905+1 thenthe migration sequence will be the positions that are aroundthe target entity But once the target entity starts to move notaccording to the predetermination SMLA will fail

DMLA can in some sense optimize the performanceof SMLA As shown in Figure 6 mobile agent tracks themovement of the target entity However the movement of

International Journal of Distributed Sensor Networks 9

t

t + 1 t + 3

t + 4

t + 5

t + 6t + 7

t + 2

t t + 1 t + 3 t + 4 t + 5 t + 6 t + 7t + 2

Figure 7 The migration process of pid-DMLA

Table 1 Basic simulation parameters of single mobile agent migra-tion algorithms

Network area 500 times 500mNumber of nodes 200 or 800Nodersquos initial energy 50 JNodersquos sensing radius 5mEnergy consumption for data processing 03 JEnergy consumption for sending beacon frames 01 JThe interval of node broadcasting beacon frames 01 sThe speed of the target entity 0sim50msThe threshold of the amount of information 18The size of a mobile agent 800 bitsSimulation time 100 s

the target entity is unpredictable so it is not accurate to decidemigration path only by the amount of information which ishowever related to the distance to the target entity

The pid-DMLA can predict movement information oftarget entity dynamically which is a better solution to suchproblems As shown in Figure 7 at time 119905 + 5 mobile agentselects the next hop according to the predicted informationwhich is different from DMLA In brief pid-DMLA rebuildsmigration sequence based on predicting themovement of thetarget entity to achieve the purpose of tracking

62 Performance Analysis on the Single Mobile Agent Migra-tion Algorithms We use simulation software NS2 to eval-uate the three single mobile agent migration algorithmsrsquoperformances The default simulation parameters are shownin Table 1

Suppose that once the agent migrates one hop all sensornodes in the network consume 02 J For SMLA the transmis-sion distance is set to 30m while for DMLA and pid-DMLAthe transmission distance is set to be 10m Therefore theconsumption of transmission for SMLA is three times morethan that of sendingThe reason for this setting is that SMLAneeds bigger transmission range to collect information Thedirection and size of movement of the target entity areconstant denoted as V (ms) Once the target entity reachesthe edge of the sensing area it reverses direction immediatelywhich ensures its being always within the sensing area

0 5 10 15 20 25 30 35 40 45 500

2

4

6

8

10

12

Target speed (ms)

Ener

gy co

nsum

ptio

n (J

)

SMLADMLApid-DMLA

Figure 8 Energy consumption of the single agent migrationalgorithms in Scenario 1

Scenario 1 Set the number of nodes as 200 the target entityspeed as 0sim50ms and the simulation time as 100 s whichcan guarantee a complete migration The simulation resultsare shown in Figures 8 and 9 It is obvious that DMLA andpid-DMLA consume less energy than SMLA in the wholenetwork This is because the communication cost of SMLAis higher than that of DMLA But with the increasing speedof the target entity the effect of SMLA algorithm in trackingthe target entity becomes less and less obvious thereforemigration may terminate and the energy consumption onthe contrary becomes slower On the other hand when thespeed of target entity is very low there are no significantdifferences among the three migration schemes

Scenario 2 Set the speed of target entity as 5ms and thenumber of nodes as 800 Generally the coverage and thenumber of nodes are positively correlated The simulationresults are shown in Figures 10 and 11 It is clear that pid-DMLA has the lowest energy consumption when the numberof nodes is more than 100 This is because with the increaseof the nodes the choices of predicting the next hop become

10 International Journal of Distributed Sensor Networks

0102030405060708090

100

Life

time (

s)

0 5 10 15 20 25 30 35 40 45 50Target speed (ms)

SMLADMLApid-DMLA

Figure 9 Lifetime of the single agent migration algorithms inScenario 1

Ener

gy co

nsum

ptio

n (J

)

SMLADMLApid-DMLA

0 100 200 300 400 500 600 700 8000

1

2

3

4

5

6

7

8

9

Target speed (ms)

Figure 10 Energy consumption of the single agent migrationalgorithms in Scenario 2

more as a result the overheads become less [23] When thenumber of nodes is less than 50 the performance of SMLAis better than those of the other two This is because withless nodes static and dynamic migration algorithms share themore similar migration path however the migration tablehas been predetermined in SMLA thus it on the contraryhas the least overheads On the other hand when the densityof nodes is high there would be more neighbors and moremigration choices Thus DMLA and pid-DMLA outperformSMLA

63 Multiple Mobile Agents Migration Algorithm PerformanceAnalysis Before the assessment of Mpid-DMLA we need toconsider two problems [24 25] The first is the effect of code

SMLADMLApid-DMLA

10

20

30

40

50

60

70

80

Life

time (

s)

0 100 200 300 400 500 600 700 800Target speed (ms)

Figure 11 Lifetime of the single agent migration algorithms inScenario 2

segment applications on the amount of communication inthe migration process To ensure the comprehensiveness ofthe evaluation set the number of communication componentagents as 1 and the number of subcomponent agents as 5 10or 20 and evaluate Mpid-DMLA based on code segmentswith different scales The second is the effect of differentenvironment and running platform in practical operation onthe amount of information and the success rate of migrationTherefore we only consider the relation of the number ofcommunication and energy instead of the influence of theagentrsquos size

The performance of the migration of mobile agent inAgilla system has been evaluated in [19] First Agilla systemis not designed to support multiple mobile agents specificallywhich means that it is not aware of the number of the agentsin the code segment application And the systemrsquos overheadsare related to the size ofmobile agent So inAgilla all kinds ofagents are considered to be identical and there is no discrim-ination between the main and communication componentagents However this does not affect the simulation resultsbecause all agents are uniformly assumed in the simulationin this paper

The purpose of multiple mobile agents migration isto complete the entire application migration with limitedenergy The default simulation parameters are shown inTable 2

And when main component agent recycles and sendsmobile agents suppose that the energy consumption of thenode where main component agent stays is neglected whilethat of other nodes is 005 J

Figures 12 13 and 14 illustrate the relation between thenumber of application code segments and the success rateof one hop migration when the numbers of subcomponentagents are 5 10 and 15 respectivelyThe success rate ofMpid-DMLA is higher when the number of subcomponent agentsis 5 than those when the numbers of subcomponent agents

International Journal of Distributed Sensor Networks 11

Table 2 Basic simulation algorithm parameters of a single mobileagent migration

Network area 100 times 100mNumber of nodes 7 times 7Number of main component agents per codesegment application 1

Number of subcomponent agents per code segmentapplication 5 10 or 15

Number of communication component agents percode segment application 1

Nodersquos initial energy 2 JEnergy consumption of a node when an agentreaches it 005 J

Interaction consumption when an agent migrates 01 J

20

30

40

50

60

70

80

90

100

Mpid-DMLAAgilla

2 25 3 35 45 554 5 6Number of migration applications

Succ

ess r

ate (

)

Figure 12 The relation between the number of code segmentapplications and the success rate of migration (I)

are 10 and 20 It is obvious that the successful rate of Mpid-DMLA is generally not less than 50 which is much largerthan that of Agilla

Moreover Figure 15 demonstrates the relationship be-tween themigration hops and the amount of communicationwhen the subcomponent agents are 5 As shown in Figure 15with the samemigration hops the amount of communicationof Mpid-DMLA is much less than that of Agilla As the num-ber of data packet reflects the amount of energy consumptionMpid-DMLA consumes less energy than Agilla

Although the simulation results show that Mpid-DMLAis a suitable migration scheme for multiple mobile agents itbases on some specific assumptions In general this schemestill encounters the following three issues

(1) When the main component agent loses efficacy inthe process of migration code segment applicationscannot be rebuilt

(2) When subcomponent agents lose efficacy afterdeployment code segment applications will keepwaiting because the applications cannot be recycled

20

30

40

50

60

70

80

90

100

Mpid-DMLAAgilla

2 25 3 35 45 554 5 6Number of migration applications

Succ

ess r

ate (

)

Figure 13 The relation between the number of code segmentapplications and the success rate of migration (II)

10

20

30

40

50

60

70

80

0

Mpid-DMLAAgilla

2 25 3 35 45 554 5 6Number of migration applications

Succ

ess r

ate (

)

Figure 14 The relation between the number of code segmentapplications and the success rate of migration (III)

1 15 2 25 3 35 420

40

60

80

100

120

140

160

180

200

Hops of main agent

Num

ber o

f dat

a pac

kets

Mpid-DMLAAgilla

Figure 15The relation between themigration hops and the amountof communication

12 International Journal of Distributed Sensor Networks

(3) If the main and the subcomponent agents which loseefficacy in issue 1 or 2 are left in the nodes withoutbeing recycled then othermobile agents cannot reachthese nodes

Therefore in designing middleware software it is nec-essary to add the feedback mechanism of main componentagent and it needs to periodically return ACK informationto the higher layer of the middleware software When themiddleware finds that the main component agent losesefficacy it needs to distribute a new one or a specializedindependent agent to recycle the useless agents in the wholenetwork

In Agilla the method of recycling the useless agentsis by adding the unified interface into all of the agentsRemoving all of the mobile agents will affect the function ofcode segment applications Therefore developing the moreappropriate middleware software is the future work whichis also related to the development of the better migrationstrategies [26]

7 Conclusion

As big data is the intensification of information architec-ture while distributed computing is the intensification ofinfrastructure and resource distributed computing is anappropriate method to deal with big data In this paperwe have firstly illustrated the mobile agent model in thedistributed computing And it led to the two considerablefactors of migration the selection of nodes and of thesequence of them Then we have proposed the assumptionsof the migration algorithms Moreover we have designedthe nodersquos sensor model and information collection modelFurthermore we have formulated the beacon frames ofmobile agent for communication Generally mobile agentmigration algorithms can be transferred as the target trackingissue Thus we have described a simple target localizationalgorithm applicable for migration On this basis we havegeneralized the whole migration process of the mobile agent

Considering the single mobile agent we have proposedthe static SMLA and the dynamic DMLA algorithms Afteranalyzing the drawbacks of SMLA and DMLA we thenproposed the pid-DMLA algorithm Section 4 has demon-strated the modeling procedure of the abovementioned threealgorithms Section 6 has emulated and evaluated their per-formances and studied their advantages and disadvantages indifferent scenarios

Considering the multiple mobile agents applied in prac-tical target tracking we have also put forward the Mpid-DMLA algorithm In the process of mobile agent componentconstructing code segment applications we defined threemobile agents with different functions to cooperate with eachother to accomplish the migration Section 5 illustrated thearchitecture modeling procedure and basic process ofMpid-DMLA Section 6 compared it with Agilla and evaluated itsadvantages and disadvantages

Acknowledgments

The paper is sponsored by the National Natural ScienceFoundation of China (61170065 61171053 61003039 6110019961003236 61202355 61201163 and 61373138) the Natu-ral Science Foundation of Jiangsu Province (BK2011755BK2011072 and BK2012436) the Scientific amp Technolog-ical Support Project of Jiangsu Province (BE2012183 andBE2012755) the Natural Science Key Fund for Collegesand Universities in Jiangsu Province (12KJA520002) theScientific Research amp Industry Promotion Project for HigherEducation Institutions (JHB2012-7) the Postdoctoral Foun-dation (2012M511753 1101011B 1202034C and 2013T60536)the Doctoral Fund of Ministry of Education of China(20113223110002 and 20123223120006) the Fund of NanjingUniversity of Posts and Telecommunications (NY212047)and the project funded by the Priority Academic Pro-gram Development of Jiangsu Higher Education Institutions(yx002001)

References

[1] Research Trends special issue on big data httpwwwresearch-trendscomwp-contentuploads201209Research Trends Is-sue30pdf 2012

[2] L Ismail and D Hagimont ldquoA performance evaluation of themobile agent paradigmrdquoACM SIGPLANNotices vol 34 no 10pp 306ndash313 1999

[3] AMainwaring J Polastre R Szewczyk D Culler and J Ander-son ldquoWireless sensor networks for habitat monitoringrdquo inProceedings of the 1st ACM International Workshop on WirelessSensor Networks and Applications pp 88ndash97 September 2002

[4] IVY-A sensor network infrastructure for the college of engi-neering httpwww-bsaceecsberkeleyeduprojectsivy 2008

[5] I F AkyildizW Su Y Sankarasubramaniam and E Cayirci ldquoAsurvey on sensor networksrdquo IEEE Communications Magazinevol 40 no 8 pp 102ndash105 2002

[6] Y-C Tseng S-P Kuo H-W Lee and C-F Huang ldquoLocationtracking in a wireless sensor network by mobile agents and itsdata fusion strategiesrdquoComputer Journal vol 47 no 4 pp 448ndash460 2004

[7] Y Xu and H Qi ldquoMobile agent migration modeling anddesign for target tracking in wireless sensor networksrdquo Ad HocNetworks vol 6 no 1 pp 1ndash16 2008

[8] Y Xu and H Qi ldquoDistributed computing paradigms for collab-orative signal and information processing in sensor networksrdquoJournal of Parallel and Distributed Computing vol 64 no 8 pp945ndash959 2004

[9] F Zhao J Shin and J Reich ldquoInformation-driven dynamicsensor collaborationrdquo IEEE Signal Processing Magazine vol 19no 2 pp 61ndash72 2002

[10] J Liu J Reich and F Zhao ldquoCollaborative in-network pro-cessing for target trackingrdquo Eurasip Journal on Applied SignalProcessing vol 2003 no 4 pp 378ndash391 2003

[11] J-W Kim J-S In K Hur J-W Kim and D-S Eom ldquoAnintelligent agent-based routing structure for mobile sinks inWSNsrdquo IEEE Transactions on Consumer Electronics vol 56 no4 pp 2310ndash2316 2010

[12] H Wang G Pottie K Yao and D Estrin ldquoEntropy-basedsensor selection heuristic for target localizationrdquo in Proceedings

International Journal of Distributed Sensor Networks 13

of the 3rd International Symposium on Information Processing inSensor Networks (IPSN rsquo04) pp 36ndash45 ACM Press April 2004

[13] T S RappaportWireless Communications Principles and Prac-tice IEEE Press 1996

[14] H Mousannif and I Khalil ldquoCooperative data management inwireless sensor networksrdquo International Journal of Computa-tional Intelligence Systems vol 5 no 3 pp 403ndash412 2012

[15] N Labraoui M Guerroui M Aliouat and T Zia ldquoDataaggregation security challenge in wireless sensor networks asurveyrdquo Ad-Hoc and Sensor Wireless Networks vol 12 no 3-4pp 295ndash324 2011

[16] M Chen L T Yang T Kwon L Zhou and M Jo ldquoItineraryplanning for energy-efficient agent communications in wirelesssensor networksrdquo IEEE Transactions on Vehicular Technologyvol 60 no 7 pp 3290ndash3299 2011

[17] Q Wu N S V Rao J Barhen et al ldquoOn computing mobileagent routes for data fusion in distributed sensor networksrdquoIEEE Transactions on Knowledge and Data Engineering vol 16no 6 pp 740ndash753 2004

[18] C-L Fok G-C Roman and C Lu ldquoMobile agent middlewarefor sensor networks an application case studyrdquo in Proceedingsof the 4th International Symposium on Information Processing inSensor Networks (IPSN rsquo05) pp 382ndash387 April 2005

[19] A Boulis C Han andM B Srivastava ldquoDesign and implemen-tation of a framework for efficient and programmable sensornetworksrdquo in Proceedings of the 1st international conference onMobile systems applications and services ACM San FranciscoCalif USA 2003

[20] P Levis and D Culler ldquoMate a tiny virtual machine for sensornetworksrdquo ACM Sigplan Notices vol 37 no 10 pp 85ndash95 2002

[21] S Suenaga N Yoshioka and S Honiden ldquoGroup migration bymobile agents in wireless sensor networksrdquo Computer Journalvol 54 no 3 pp 345ndash355 2011

[22] P Bellavista Y Cai and T Magedanz ldquoRecent advances inmobile middleware for wireless systems and servicesrdquo MobileNetworks and Applications vol 16 no 3 pp 267ndash269 2011

[23] Z Zhirou and C Wengang ldquoModel and optimization ofmobile agentrsquos migration in gridrdquo in Proceedings of the 2ndInternational Conference on Bio-Inspired Computing Theoriesand Applications (BICTA rsquo07) pp 127ndash130 September 2007

[24] J-P Zhang M Jun J Yang and L-L Cheng ldquoAn improvedmigration strategy of mobile agentrdquo in Proceedings of the4th International Conference on Fuzzy Systems and KnowledgeDiscovery (FSKD rsquo07) pp 577ndash581 August 2007

[25] Y Lee and K Kim ldquoOptimal migration path searching usingpath adjustment and reassignment for mobile agentrdquo in Pro-ceedings of the 4th International Conference on NetworkedComputing and Advanced InformationManagement (NCM rsquo08)pp 564ndash569 Gyeongju South Korea September 2008

[26] M Tentori M Rodrıguez and J Favela Vara ldquoAn agent-basedmiddleware for the design of activity-aware applicationsrdquo IEEEIntelligent Systems vol 26 no 3 pp 15ndash23 2011

International Journal of

AerospaceEngineeringHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

RoboticsJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Active and Passive Electronic Components

Control Scienceand Engineering

Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

International Journal of

RotatingMachinery

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporation httpwwwhindawicom

Journal ofEngineeringVolume 2014

Submit your manuscripts athttpwwwhindawicom

VLSI Design

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Shock and Vibration

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Civil EngineeringAdvances in

Acoustics and VibrationAdvances in

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Electrical and Computer Engineering

Journal of

Advances inOptoElectronics

Hindawi Publishing Corporation httpwwwhindawicom

Volume 2014

The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014

SensorsJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Modelling amp Simulation in EngineeringHindawi Publishing Corporation httpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Chemical EngineeringInternational Journal of Antennas and

Propagation

International Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Navigation and Observation

International Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

DistributedSensor Networks

International Journal of

8 International Journal of Distributed Sensor Networks

t

t

t + 1

t + 1

t + 3 t + 4 t + 5 t + 6 t + 7t + 2

t + 3

t + 4

t + 5

t + 6

t + 7

t + 2

Figure 5 The migration process of SMLA

t t + 1 t + 3 t + 4 t + 5 t + 6 t + 7t + 2

tt + 1 t + 3

t + 4

t + 5

t + 6t + 7

t + 2

Figure 6 The migration process of DMLA

(5) Communication component agent stores the infor-mation of the migration path (119868

0(119905 = 0) 119868

1(119905 =

1) 119868119896(119905 = 119896) 119883 (119905 = 0) 119883 (119905 = 1) 119883 (119905 =

119896)) and other relevant information

And it interacts with the main component agent

(6) The main component agent generates the migrationsequence 119873

1 1198732 119873

119896 according to location of

target entity

Send it to each subcomponent agent (this step isoptional for different applications)

(7) The main component agent sends complete instruc-tions to the lower-layered middleware software

Method functions which are predefined in the mid-dleware software recycle the communication compo-nent agent and the subcomponent agents

(8) The main component agent goes to the destinationnode based on the migration sequence

If the migration fails send failure information toCIS to terminate this migration otherwise generatenew communication component agent and subcom-ponent agents and this migration ends

6 Simulation and Analyses

61 The Effect of Single Mobile Agent in Target TrackingBefore the simulation experiments we firstly use targettracking as a sample to demonstrate the migration effect ofa single mobile agent of different migration algorithms Asshown in Figures 5 6 and 7 suppose that a car is runningfrom left to right through the area covered byWSNThe smallcircle represents the sensor node solid small circle representsthe cluster head which distributes agents large dotted circlerepresents the communication range of sensor node and eachnode selects the next hop only within its communicationrange The above-mentioned figures illustrate the specificlocations of the car and agent from time 119905 to time 119905 + 7

The migration process of SMLA is shown in Figure 5 Attime 119905 CIS distributes the mobile agent to the node whichis the nearest to the target entity according to the prede-termined migration sequence When the object is stationaryor its motion can be estimated this static approach is verysuitable However it loses efficacy in some conditions Forexample if the object remains in the location at time 119905+1 thenthe migration sequence will be the positions that are aroundthe target entity But once the target entity starts to move notaccording to the predetermination SMLA will fail

DMLA can in some sense optimize the performanceof SMLA As shown in Figure 6 mobile agent tracks themovement of the target entity However the movement of

International Journal of Distributed Sensor Networks 9

t

t + 1 t + 3

t + 4

t + 5

t + 6t + 7

t + 2

t t + 1 t + 3 t + 4 t + 5 t + 6 t + 7t + 2

Figure 7 The migration process of pid-DMLA

Table 1 Basic simulation parameters of single mobile agent migra-tion algorithms

Network area 500 times 500mNumber of nodes 200 or 800Nodersquos initial energy 50 JNodersquos sensing radius 5mEnergy consumption for data processing 03 JEnergy consumption for sending beacon frames 01 JThe interval of node broadcasting beacon frames 01 sThe speed of the target entity 0sim50msThe threshold of the amount of information 18The size of a mobile agent 800 bitsSimulation time 100 s

the target entity is unpredictable so it is not accurate to decidemigration path only by the amount of information which ishowever related to the distance to the target entity

The pid-DMLA can predict movement information oftarget entity dynamically which is a better solution to suchproblems As shown in Figure 7 at time 119905 + 5 mobile agentselects the next hop according to the predicted informationwhich is different from DMLA In brief pid-DMLA rebuildsmigration sequence based on predicting themovement of thetarget entity to achieve the purpose of tracking

62 Performance Analysis on the Single Mobile Agent Migra-tion Algorithms We use simulation software NS2 to eval-uate the three single mobile agent migration algorithmsrsquoperformances The default simulation parameters are shownin Table 1

Suppose that once the agent migrates one hop all sensornodes in the network consume 02 J For SMLA the transmis-sion distance is set to 30m while for DMLA and pid-DMLAthe transmission distance is set to be 10m Therefore theconsumption of transmission for SMLA is three times morethan that of sendingThe reason for this setting is that SMLAneeds bigger transmission range to collect information Thedirection and size of movement of the target entity areconstant denoted as V (ms) Once the target entity reachesthe edge of the sensing area it reverses direction immediatelywhich ensures its being always within the sensing area

0 5 10 15 20 25 30 35 40 45 500

2

4

6

8

10

12

Target speed (ms)

Ener

gy co

nsum

ptio

n (J

)

SMLADMLApid-DMLA

Figure 8 Energy consumption of the single agent migrationalgorithms in Scenario 1

Scenario 1 Set the number of nodes as 200 the target entityspeed as 0sim50ms and the simulation time as 100 s whichcan guarantee a complete migration The simulation resultsare shown in Figures 8 and 9 It is obvious that DMLA andpid-DMLA consume less energy than SMLA in the wholenetwork This is because the communication cost of SMLAis higher than that of DMLA But with the increasing speedof the target entity the effect of SMLA algorithm in trackingthe target entity becomes less and less obvious thereforemigration may terminate and the energy consumption onthe contrary becomes slower On the other hand when thespeed of target entity is very low there are no significantdifferences among the three migration schemes

Scenario 2 Set the speed of target entity as 5ms and thenumber of nodes as 800 Generally the coverage and thenumber of nodes are positively correlated The simulationresults are shown in Figures 10 and 11 It is clear that pid-DMLA has the lowest energy consumption when the numberof nodes is more than 100 This is because with the increaseof the nodes the choices of predicting the next hop become

10 International Journal of Distributed Sensor Networks

0102030405060708090

100

Life

time (

s)

0 5 10 15 20 25 30 35 40 45 50Target speed (ms)

SMLADMLApid-DMLA

Figure 9 Lifetime of the single agent migration algorithms inScenario 1

Ener

gy co

nsum

ptio

n (J

)

SMLADMLApid-DMLA

0 100 200 300 400 500 600 700 8000

1

2

3

4

5

6

7

8

9

Target speed (ms)

Figure 10 Energy consumption of the single agent migrationalgorithms in Scenario 2

more as a result the overheads become less [23] When thenumber of nodes is less than 50 the performance of SMLAis better than those of the other two This is because withless nodes static and dynamic migration algorithms share themore similar migration path however the migration tablehas been predetermined in SMLA thus it on the contraryhas the least overheads On the other hand when the densityof nodes is high there would be more neighbors and moremigration choices Thus DMLA and pid-DMLA outperformSMLA

63 Multiple Mobile Agents Migration Algorithm PerformanceAnalysis Before the assessment of Mpid-DMLA we need toconsider two problems [24 25] The first is the effect of code

SMLADMLApid-DMLA

10

20

30

40

50

60

70

80

Life

time (

s)

0 100 200 300 400 500 600 700 800Target speed (ms)

Figure 11 Lifetime of the single agent migration algorithms inScenario 2

segment applications on the amount of communication inthe migration process To ensure the comprehensiveness ofthe evaluation set the number of communication componentagents as 1 and the number of subcomponent agents as 5 10or 20 and evaluate Mpid-DMLA based on code segmentswith different scales The second is the effect of differentenvironment and running platform in practical operation onthe amount of information and the success rate of migrationTherefore we only consider the relation of the number ofcommunication and energy instead of the influence of theagentrsquos size

The performance of the migration of mobile agent inAgilla system has been evaluated in [19] First Agilla systemis not designed to support multiple mobile agents specificallywhich means that it is not aware of the number of the agentsin the code segment application And the systemrsquos overheadsare related to the size ofmobile agent So inAgilla all kinds ofagents are considered to be identical and there is no discrim-ination between the main and communication componentagents However this does not affect the simulation resultsbecause all agents are uniformly assumed in the simulationin this paper

The purpose of multiple mobile agents migration isto complete the entire application migration with limitedenergy The default simulation parameters are shown inTable 2

And when main component agent recycles and sendsmobile agents suppose that the energy consumption of thenode where main component agent stays is neglected whilethat of other nodes is 005 J

Figures 12 13 and 14 illustrate the relation between thenumber of application code segments and the success rateof one hop migration when the numbers of subcomponentagents are 5 10 and 15 respectivelyThe success rate ofMpid-DMLA is higher when the number of subcomponent agentsis 5 than those when the numbers of subcomponent agents

International Journal of Distributed Sensor Networks 11

Table 2 Basic simulation algorithm parameters of a single mobileagent migration

Network area 100 times 100mNumber of nodes 7 times 7Number of main component agents per codesegment application 1

Number of subcomponent agents per code segmentapplication 5 10 or 15

Number of communication component agents percode segment application 1

Nodersquos initial energy 2 JEnergy consumption of a node when an agentreaches it 005 J

Interaction consumption when an agent migrates 01 J

20

30

40

50

60

70

80

90

100

Mpid-DMLAAgilla

2 25 3 35 45 554 5 6Number of migration applications

Succ

ess r

ate (

)

Figure 12 The relation between the number of code segmentapplications and the success rate of migration (I)

are 10 and 20 It is obvious that the successful rate of Mpid-DMLA is generally not less than 50 which is much largerthan that of Agilla

Moreover Figure 15 demonstrates the relationship be-tween themigration hops and the amount of communicationwhen the subcomponent agents are 5 As shown in Figure 15with the samemigration hops the amount of communicationof Mpid-DMLA is much less than that of Agilla As the num-ber of data packet reflects the amount of energy consumptionMpid-DMLA consumes less energy than Agilla

Although the simulation results show that Mpid-DMLAis a suitable migration scheme for multiple mobile agents itbases on some specific assumptions In general this schemestill encounters the following three issues

(1) When the main component agent loses efficacy inthe process of migration code segment applicationscannot be rebuilt

(2) When subcomponent agents lose efficacy afterdeployment code segment applications will keepwaiting because the applications cannot be recycled

20

30

40

50

60

70

80

90

100

Mpid-DMLAAgilla

2 25 3 35 45 554 5 6Number of migration applications

Succ

ess r

ate (

)

Figure 13 The relation between the number of code segmentapplications and the success rate of migration (II)

10

20

30

40

50

60

70

80

0

Mpid-DMLAAgilla

2 25 3 35 45 554 5 6Number of migration applications

Succ

ess r

ate (

)

Figure 14 The relation between the number of code segmentapplications and the success rate of migration (III)

1 15 2 25 3 35 420

40

60

80

100

120

140

160

180

200

Hops of main agent

Num

ber o

f dat

a pac

kets

Mpid-DMLAAgilla

Figure 15The relation between themigration hops and the amountof communication

12 International Journal of Distributed Sensor Networks

(3) If the main and the subcomponent agents which loseefficacy in issue 1 or 2 are left in the nodes withoutbeing recycled then othermobile agents cannot reachthese nodes

Therefore in designing middleware software it is nec-essary to add the feedback mechanism of main componentagent and it needs to periodically return ACK informationto the higher layer of the middleware software When themiddleware finds that the main component agent losesefficacy it needs to distribute a new one or a specializedindependent agent to recycle the useless agents in the wholenetwork

In Agilla the method of recycling the useless agentsis by adding the unified interface into all of the agentsRemoving all of the mobile agents will affect the function ofcode segment applications Therefore developing the moreappropriate middleware software is the future work whichis also related to the development of the better migrationstrategies [26]

7 Conclusion

As big data is the intensification of information architec-ture while distributed computing is the intensification ofinfrastructure and resource distributed computing is anappropriate method to deal with big data In this paperwe have firstly illustrated the mobile agent model in thedistributed computing And it led to the two considerablefactors of migration the selection of nodes and of thesequence of them Then we have proposed the assumptionsof the migration algorithms Moreover we have designedthe nodersquos sensor model and information collection modelFurthermore we have formulated the beacon frames ofmobile agent for communication Generally mobile agentmigration algorithms can be transferred as the target trackingissue Thus we have described a simple target localizationalgorithm applicable for migration On this basis we havegeneralized the whole migration process of the mobile agent

Considering the single mobile agent we have proposedthe static SMLA and the dynamic DMLA algorithms Afteranalyzing the drawbacks of SMLA and DMLA we thenproposed the pid-DMLA algorithm Section 4 has demon-strated the modeling procedure of the abovementioned threealgorithms Section 6 has emulated and evaluated their per-formances and studied their advantages and disadvantages indifferent scenarios

Considering the multiple mobile agents applied in prac-tical target tracking we have also put forward the Mpid-DMLA algorithm In the process of mobile agent componentconstructing code segment applications we defined threemobile agents with different functions to cooperate with eachother to accomplish the migration Section 5 illustrated thearchitecture modeling procedure and basic process ofMpid-DMLA Section 6 compared it with Agilla and evaluated itsadvantages and disadvantages

Acknowledgments

The paper is sponsored by the National Natural ScienceFoundation of China (61170065 61171053 61003039 6110019961003236 61202355 61201163 and 61373138) the Natu-ral Science Foundation of Jiangsu Province (BK2011755BK2011072 and BK2012436) the Scientific amp Technolog-ical Support Project of Jiangsu Province (BE2012183 andBE2012755) the Natural Science Key Fund for Collegesand Universities in Jiangsu Province (12KJA520002) theScientific Research amp Industry Promotion Project for HigherEducation Institutions (JHB2012-7) the Postdoctoral Foun-dation (2012M511753 1101011B 1202034C and 2013T60536)the Doctoral Fund of Ministry of Education of China(20113223110002 and 20123223120006) the Fund of NanjingUniversity of Posts and Telecommunications (NY212047)and the project funded by the Priority Academic Pro-gram Development of Jiangsu Higher Education Institutions(yx002001)

References

[1] Research Trends special issue on big data httpwwwresearch-trendscomwp-contentuploads201209Research Trends Is-sue30pdf 2012

[2] L Ismail and D Hagimont ldquoA performance evaluation of themobile agent paradigmrdquoACM SIGPLANNotices vol 34 no 10pp 306ndash313 1999

[3] AMainwaring J Polastre R Szewczyk D Culler and J Ander-son ldquoWireless sensor networks for habitat monitoringrdquo inProceedings of the 1st ACM International Workshop on WirelessSensor Networks and Applications pp 88ndash97 September 2002

[4] IVY-A sensor network infrastructure for the college of engi-neering httpwww-bsaceecsberkeleyeduprojectsivy 2008

[5] I F AkyildizW Su Y Sankarasubramaniam and E Cayirci ldquoAsurvey on sensor networksrdquo IEEE Communications Magazinevol 40 no 8 pp 102ndash105 2002

[6] Y-C Tseng S-P Kuo H-W Lee and C-F Huang ldquoLocationtracking in a wireless sensor network by mobile agents and itsdata fusion strategiesrdquoComputer Journal vol 47 no 4 pp 448ndash460 2004

[7] Y Xu and H Qi ldquoMobile agent migration modeling anddesign for target tracking in wireless sensor networksrdquo Ad HocNetworks vol 6 no 1 pp 1ndash16 2008

[8] Y Xu and H Qi ldquoDistributed computing paradigms for collab-orative signal and information processing in sensor networksrdquoJournal of Parallel and Distributed Computing vol 64 no 8 pp945ndash959 2004

[9] F Zhao J Shin and J Reich ldquoInformation-driven dynamicsensor collaborationrdquo IEEE Signal Processing Magazine vol 19no 2 pp 61ndash72 2002

[10] J Liu J Reich and F Zhao ldquoCollaborative in-network pro-cessing for target trackingrdquo Eurasip Journal on Applied SignalProcessing vol 2003 no 4 pp 378ndash391 2003

[11] J-W Kim J-S In K Hur J-W Kim and D-S Eom ldquoAnintelligent agent-based routing structure for mobile sinks inWSNsrdquo IEEE Transactions on Consumer Electronics vol 56 no4 pp 2310ndash2316 2010

[12] H Wang G Pottie K Yao and D Estrin ldquoEntropy-basedsensor selection heuristic for target localizationrdquo in Proceedings

International Journal of Distributed Sensor Networks 13

of the 3rd International Symposium on Information Processing inSensor Networks (IPSN rsquo04) pp 36ndash45 ACM Press April 2004

[13] T S RappaportWireless Communications Principles and Prac-tice IEEE Press 1996

[14] H Mousannif and I Khalil ldquoCooperative data management inwireless sensor networksrdquo International Journal of Computa-tional Intelligence Systems vol 5 no 3 pp 403ndash412 2012

[15] N Labraoui M Guerroui M Aliouat and T Zia ldquoDataaggregation security challenge in wireless sensor networks asurveyrdquo Ad-Hoc and Sensor Wireless Networks vol 12 no 3-4pp 295ndash324 2011

[16] M Chen L T Yang T Kwon L Zhou and M Jo ldquoItineraryplanning for energy-efficient agent communications in wirelesssensor networksrdquo IEEE Transactions on Vehicular Technologyvol 60 no 7 pp 3290ndash3299 2011

[17] Q Wu N S V Rao J Barhen et al ldquoOn computing mobileagent routes for data fusion in distributed sensor networksrdquoIEEE Transactions on Knowledge and Data Engineering vol 16no 6 pp 740ndash753 2004

[18] C-L Fok G-C Roman and C Lu ldquoMobile agent middlewarefor sensor networks an application case studyrdquo in Proceedingsof the 4th International Symposium on Information Processing inSensor Networks (IPSN rsquo05) pp 382ndash387 April 2005

[19] A Boulis C Han andM B Srivastava ldquoDesign and implemen-tation of a framework for efficient and programmable sensornetworksrdquo in Proceedings of the 1st international conference onMobile systems applications and services ACM San FranciscoCalif USA 2003

[20] P Levis and D Culler ldquoMate a tiny virtual machine for sensornetworksrdquo ACM Sigplan Notices vol 37 no 10 pp 85ndash95 2002

[21] S Suenaga N Yoshioka and S Honiden ldquoGroup migration bymobile agents in wireless sensor networksrdquo Computer Journalvol 54 no 3 pp 345ndash355 2011

[22] P Bellavista Y Cai and T Magedanz ldquoRecent advances inmobile middleware for wireless systems and servicesrdquo MobileNetworks and Applications vol 16 no 3 pp 267ndash269 2011

[23] Z Zhirou and C Wengang ldquoModel and optimization ofmobile agentrsquos migration in gridrdquo in Proceedings of the 2ndInternational Conference on Bio-Inspired Computing Theoriesand Applications (BICTA rsquo07) pp 127ndash130 September 2007

[24] J-P Zhang M Jun J Yang and L-L Cheng ldquoAn improvedmigration strategy of mobile agentrdquo in Proceedings of the4th International Conference on Fuzzy Systems and KnowledgeDiscovery (FSKD rsquo07) pp 577ndash581 August 2007

[25] Y Lee and K Kim ldquoOptimal migration path searching usingpath adjustment and reassignment for mobile agentrdquo in Pro-ceedings of the 4th International Conference on NetworkedComputing and Advanced InformationManagement (NCM rsquo08)pp 564ndash569 Gyeongju South Korea September 2008

[26] M Tentori M Rodrıguez and J Favela Vara ldquoAn agent-basedmiddleware for the design of activity-aware applicationsrdquo IEEEIntelligent Systems vol 26 no 3 pp 15ndash23 2011

International Journal of

AerospaceEngineeringHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

RoboticsJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Active and Passive Electronic Components

Control Scienceand Engineering

Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

International Journal of

RotatingMachinery

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporation httpwwwhindawicom

Journal ofEngineeringVolume 2014

Submit your manuscripts athttpwwwhindawicom

VLSI Design

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Shock and Vibration

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Civil EngineeringAdvances in

Acoustics and VibrationAdvances in

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Electrical and Computer Engineering

Journal of

Advances inOptoElectronics

Hindawi Publishing Corporation httpwwwhindawicom

Volume 2014

The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014

SensorsJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Modelling amp Simulation in EngineeringHindawi Publishing Corporation httpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Chemical EngineeringInternational Journal of Antennas and

Propagation

International Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Navigation and Observation

International Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

DistributedSensor Networks

International Journal of

International Journal of Distributed Sensor Networks 9

t

t + 1 t + 3

t + 4

t + 5

t + 6t + 7

t + 2

t t + 1 t + 3 t + 4 t + 5 t + 6 t + 7t + 2

Figure 7 The migration process of pid-DMLA

Table 1 Basic simulation parameters of single mobile agent migra-tion algorithms

Network area 500 times 500mNumber of nodes 200 or 800Nodersquos initial energy 50 JNodersquos sensing radius 5mEnergy consumption for data processing 03 JEnergy consumption for sending beacon frames 01 JThe interval of node broadcasting beacon frames 01 sThe speed of the target entity 0sim50msThe threshold of the amount of information 18The size of a mobile agent 800 bitsSimulation time 100 s

the target entity is unpredictable so it is not accurate to decidemigration path only by the amount of information which ishowever related to the distance to the target entity

The pid-DMLA can predict movement information oftarget entity dynamically which is a better solution to suchproblems As shown in Figure 7 at time 119905 + 5 mobile agentselects the next hop according to the predicted informationwhich is different from DMLA In brief pid-DMLA rebuildsmigration sequence based on predicting themovement of thetarget entity to achieve the purpose of tracking

62 Performance Analysis on the Single Mobile Agent Migra-tion Algorithms We use simulation software NS2 to eval-uate the three single mobile agent migration algorithmsrsquoperformances The default simulation parameters are shownin Table 1

Suppose that once the agent migrates one hop all sensornodes in the network consume 02 J For SMLA the transmis-sion distance is set to 30m while for DMLA and pid-DMLAthe transmission distance is set to be 10m Therefore theconsumption of transmission for SMLA is three times morethan that of sendingThe reason for this setting is that SMLAneeds bigger transmission range to collect information Thedirection and size of movement of the target entity areconstant denoted as V (ms) Once the target entity reachesthe edge of the sensing area it reverses direction immediatelywhich ensures its being always within the sensing area

0 5 10 15 20 25 30 35 40 45 500

2

4

6

8

10

12

Target speed (ms)

Ener

gy co

nsum

ptio

n (J

)

SMLADMLApid-DMLA

Figure 8 Energy consumption of the single agent migrationalgorithms in Scenario 1

Scenario 1 Set the number of nodes as 200 the target entityspeed as 0sim50ms and the simulation time as 100 s whichcan guarantee a complete migration The simulation resultsare shown in Figures 8 and 9 It is obvious that DMLA andpid-DMLA consume less energy than SMLA in the wholenetwork This is because the communication cost of SMLAis higher than that of DMLA But with the increasing speedof the target entity the effect of SMLA algorithm in trackingthe target entity becomes less and less obvious thereforemigration may terminate and the energy consumption onthe contrary becomes slower On the other hand when thespeed of target entity is very low there are no significantdifferences among the three migration schemes

Scenario 2 Set the speed of target entity as 5ms and thenumber of nodes as 800 Generally the coverage and thenumber of nodes are positively correlated The simulationresults are shown in Figures 10 and 11 It is clear that pid-DMLA has the lowest energy consumption when the numberof nodes is more than 100 This is because with the increaseof the nodes the choices of predicting the next hop become

10 International Journal of Distributed Sensor Networks

0102030405060708090

100

Life

time (

s)

0 5 10 15 20 25 30 35 40 45 50Target speed (ms)

SMLADMLApid-DMLA

Figure 9 Lifetime of the single agent migration algorithms inScenario 1

Ener

gy co

nsum

ptio

n (J

)

SMLADMLApid-DMLA

0 100 200 300 400 500 600 700 8000

1

2

3

4

5

6

7

8

9

Target speed (ms)

Figure 10 Energy consumption of the single agent migrationalgorithms in Scenario 2

more as a result the overheads become less [23] When thenumber of nodes is less than 50 the performance of SMLAis better than those of the other two This is because withless nodes static and dynamic migration algorithms share themore similar migration path however the migration tablehas been predetermined in SMLA thus it on the contraryhas the least overheads On the other hand when the densityof nodes is high there would be more neighbors and moremigration choices Thus DMLA and pid-DMLA outperformSMLA

63 Multiple Mobile Agents Migration Algorithm PerformanceAnalysis Before the assessment of Mpid-DMLA we need toconsider two problems [24 25] The first is the effect of code

SMLADMLApid-DMLA

10

20

30

40

50

60

70

80

Life

time (

s)

0 100 200 300 400 500 600 700 800Target speed (ms)

Figure 11 Lifetime of the single agent migration algorithms inScenario 2

segment applications on the amount of communication inthe migration process To ensure the comprehensiveness ofthe evaluation set the number of communication componentagents as 1 and the number of subcomponent agents as 5 10or 20 and evaluate Mpid-DMLA based on code segmentswith different scales The second is the effect of differentenvironment and running platform in practical operation onthe amount of information and the success rate of migrationTherefore we only consider the relation of the number ofcommunication and energy instead of the influence of theagentrsquos size

The performance of the migration of mobile agent inAgilla system has been evaluated in [19] First Agilla systemis not designed to support multiple mobile agents specificallywhich means that it is not aware of the number of the agentsin the code segment application And the systemrsquos overheadsare related to the size ofmobile agent So inAgilla all kinds ofagents are considered to be identical and there is no discrim-ination between the main and communication componentagents However this does not affect the simulation resultsbecause all agents are uniformly assumed in the simulationin this paper

The purpose of multiple mobile agents migration isto complete the entire application migration with limitedenergy The default simulation parameters are shown inTable 2

And when main component agent recycles and sendsmobile agents suppose that the energy consumption of thenode where main component agent stays is neglected whilethat of other nodes is 005 J

Figures 12 13 and 14 illustrate the relation between thenumber of application code segments and the success rateof one hop migration when the numbers of subcomponentagents are 5 10 and 15 respectivelyThe success rate ofMpid-DMLA is higher when the number of subcomponent agentsis 5 than those when the numbers of subcomponent agents

International Journal of Distributed Sensor Networks 11

Table 2 Basic simulation algorithm parameters of a single mobileagent migration

Network area 100 times 100mNumber of nodes 7 times 7Number of main component agents per codesegment application 1

Number of subcomponent agents per code segmentapplication 5 10 or 15

Number of communication component agents percode segment application 1

Nodersquos initial energy 2 JEnergy consumption of a node when an agentreaches it 005 J

Interaction consumption when an agent migrates 01 J

20

30

40

50

60

70

80

90

100

Mpid-DMLAAgilla

2 25 3 35 45 554 5 6Number of migration applications

Succ

ess r

ate (

)

Figure 12 The relation between the number of code segmentapplications and the success rate of migration (I)

are 10 and 20 It is obvious that the successful rate of Mpid-DMLA is generally not less than 50 which is much largerthan that of Agilla

Moreover Figure 15 demonstrates the relationship be-tween themigration hops and the amount of communicationwhen the subcomponent agents are 5 As shown in Figure 15with the samemigration hops the amount of communicationof Mpid-DMLA is much less than that of Agilla As the num-ber of data packet reflects the amount of energy consumptionMpid-DMLA consumes less energy than Agilla

Although the simulation results show that Mpid-DMLAis a suitable migration scheme for multiple mobile agents itbases on some specific assumptions In general this schemestill encounters the following three issues

(1) When the main component agent loses efficacy inthe process of migration code segment applicationscannot be rebuilt

(2) When subcomponent agents lose efficacy afterdeployment code segment applications will keepwaiting because the applications cannot be recycled

20

30

40

50

60

70

80

90

100

Mpid-DMLAAgilla

2 25 3 35 45 554 5 6Number of migration applications

Succ

ess r

ate (

)

Figure 13 The relation between the number of code segmentapplications and the success rate of migration (II)

10

20

30

40

50

60

70

80

0

Mpid-DMLAAgilla

2 25 3 35 45 554 5 6Number of migration applications

Succ

ess r

ate (

)

Figure 14 The relation between the number of code segmentapplications and the success rate of migration (III)

1 15 2 25 3 35 420

40

60

80

100

120

140

160

180

200

Hops of main agent

Num

ber o

f dat

a pac

kets

Mpid-DMLAAgilla

Figure 15The relation between themigration hops and the amountof communication

12 International Journal of Distributed Sensor Networks

(3) If the main and the subcomponent agents which loseefficacy in issue 1 or 2 are left in the nodes withoutbeing recycled then othermobile agents cannot reachthese nodes

Therefore in designing middleware software it is nec-essary to add the feedback mechanism of main componentagent and it needs to periodically return ACK informationto the higher layer of the middleware software When themiddleware finds that the main component agent losesefficacy it needs to distribute a new one or a specializedindependent agent to recycle the useless agents in the wholenetwork

In Agilla the method of recycling the useless agentsis by adding the unified interface into all of the agentsRemoving all of the mobile agents will affect the function ofcode segment applications Therefore developing the moreappropriate middleware software is the future work whichis also related to the development of the better migrationstrategies [26]

7 Conclusion

As big data is the intensification of information architec-ture while distributed computing is the intensification ofinfrastructure and resource distributed computing is anappropriate method to deal with big data In this paperwe have firstly illustrated the mobile agent model in thedistributed computing And it led to the two considerablefactors of migration the selection of nodes and of thesequence of them Then we have proposed the assumptionsof the migration algorithms Moreover we have designedthe nodersquos sensor model and information collection modelFurthermore we have formulated the beacon frames ofmobile agent for communication Generally mobile agentmigration algorithms can be transferred as the target trackingissue Thus we have described a simple target localizationalgorithm applicable for migration On this basis we havegeneralized the whole migration process of the mobile agent

Considering the single mobile agent we have proposedthe static SMLA and the dynamic DMLA algorithms Afteranalyzing the drawbacks of SMLA and DMLA we thenproposed the pid-DMLA algorithm Section 4 has demon-strated the modeling procedure of the abovementioned threealgorithms Section 6 has emulated and evaluated their per-formances and studied their advantages and disadvantages indifferent scenarios

Considering the multiple mobile agents applied in prac-tical target tracking we have also put forward the Mpid-DMLA algorithm In the process of mobile agent componentconstructing code segment applications we defined threemobile agents with different functions to cooperate with eachother to accomplish the migration Section 5 illustrated thearchitecture modeling procedure and basic process ofMpid-DMLA Section 6 compared it with Agilla and evaluated itsadvantages and disadvantages

Acknowledgments

The paper is sponsored by the National Natural ScienceFoundation of China (61170065 61171053 61003039 6110019961003236 61202355 61201163 and 61373138) the Natu-ral Science Foundation of Jiangsu Province (BK2011755BK2011072 and BK2012436) the Scientific amp Technolog-ical Support Project of Jiangsu Province (BE2012183 andBE2012755) the Natural Science Key Fund for Collegesand Universities in Jiangsu Province (12KJA520002) theScientific Research amp Industry Promotion Project for HigherEducation Institutions (JHB2012-7) the Postdoctoral Foun-dation (2012M511753 1101011B 1202034C and 2013T60536)the Doctoral Fund of Ministry of Education of China(20113223110002 and 20123223120006) the Fund of NanjingUniversity of Posts and Telecommunications (NY212047)and the project funded by the Priority Academic Pro-gram Development of Jiangsu Higher Education Institutions(yx002001)

References

[1] Research Trends special issue on big data httpwwwresearch-trendscomwp-contentuploads201209Research Trends Is-sue30pdf 2012

[2] L Ismail and D Hagimont ldquoA performance evaluation of themobile agent paradigmrdquoACM SIGPLANNotices vol 34 no 10pp 306ndash313 1999

[3] AMainwaring J Polastre R Szewczyk D Culler and J Ander-son ldquoWireless sensor networks for habitat monitoringrdquo inProceedings of the 1st ACM International Workshop on WirelessSensor Networks and Applications pp 88ndash97 September 2002

[4] IVY-A sensor network infrastructure for the college of engi-neering httpwww-bsaceecsberkeleyeduprojectsivy 2008

[5] I F AkyildizW Su Y Sankarasubramaniam and E Cayirci ldquoAsurvey on sensor networksrdquo IEEE Communications Magazinevol 40 no 8 pp 102ndash105 2002

[6] Y-C Tseng S-P Kuo H-W Lee and C-F Huang ldquoLocationtracking in a wireless sensor network by mobile agents and itsdata fusion strategiesrdquoComputer Journal vol 47 no 4 pp 448ndash460 2004

[7] Y Xu and H Qi ldquoMobile agent migration modeling anddesign for target tracking in wireless sensor networksrdquo Ad HocNetworks vol 6 no 1 pp 1ndash16 2008

[8] Y Xu and H Qi ldquoDistributed computing paradigms for collab-orative signal and information processing in sensor networksrdquoJournal of Parallel and Distributed Computing vol 64 no 8 pp945ndash959 2004

[9] F Zhao J Shin and J Reich ldquoInformation-driven dynamicsensor collaborationrdquo IEEE Signal Processing Magazine vol 19no 2 pp 61ndash72 2002

[10] J Liu J Reich and F Zhao ldquoCollaborative in-network pro-cessing for target trackingrdquo Eurasip Journal on Applied SignalProcessing vol 2003 no 4 pp 378ndash391 2003

[11] J-W Kim J-S In K Hur J-W Kim and D-S Eom ldquoAnintelligent agent-based routing structure for mobile sinks inWSNsrdquo IEEE Transactions on Consumer Electronics vol 56 no4 pp 2310ndash2316 2010

[12] H Wang G Pottie K Yao and D Estrin ldquoEntropy-basedsensor selection heuristic for target localizationrdquo in Proceedings

International Journal of Distributed Sensor Networks 13

of the 3rd International Symposium on Information Processing inSensor Networks (IPSN rsquo04) pp 36ndash45 ACM Press April 2004

[13] T S RappaportWireless Communications Principles and Prac-tice IEEE Press 1996

[14] H Mousannif and I Khalil ldquoCooperative data management inwireless sensor networksrdquo International Journal of Computa-tional Intelligence Systems vol 5 no 3 pp 403ndash412 2012

[15] N Labraoui M Guerroui M Aliouat and T Zia ldquoDataaggregation security challenge in wireless sensor networks asurveyrdquo Ad-Hoc and Sensor Wireless Networks vol 12 no 3-4pp 295ndash324 2011

[16] M Chen L T Yang T Kwon L Zhou and M Jo ldquoItineraryplanning for energy-efficient agent communications in wirelesssensor networksrdquo IEEE Transactions on Vehicular Technologyvol 60 no 7 pp 3290ndash3299 2011

[17] Q Wu N S V Rao J Barhen et al ldquoOn computing mobileagent routes for data fusion in distributed sensor networksrdquoIEEE Transactions on Knowledge and Data Engineering vol 16no 6 pp 740ndash753 2004

[18] C-L Fok G-C Roman and C Lu ldquoMobile agent middlewarefor sensor networks an application case studyrdquo in Proceedingsof the 4th International Symposium on Information Processing inSensor Networks (IPSN rsquo05) pp 382ndash387 April 2005

[19] A Boulis C Han andM B Srivastava ldquoDesign and implemen-tation of a framework for efficient and programmable sensornetworksrdquo in Proceedings of the 1st international conference onMobile systems applications and services ACM San FranciscoCalif USA 2003

[20] P Levis and D Culler ldquoMate a tiny virtual machine for sensornetworksrdquo ACM Sigplan Notices vol 37 no 10 pp 85ndash95 2002

[21] S Suenaga N Yoshioka and S Honiden ldquoGroup migration bymobile agents in wireless sensor networksrdquo Computer Journalvol 54 no 3 pp 345ndash355 2011

[22] P Bellavista Y Cai and T Magedanz ldquoRecent advances inmobile middleware for wireless systems and servicesrdquo MobileNetworks and Applications vol 16 no 3 pp 267ndash269 2011

[23] Z Zhirou and C Wengang ldquoModel and optimization ofmobile agentrsquos migration in gridrdquo in Proceedings of the 2ndInternational Conference on Bio-Inspired Computing Theoriesand Applications (BICTA rsquo07) pp 127ndash130 September 2007

[24] J-P Zhang M Jun J Yang and L-L Cheng ldquoAn improvedmigration strategy of mobile agentrdquo in Proceedings of the4th International Conference on Fuzzy Systems and KnowledgeDiscovery (FSKD rsquo07) pp 577ndash581 August 2007

[25] Y Lee and K Kim ldquoOptimal migration path searching usingpath adjustment and reassignment for mobile agentrdquo in Pro-ceedings of the 4th International Conference on NetworkedComputing and Advanced InformationManagement (NCM rsquo08)pp 564ndash569 Gyeongju South Korea September 2008

[26] M Tentori M Rodrıguez and J Favela Vara ldquoAn agent-basedmiddleware for the design of activity-aware applicationsrdquo IEEEIntelligent Systems vol 26 no 3 pp 15ndash23 2011

International Journal of

AerospaceEngineeringHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

RoboticsJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Active and Passive Electronic Components

Control Scienceand Engineering

Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

International Journal of

RotatingMachinery

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporation httpwwwhindawicom

Journal ofEngineeringVolume 2014

Submit your manuscripts athttpwwwhindawicom

VLSI Design

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Shock and Vibration

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Civil EngineeringAdvances in

Acoustics and VibrationAdvances in

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Electrical and Computer Engineering

Journal of

Advances inOptoElectronics

Hindawi Publishing Corporation httpwwwhindawicom

Volume 2014

The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014

SensorsJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Modelling amp Simulation in EngineeringHindawi Publishing Corporation httpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Chemical EngineeringInternational Journal of Antennas and

Propagation

International Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Navigation and Observation

International Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

DistributedSensor Networks

International Journal of

10 International Journal of Distributed Sensor Networks

0102030405060708090

100

Life

time (

s)

0 5 10 15 20 25 30 35 40 45 50Target speed (ms)

SMLADMLApid-DMLA

Figure 9 Lifetime of the single agent migration algorithms inScenario 1

Ener

gy co

nsum

ptio

n (J

)

SMLADMLApid-DMLA

0 100 200 300 400 500 600 700 8000

1

2

3

4

5

6

7

8

9

Target speed (ms)

Figure 10 Energy consumption of the single agent migrationalgorithms in Scenario 2

more as a result the overheads become less [23] When thenumber of nodes is less than 50 the performance of SMLAis better than those of the other two This is because withless nodes static and dynamic migration algorithms share themore similar migration path however the migration tablehas been predetermined in SMLA thus it on the contraryhas the least overheads On the other hand when the densityof nodes is high there would be more neighbors and moremigration choices Thus DMLA and pid-DMLA outperformSMLA

63 Multiple Mobile Agents Migration Algorithm PerformanceAnalysis Before the assessment of Mpid-DMLA we need toconsider two problems [24 25] The first is the effect of code

SMLADMLApid-DMLA

10

20

30

40

50

60

70

80

Life

time (

s)

0 100 200 300 400 500 600 700 800Target speed (ms)

Figure 11 Lifetime of the single agent migration algorithms inScenario 2

segment applications on the amount of communication inthe migration process To ensure the comprehensiveness ofthe evaluation set the number of communication componentagents as 1 and the number of subcomponent agents as 5 10or 20 and evaluate Mpid-DMLA based on code segmentswith different scales The second is the effect of differentenvironment and running platform in practical operation onthe amount of information and the success rate of migrationTherefore we only consider the relation of the number ofcommunication and energy instead of the influence of theagentrsquos size

The performance of the migration of mobile agent inAgilla system has been evaluated in [19] First Agilla systemis not designed to support multiple mobile agents specificallywhich means that it is not aware of the number of the agentsin the code segment application And the systemrsquos overheadsare related to the size ofmobile agent So inAgilla all kinds ofagents are considered to be identical and there is no discrim-ination between the main and communication componentagents However this does not affect the simulation resultsbecause all agents are uniformly assumed in the simulationin this paper

The purpose of multiple mobile agents migration isto complete the entire application migration with limitedenergy The default simulation parameters are shown inTable 2

And when main component agent recycles and sendsmobile agents suppose that the energy consumption of thenode where main component agent stays is neglected whilethat of other nodes is 005 J

Figures 12 13 and 14 illustrate the relation between thenumber of application code segments and the success rateof one hop migration when the numbers of subcomponentagents are 5 10 and 15 respectivelyThe success rate ofMpid-DMLA is higher when the number of subcomponent agentsis 5 than those when the numbers of subcomponent agents

International Journal of Distributed Sensor Networks 11

Table 2 Basic simulation algorithm parameters of a single mobileagent migration

Network area 100 times 100mNumber of nodes 7 times 7Number of main component agents per codesegment application 1

Number of subcomponent agents per code segmentapplication 5 10 or 15

Number of communication component agents percode segment application 1

Nodersquos initial energy 2 JEnergy consumption of a node when an agentreaches it 005 J

Interaction consumption when an agent migrates 01 J

20

30

40

50

60

70

80

90

100

Mpid-DMLAAgilla

2 25 3 35 45 554 5 6Number of migration applications

Succ

ess r

ate (

)

Figure 12 The relation between the number of code segmentapplications and the success rate of migration (I)

are 10 and 20 It is obvious that the successful rate of Mpid-DMLA is generally not less than 50 which is much largerthan that of Agilla

Moreover Figure 15 demonstrates the relationship be-tween themigration hops and the amount of communicationwhen the subcomponent agents are 5 As shown in Figure 15with the samemigration hops the amount of communicationof Mpid-DMLA is much less than that of Agilla As the num-ber of data packet reflects the amount of energy consumptionMpid-DMLA consumes less energy than Agilla

Although the simulation results show that Mpid-DMLAis a suitable migration scheme for multiple mobile agents itbases on some specific assumptions In general this schemestill encounters the following three issues

(1) When the main component agent loses efficacy inthe process of migration code segment applicationscannot be rebuilt

(2) When subcomponent agents lose efficacy afterdeployment code segment applications will keepwaiting because the applications cannot be recycled

20

30

40

50

60

70

80

90

100

Mpid-DMLAAgilla

2 25 3 35 45 554 5 6Number of migration applications

Succ

ess r

ate (

)

Figure 13 The relation between the number of code segmentapplications and the success rate of migration (II)

10

20

30

40

50

60

70

80

0

Mpid-DMLAAgilla

2 25 3 35 45 554 5 6Number of migration applications

Succ

ess r

ate (

)

Figure 14 The relation between the number of code segmentapplications and the success rate of migration (III)

1 15 2 25 3 35 420

40

60

80

100

120

140

160

180

200

Hops of main agent

Num

ber o

f dat

a pac

kets

Mpid-DMLAAgilla

Figure 15The relation between themigration hops and the amountof communication

12 International Journal of Distributed Sensor Networks

(3) If the main and the subcomponent agents which loseefficacy in issue 1 or 2 are left in the nodes withoutbeing recycled then othermobile agents cannot reachthese nodes

Therefore in designing middleware software it is nec-essary to add the feedback mechanism of main componentagent and it needs to periodically return ACK informationto the higher layer of the middleware software When themiddleware finds that the main component agent losesefficacy it needs to distribute a new one or a specializedindependent agent to recycle the useless agents in the wholenetwork

In Agilla the method of recycling the useless agentsis by adding the unified interface into all of the agentsRemoving all of the mobile agents will affect the function ofcode segment applications Therefore developing the moreappropriate middleware software is the future work whichis also related to the development of the better migrationstrategies [26]

7 Conclusion

As big data is the intensification of information architec-ture while distributed computing is the intensification ofinfrastructure and resource distributed computing is anappropriate method to deal with big data In this paperwe have firstly illustrated the mobile agent model in thedistributed computing And it led to the two considerablefactors of migration the selection of nodes and of thesequence of them Then we have proposed the assumptionsof the migration algorithms Moreover we have designedthe nodersquos sensor model and information collection modelFurthermore we have formulated the beacon frames ofmobile agent for communication Generally mobile agentmigration algorithms can be transferred as the target trackingissue Thus we have described a simple target localizationalgorithm applicable for migration On this basis we havegeneralized the whole migration process of the mobile agent

Considering the single mobile agent we have proposedthe static SMLA and the dynamic DMLA algorithms Afteranalyzing the drawbacks of SMLA and DMLA we thenproposed the pid-DMLA algorithm Section 4 has demon-strated the modeling procedure of the abovementioned threealgorithms Section 6 has emulated and evaluated their per-formances and studied their advantages and disadvantages indifferent scenarios

Considering the multiple mobile agents applied in prac-tical target tracking we have also put forward the Mpid-DMLA algorithm In the process of mobile agent componentconstructing code segment applications we defined threemobile agents with different functions to cooperate with eachother to accomplish the migration Section 5 illustrated thearchitecture modeling procedure and basic process ofMpid-DMLA Section 6 compared it with Agilla and evaluated itsadvantages and disadvantages

Acknowledgments

The paper is sponsored by the National Natural ScienceFoundation of China (61170065 61171053 61003039 6110019961003236 61202355 61201163 and 61373138) the Natu-ral Science Foundation of Jiangsu Province (BK2011755BK2011072 and BK2012436) the Scientific amp Technolog-ical Support Project of Jiangsu Province (BE2012183 andBE2012755) the Natural Science Key Fund for Collegesand Universities in Jiangsu Province (12KJA520002) theScientific Research amp Industry Promotion Project for HigherEducation Institutions (JHB2012-7) the Postdoctoral Foun-dation (2012M511753 1101011B 1202034C and 2013T60536)the Doctoral Fund of Ministry of Education of China(20113223110002 and 20123223120006) the Fund of NanjingUniversity of Posts and Telecommunications (NY212047)and the project funded by the Priority Academic Pro-gram Development of Jiangsu Higher Education Institutions(yx002001)

References

[1] Research Trends special issue on big data httpwwwresearch-trendscomwp-contentuploads201209Research Trends Is-sue30pdf 2012

[2] L Ismail and D Hagimont ldquoA performance evaluation of themobile agent paradigmrdquoACM SIGPLANNotices vol 34 no 10pp 306ndash313 1999

[3] AMainwaring J Polastre R Szewczyk D Culler and J Ander-son ldquoWireless sensor networks for habitat monitoringrdquo inProceedings of the 1st ACM International Workshop on WirelessSensor Networks and Applications pp 88ndash97 September 2002

[4] IVY-A sensor network infrastructure for the college of engi-neering httpwww-bsaceecsberkeleyeduprojectsivy 2008

[5] I F AkyildizW Su Y Sankarasubramaniam and E Cayirci ldquoAsurvey on sensor networksrdquo IEEE Communications Magazinevol 40 no 8 pp 102ndash105 2002

[6] Y-C Tseng S-P Kuo H-W Lee and C-F Huang ldquoLocationtracking in a wireless sensor network by mobile agents and itsdata fusion strategiesrdquoComputer Journal vol 47 no 4 pp 448ndash460 2004

[7] Y Xu and H Qi ldquoMobile agent migration modeling anddesign for target tracking in wireless sensor networksrdquo Ad HocNetworks vol 6 no 1 pp 1ndash16 2008

[8] Y Xu and H Qi ldquoDistributed computing paradigms for collab-orative signal and information processing in sensor networksrdquoJournal of Parallel and Distributed Computing vol 64 no 8 pp945ndash959 2004

[9] F Zhao J Shin and J Reich ldquoInformation-driven dynamicsensor collaborationrdquo IEEE Signal Processing Magazine vol 19no 2 pp 61ndash72 2002

[10] J Liu J Reich and F Zhao ldquoCollaborative in-network pro-cessing for target trackingrdquo Eurasip Journal on Applied SignalProcessing vol 2003 no 4 pp 378ndash391 2003

[11] J-W Kim J-S In K Hur J-W Kim and D-S Eom ldquoAnintelligent agent-based routing structure for mobile sinks inWSNsrdquo IEEE Transactions on Consumer Electronics vol 56 no4 pp 2310ndash2316 2010

[12] H Wang G Pottie K Yao and D Estrin ldquoEntropy-basedsensor selection heuristic for target localizationrdquo in Proceedings

International Journal of Distributed Sensor Networks 13

of the 3rd International Symposium on Information Processing inSensor Networks (IPSN rsquo04) pp 36ndash45 ACM Press April 2004

[13] T S RappaportWireless Communications Principles and Prac-tice IEEE Press 1996

[14] H Mousannif and I Khalil ldquoCooperative data management inwireless sensor networksrdquo International Journal of Computa-tional Intelligence Systems vol 5 no 3 pp 403ndash412 2012

[15] N Labraoui M Guerroui M Aliouat and T Zia ldquoDataaggregation security challenge in wireless sensor networks asurveyrdquo Ad-Hoc and Sensor Wireless Networks vol 12 no 3-4pp 295ndash324 2011

[16] M Chen L T Yang T Kwon L Zhou and M Jo ldquoItineraryplanning for energy-efficient agent communications in wirelesssensor networksrdquo IEEE Transactions on Vehicular Technologyvol 60 no 7 pp 3290ndash3299 2011

[17] Q Wu N S V Rao J Barhen et al ldquoOn computing mobileagent routes for data fusion in distributed sensor networksrdquoIEEE Transactions on Knowledge and Data Engineering vol 16no 6 pp 740ndash753 2004

[18] C-L Fok G-C Roman and C Lu ldquoMobile agent middlewarefor sensor networks an application case studyrdquo in Proceedingsof the 4th International Symposium on Information Processing inSensor Networks (IPSN rsquo05) pp 382ndash387 April 2005

[19] A Boulis C Han andM B Srivastava ldquoDesign and implemen-tation of a framework for efficient and programmable sensornetworksrdquo in Proceedings of the 1st international conference onMobile systems applications and services ACM San FranciscoCalif USA 2003

[20] P Levis and D Culler ldquoMate a tiny virtual machine for sensornetworksrdquo ACM Sigplan Notices vol 37 no 10 pp 85ndash95 2002

[21] S Suenaga N Yoshioka and S Honiden ldquoGroup migration bymobile agents in wireless sensor networksrdquo Computer Journalvol 54 no 3 pp 345ndash355 2011

[22] P Bellavista Y Cai and T Magedanz ldquoRecent advances inmobile middleware for wireless systems and servicesrdquo MobileNetworks and Applications vol 16 no 3 pp 267ndash269 2011

[23] Z Zhirou and C Wengang ldquoModel and optimization ofmobile agentrsquos migration in gridrdquo in Proceedings of the 2ndInternational Conference on Bio-Inspired Computing Theoriesand Applications (BICTA rsquo07) pp 127ndash130 September 2007

[24] J-P Zhang M Jun J Yang and L-L Cheng ldquoAn improvedmigration strategy of mobile agentrdquo in Proceedings of the4th International Conference on Fuzzy Systems and KnowledgeDiscovery (FSKD rsquo07) pp 577ndash581 August 2007

[25] Y Lee and K Kim ldquoOptimal migration path searching usingpath adjustment and reassignment for mobile agentrdquo in Pro-ceedings of the 4th International Conference on NetworkedComputing and Advanced InformationManagement (NCM rsquo08)pp 564ndash569 Gyeongju South Korea September 2008

[26] M Tentori M Rodrıguez and J Favela Vara ldquoAn agent-basedmiddleware for the design of activity-aware applicationsrdquo IEEEIntelligent Systems vol 26 no 3 pp 15ndash23 2011

International Journal of

AerospaceEngineeringHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

RoboticsJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Active and Passive Electronic Components

Control Scienceand Engineering

Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

International Journal of

RotatingMachinery

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporation httpwwwhindawicom

Journal ofEngineeringVolume 2014

Submit your manuscripts athttpwwwhindawicom

VLSI Design

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Shock and Vibration

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Civil EngineeringAdvances in

Acoustics and VibrationAdvances in

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Electrical and Computer Engineering

Journal of

Advances inOptoElectronics

Hindawi Publishing Corporation httpwwwhindawicom

Volume 2014

The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014

SensorsJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Modelling amp Simulation in EngineeringHindawi Publishing Corporation httpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Chemical EngineeringInternational Journal of Antennas and

Propagation

International Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Navigation and Observation

International Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

DistributedSensor Networks

International Journal of

International Journal of Distributed Sensor Networks 11

Table 2 Basic simulation algorithm parameters of a single mobileagent migration

Network area 100 times 100mNumber of nodes 7 times 7Number of main component agents per codesegment application 1

Number of subcomponent agents per code segmentapplication 5 10 or 15

Number of communication component agents percode segment application 1

Nodersquos initial energy 2 JEnergy consumption of a node when an agentreaches it 005 J

Interaction consumption when an agent migrates 01 J

20

30

40

50

60

70

80

90

100

Mpid-DMLAAgilla

2 25 3 35 45 554 5 6Number of migration applications

Succ

ess r

ate (

)

Figure 12 The relation between the number of code segmentapplications and the success rate of migration (I)

are 10 and 20 It is obvious that the successful rate of Mpid-DMLA is generally not less than 50 which is much largerthan that of Agilla

Moreover Figure 15 demonstrates the relationship be-tween themigration hops and the amount of communicationwhen the subcomponent agents are 5 As shown in Figure 15with the samemigration hops the amount of communicationof Mpid-DMLA is much less than that of Agilla As the num-ber of data packet reflects the amount of energy consumptionMpid-DMLA consumes less energy than Agilla

Although the simulation results show that Mpid-DMLAis a suitable migration scheme for multiple mobile agents itbases on some specific assumptions In general this schemestill encounters the following three issues

(1) When the main component agent loses efficacy inthe process of migration code segment applicationscannot be rebuilt

(2) When subcomponent agents lose efficacy afterdeployment code segment applications will keepwaiting because the applications cannot be recycled

20

30

40

50

60

70

80

90

100

Mpid-DMLAAgilla

2 25 3 35 45 554 5 6Number of migration applications

Succ

ess r

ate (

)

Figure 13 The relation between the number of code segmentapplications and the success rate of migration (II)

10

20

30

40

50

60

70

80

0

Mpid-DMLAAgilla

2 25 3 35 45 554 5 6Number of migration applications

Succ

ess r

ate (

)

Figure 14 The relation between the number of code segmentapplications and the success rate of migration (III)

1 15 2 25 3 35 420

40

60

80

100

120

140

160

180

200

Hops of main agent

Num

ber o

f dat

a pac

kets

Mpid-DMLAAgilla

Figure 15The relation between themigration hops and the amountof communication

12 International Journal of Distributed Sensor Networks

(3) If the main and the subcomponent agents which loseefficacy in issue 1 or 2 are left in the nodes withoutbeing recycled then othermobile agents cannot reachthese nodes

Therefore in designing middleware software it is nec-essary to add the feedback mechanism of main componentagent and it needs to periodically return ACK informationto the higher layer of the middleware software When themiddleware finds that the main component agent losesefficacy it needs to distribute a new one or a specializedindependent agent to recycle the useless agents in the wholenetwork

In Agilla the method of recycling the useless agentsis by adding the unified interface into all of the agentsRemoving all of the mobile agents will affect the function ofcode segment applications Therefore developing the moreappropriate middleware software is the future work whichis also related to the development of the better migrationstrategies [26]

7 Conclusion

As big data is the intensification of information architec-ture while distributed computing is the intensification ofinfrastructure and resource distributed computing is anappropriate method to deal with big data In this paperwe have firstly illustrated the mobile agent model in thedistributed computing And it led to the two considerablefactors of migration the selection of nodes and of thesequence of them Then we have proposed the assumptionsof the migration algorithms Moreover we have designedthe nodersquos sensor model and information collection modelFurthermore we have formulated the beacon frames ofmobile agent for communication Generally mobile agentmigration algorithms can be transferred as the target trackingissue Thus we have described a simple target localizationalgorithm applicable for migration On this basis we havegeneralized the whole migration process of the mobile agent

Considering the single mobile agent we have proposedthe static SMLA and the dynamic DMLA algorithms Afteranalyzing the drawbacks of SMLA and DMLA we thenproposed the pid-DMLA algorithm Section 4 has demon-strated the modeling procedure of the abovementioned threealgorithms Section 6 has emulated and evaluated their per-formances and studied their advantages and disadvantages indifferent scenarios

Considering the multiple mobile agents applied in prac-tical target tracking we have also put forward the Mpid-DMLA algorithm In the process of mobile agent componentconstructing code segment applications we defined threemobile agents with different functions to cooperate with eachother to accomplish the migration Section 5 illustrated thearchitecture modeling procedure and basic process ofMpid-DMLA Section 6 compared it with Agilla and evaluated itsadvantages and disadvantages

Acknowledgments

The paper is sponsored by the National Natural ScienceFoundation of China (61170065 61171053 61003039 6110019961003236 61202355 61201163 and 61373138) the Natu-ral Science Foundation of Jiangsu Province (BK2011755BK2011072 and BK2012436) the Scientific amp Technolog-ical Support Project of Jiangsu Province (BE2012183 andBE2012755) the Natural Science Key Fund for Collegesand Universities in Jiangsu Province (12KJA520002) theScientific Research amp Industry Promotion Project for HigherEducation Institutions (JHB2012-7) the Postdoctoral Foun-dation (2012M511753 1101011B 1202034C and 2013T60536)the Doctoral Fund of Ministry of Education of China(20113223110002 and 20123223120006) the Fund of NanjingUniversity of Posts and Telecommunications (NY212047)and the project funded by the Priority Academic Pro-gram Development of Jiangsu Higher Education Institutions(yx002001)

References

[1] Research Trends special issue on big data httpwwwresearch-trendscomwp-contentuploads201209Research Trends Is-sue30pdf 2012

[2] L Ismail and D Hagimont ldquoA performance evaluation of themobile agent paradigmrdquoACM SIGPLANNotices vol 34 no 10pp 306ndash313 1999

[3] AMainwaring J Polastre R Szewczyk D Culler and J Ander-son ldquoWireless sensor networks for habitat monitoringrdquo inProceedings of the 1st ACM International Workshop on WirelessSensor Networks and Applications pp 88ndash97 September 2002

[4] IVY-A sensor network infrastructure for the college of engi-neering httpwww-bsaceecsberkeleyeduprojectsivy 2008

[5] I F AkyildizW Su Y Sankarasubramaniam and E Cayirci ldquoAsurvey on sensor networksrdquo IEEE Communications Magazinevol 40 no 8 pp 102ndash105 2002

[6] Y-C Tseng S-P Kuo H-W Lee and C-F Huang ldquoLocationtracking in a wireless sensor network by mobile agents and itsdata fusion strategiesrdquoComputer Journal vol 47 no 4 pp 448ndash460 2004

[7] Y Xu and H Qi ldquoMobile agent migration modeling anddesign for target tracking in wireless sensor networksrdquo Ad HocNetworks vol 6 no 1 pp 1ndash16 2008

[8] Y Xu and H Qi ldquoDistributed computing paradigms for collab-orative signal and information processing in sensor networksrdquoJournal of Parallel and Distributed Computing vol 64 no 8 pp945ndash959 2004

[9] F Zhao J Shin and J Reich ldquoInformation-driven dynamicsensor collaborationrdquo IEEE Signal Processing Magazine vol 19no 2 pp 61ndash72 2002

[10] J Liu J Reich and F Zhao ldquoCollaborative in-network pro-cessing for target trackingrdquo Eurasip Journal on Applied SignalProcessing vol 2003 no 4 pp 378ndash391 2003

[11] J-W Kim J-S In K Hur J-W Kim and D-S Eom ldquoAnintelligent agent-based routing structure for mobile sinks inWSNsrdquo IEEE Transactions on Consumer Electronics vol 56 no4 pp 2310ndash2316 2010

[12] H Wang G Pottie K Yao and D Estrin ldquoEntropy-basedsensor selection heuristic for target localizationrdquo in Proceedings

International Journal of Distributed Sensor Networks 13

of the 3rd International Symposium on Information Processing inSensor Networks (IPSN rsquo04) pp 36ndash45 ACM Press April 2004

[13] T S RappaportWireless Communications Principles and Prac-tice IEEE Press 1996

[14] H Mousannif and I Khalil ldquoCooperative data management inwireless sensor networksrdquo International Journal of Computa-tional Intelligence Systems vol 5 no 3 pp 403ndash412 2012

[15] N Labraoui M Guerroui M Aliouat and T Zia ldquoDataaggregation security challenge in wireless sensor networks asurveyrdquo Ad-Hoc and Sensor Wireless Networks vol 12 no 3-4pp 295ndash324 2011

[16] M Chen L T Yang T Kwon L Zhou and M Jo ldquoItineraryplanning for energy-efficient agent communications in wirelesssensor networksrdquo IEEE Transactions on Vehicular Technologyvol 60 no 7 pp 3290ndash3299 2011

[17] Q Wu N S V Rao J Barhen et al ldquoOn computing mobileagent routes for data fusion in distributed sensor networksrdquoIEEE Transactions on Knowledge and Data Engineering vol 16no 6 pp 740ndash753 2004

[18] C-L Fok G-C Roman and C Lu ldquoMobile agent middlewarefor sensor networks an application case studyrdquo in Proceedingsof the 4th International Symposium on Information Processing inSensor Networks (IPSN rsquo05) pp 382ndash387 April 2005

[19] A Boulis C Han andM B Srivastava ldquoDesign and implemen-tation of a framework for efficient and programmable sensornetworksrdquo in Proceedings of the 1st international conference onMobile systems applications and services ACM San FranciscoCalif USA 2003

[20] P Levis and D Culler ldquoMate a tiny virtual machine for sensornetworksrdquo ACM Sigplan Notices vol 37 no 10 pp 85ndash95 2002

[21] S Suenaga N Yoshioka and S Honiden ldquoGroup migration bymobile agents in wireless sensor networksrdquo Computer Journalvol 54 no 3 pp 345ndash355 2011

[22] P Bellavista Y Cai and T Magedanz ldquoRecent advances inmobile middleware for wireless systems and servicesrdquo MobileNetworks and Applications vol 16 no 3 pp 267ndash269 2011

[23] Z Zhirou and C Wengang ldquoModel and optimization ofmobile agentrsquos migration in gridrdquo in Proceedings of the 2ndInternational Conference on Bio-Inspired Computing Theoriesand Applications (BICTA rsquo07) pp 127ndash130 September 2007

[24] J-P Zhang M Jun J Yang and L-L Cheng ldquoAn improvedmigration strategy of mobile agentrdquo in Proceedings of the4th International Conference on Fuzzy Systems and KnowledgeDiscovery (FSKD rsquo07) pp 577ndash581 August 2007

[25] Y Lee and K Kim ldquoOptimal migration path searching usingpath adjustment and reassignment for mobile agentrdquo in Pro-ceedings of the 4th International Conference on NetworkedComputing and Advanced InformationManagement (NCM rsquo08)pp 564ndash569 Gyeongju South Korea September 2008

[26] M Tentori M Rodrıguez and J Favela Vara ldquoAn agent-basedmiddleware for the design of activity-aware applicationsrdquo IEEEIntelligent Systems vol 26 no 3 pp 15ndash23 2011

International Journal of

AerospaceEngineeringHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

RoboticsJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Active and Passive Electronic Components

Control Scienceand Engineering

Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

International Journal of

RotatingMachinery

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporation httpwwwhindawicom

Journal ofEngineeringVolume 2014

Submit your manuscripts athttpwwwhindawicom

VLSI Design

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Shock and Vibration

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Civil EngineeringAdvances in

Acoustics and VibrationAdvances in

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Electrical and Computer Engineering

Journal of

Advances inOptoElectronics

Hindawi Publishing Corporation httpwwwhindawicom

Volume 2014

The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014

SensorsJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Modelling amp Simulation in EngineeringHindawi Publishing Corporation httpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Chemical EngineeringInternational Journal of Antennas and

Propagation

International Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Navigation and Observation

International Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

DistributedSensor Networks

International Journal of

12 International Journal of Distributed Sensor Networks

(3) If the main and the subcomponent agents which loseefficacy in issue 1 or 2 are left in the nodes withoutbeing recycled then othermobile agents cannot reachthese nodes

Therefore in designing middleware software it is nec-essary to add the feedback mechanism of main componentagent and it needs to periodically return ACK informationto the higher layer of the middleware software When themiddleware finds that the main component agent losesefficacy it needs to distribute a new one or a specializedindependent agent to recycle the useless agents in the wholenetwork

In Agilla the method of recycling the useless agentsis by adding the unified interface into all of the agentsRemoving all of the mobile agents will affect the function ofcode segment applications Therefore developing the moreappropriate middleware software is the future work whichis also related to the development of the better migrationstrategies [26]

7 Conclusion

As big data is the intensification of information architec-ture while distributed computing is the intensification ofinfrastructure and resource distributed computing is anappropriate method to deal with big data In this paperwe have firstly illustrated the mobile agent model in thedistributed computing And it led to the two considerablefactors of migration the selection of nodes and of thesequence of them Then we have proposed the assumptionsof the migration algorithms Moreover we have designedthe nodersquos sensor model and information collection modelFurthermore we have formulated the beacon frames ofmobile agent for communication Generally mobile agentmigration algorithms can be transferred as the target trackingissue Thus we have described a simple target localizationalgorithm applicable for migration On this basis we havegeneralized the whole migration process of the mobile agent

Considering the single mobile agent we have proposedthe static SMLA and the dynamic DMLA algorithms Afteranalyzing the drawbacks of SMLA and DMLA we thenproposed the pid-DMLA algorithm Section 4 has demon-strated the modeling procedure of the abovementioned threealgorithms Section 6 has emulated and evaluated their per-formances and studied their advantages and disadvantages indifferent scenarios

Considering the multiple mobile agents applied in prac-tical target tracking we have also put forward the Mpid-DMLA algorithm In the process of mobile agent componentconstructing code segment applications we defined threemobile agents with different functions to cooperate with eachother to accomplish the migration Section 5 illustrated thearchitecture modeling procedure and basic process ofMpid-DMLA Section 6 compared it with Agilla and evaluated itsadvantages and disadvantages

Acknowledgments

The paper is sponsored by the National Natural ScienceFoundation of China (61170065 61171053 61003039 6110019961003236 61202355 61201163 and 61373138) the Natu-ral Science Foundation of Jiangsu Province (BK2011755BK2011072 and BK2012436) the Scientific amp Technolog-ical Support Project of Jiangsu Province (BE2012183 andBE2012755) the Natural Science Key Fund for Collegesand Universities in Jiangsu Province (12KJA520002) theScientific Research amp Industry Promotion Project for HigherEducation Institutions (JHB2012-7) the Postdoctoral Foun-dation (2012M511753 1101011B 1202034C and 2013T60536)the Doctoral Fund of Ministry of Education of China(20113223110002 and 20123223120006) the Fund of NanjingUniversity of Posts and Telecommunications (NY212047)and the project funded by the Priority Academic Pro-gram Development of Jiangsu Higher Education Institutions(yx002001)

References

[1] Research Trends special issue on big data httpwwwresearch-trendscomwp-contentuploads201209Research Trends Is-sue30pdf 2012

[2] L Ismail and D Hagimont ldquoA performance evaluation of themobile agent paradigmrdquoACM SIGPLANNotices vol 34 no 10pp 306ndash313 1999

[3] AMainwaring J Polastre R Szewczyk D Culler and J Ander-son ldquoWireless sensor networks for habitat monitoringrdquo inProceedings of the 1st ACM International Workshop on WirelessSensor Networks and Applications pp 88ndash97 September 2002

[4] IVY-A sensor network infrastructure for the college of engi-neering httpwww-bsaceecsberkeleyeduprojectsivy 2008

[5] I F AkyildizW Su Y Sankarasubramaniam and E Cayirci ldquoAsurvey on sensor networksrdquo IEEE Communications Magazinevol 40 no 8 pp 102ndash105 2002

[6] Y-C Tseng S-P Kuo H-W Lee and C-F Huang ldquoLocationtracking in a wireless sensor network by mobile agents and itsdata fusion strategiesrdquoComputer Journal vol 47 no 4 pp 448ndash460 2004

[7] Y Xu and H Qi ldquoMobile agent migration modeling anddesign for target tracking in wireless sensor networksrdquo Ad HocNetworks vol 6 no 1 pp 1ndash16 2008

[8] Y Xu and H Qi ldquoDistributed computing paradigms for collab-orative signal and information processing in sensor networksrdquoJournal of Parallel and Distributed Computing vol 64 no 8 pp945ndash959 2004

[9] F Zhao J Shin and J Reich ldquoInformation-driven dynamicsensor collaborationrdquo IEEE Signal Processing Magazine vol 19no 2 pp 61ndash72 2002

[10] J Liu J Reich and F Zhao ldquoCollaborative in-network pro-cessing for target trackingrdquo Eurasip Journal on Applied SignalProcessing vol 2003 no 4 pp 378ndash391 2003

[11] J-W Kim J-S In K Hur J-W Kim and D-S Eom ldquoAnintelligent agent-based routing structure for mobile sinks inWSNsrdquo IEEE Transactions on Consumer Electronics vol 56 no4 pp 2310ndash2316 2010

[12] H Wang G Pottie K Yao and D Estrin ldquoEntropy-basedsensor selection heuristic for target localizationrdquo in Proceedings

International Journal of Distributed Sensor Networks 13

of the 3rd International Symposium on Information Processing inSensor Networks (IPSN rsquo04) pp 36ndash45 ACM Press April 2004

[13] T S RappaportWireless Communications Principles and Prac-tice IEEE Press 1996

[14] H Mousannif and I Khalil ldquoCooperative data management inwireless sensor networksrdquo International Journal of Computa-tional Intelligence Systems vol 5 no 3 pp 403ndash412 2012

[15] N Labraoui M Guerroui M Aliouat and T Zia ldquoDataaggregation security challenge in wireless sensor networks asurveyrdquo Ad-Hoc and Sensor Wireless Networks vol 12 no 3-4pp 295ndash324 2011

[16] M Chen L T Yang T Kwon L Zhou and M Jo ldquoItineraryplanning for energy-efficient agent communications in wirelesssensor networksrdquo IEEE Transactions on Vehicular Technologyvol 60 no 7 pp 3290ndash3299 2011

[17] Q Wu N S V Rao J Barhen et al ldquoOn computing mobileagent routes for data fusion in distributed sensor networksrdquoIEEE Transactions on Knowledge and Data Engineering vol 16no 6 pp 740ndash753 2004

[18] C-L Fok G-C Roman and C Lu ldquoMobile agent middlewarefor sensor networks an application case studyrdquo in Proceedingsof the 4th International Symposium on Information Processing inSensor Networks (IPSN rsquo05) pp 382ndash387 April 2005

[19] A Boulis C Han andM B Srivastava ldquoDesign and implemen-tation of a framework for efficient and programmable sensornetworksrdquo in Proceedings of the 1st international conference onMobile systems applications and services ACM San FranciscoCalif USA 2003

[20] P Levis and D Culler ldquoMate a tiny virtual machine for sensornetworksrdquo ACM Sigplan Notices vol 37 no 10 pp 85ndash95 2002

[21] S Suenaga N Yoshioka and S Honiden ldquoGroup migration bymobile agents in wireless sensor networksrdquo Computer Journalvol 54 no 3 pp 345ndash355 2011

[22] P Bellavista Y Cai and T Magedanz ldquoRecent advances inmobile middleware for wireless systems and servicesrdquo MobileNetworks and Applications vol 16 no 3 pp 267ndash269 2011

[23] Z Zhirou and C Wengang ldquoModel and optimization ofmobile agentrsquos migration in gridrdquo in Proceedings of the 2ndInternational Conference on Bio-Inspired Computing Theoriesand Applications (BICTA rsquo07) pp 127ndash130 September 2007

[24] J-P Zhang M Jun J Yang and L-L Cheng ldquoAn improvedmigration strategy of mobile agentrdquo in Proceedings of the4th International Conference on Fuzzy Systems and KnowledgeDiscovery (FSKD rsquo07) pp 577ndash581 August 2007

[25] Y Lee and K Kim ldquoOptimal migration path searching usingpath adjustment and reassignment for mobile agentrdquo in Pro-ceedings of the 4th International Conference on NetworkedComputing and Advanced InformationManagement (NCM rsquo08)pp 564ndash569 Gyeongju South Korea September 2008

[26] M Tentori M Rodrıguez and J Favela Vara ldquoAn agent-basedmiddleware for the design of activity-aware applicationsrdquo IEEEIntelligent Systems vol 26 no 3 pp 15ndash23 2011

International Journal of

AerospaceEngineeringHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

RoboticsJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Active and Passive Electronic Components

Control Scienceand Engineering

Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

International Journal of

RotatingMachinery

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporation httpwwwhindawicom

Journal ofEngineeringVolume 2014

Submit your manuscripts athttpwwwhindawicom

VLSI Design

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Shock and Vibration

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Civil EngineeringAdvances in

Acoustics and VibrationAdvances in

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Electrical and Computer Engineering

Journal of

Advances inOptoElectronics

Hindawi Publishing Corporation httpwwwhindawicom

Volume 2014

The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014

SensorsJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Modelling amp Simulation in EngineeringHindawi Publishing Corporation httpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Chemical EngineeringInternational Journal of Antennas and

Propagation

International Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Navigation and Observation

International Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

DistributedSensor Networks

International Journal of

International Journal of Distributed Sensor Networks 13

of the 3rd International Symposium on Information Processing inSensor Networks (IPSN rsquo04) pp 36ndash45 ACM Press April 2004

[13] T S RappaportWireless Communications Principles and Prac-tice IEEE Press 1996

[14] H Mousannif and I Khalil ldquoCooperative data management inwireless sensor networksrdquo International Journal of Computa-tional Intelligence Systems vol 5 no 3 pp 403ndash412 2012

[15] N Labraoui M Guerroui M Aliouat and T Zia ldquoDataaggregation security challenge in wireless sensor networks asurveyrdquo Ad-Hoc and Sensor Wireless Networks vol 12 no 3-4pp 295ndash324 2011

[16] M Chen L T Yang T Kwon L Zhou and M Jo ldquoItineraryplanning for energy-efficient agent communications in wirelesssensor networksrdquo IEEE Transactions on Vehicular Technologyvol 60 no 7 pp 3290ndash3299 2011

[17] Q Wu N S V Rao J Barhen et al ldquoOn computing mobileagent routes for data fusion in distributed sensor networksrdquoIEEE Transactions on Knowledge and Data Engineering vol 16no 6 pp 740ndash753 2004

[18] C-L Fok G-C Roman and C Lu ldquoMobile agent middlewarefor sensor networks an application case studyrdquo in Proceedingsof the 4th International Symposium on Information Processing inSensor Networks (IPSN rsquo05) pp 382ndash387 April 2005

[19] A Boulis C Han andM B Srivastava ldquoDesign and implemen-tation of a framework for efficient and programmable sensornetworksrdquo in Proceedings of the 1st international conference onMobile systems applications and services ACM San FranciscoCalif USA 2003

[20] P Levis and D Culler ldquoMate a tiny virtual machine for sensornetworksrdquo ACM Sigplan Notices vol 37 no 10 pp 85ndash95 2002

[21] S Suenaga N Yoshioka and S Honiden ldquoGroup migration bymobile agents in wireless sensor networksrdquo Computer Journalvol 54 no 3 pp 345ndash355 2011

[22] P Bellavista Y Cai and T Magedanz ldquoRecent advances inmobile middleware for wireless systems and servicesrdquo MobileNetworks and Applications vol 16 no 3 pp 267ndash269 2011

[23] Z Zhirou and C Wengang ldquoModel and optimization ofmobile agentrsquos migration in gridrdquo in Proceedings of the 2ndInternational Conference on Bio-Inspired Computing Theoriesand Applications (BICTA rsquo07) pp 127ndash130 September 2007

[24] J-P Zhang M Jun J Yang and L-L Cheng ldquoAn improvedmigration strategy of mobile agentrdquo in Proceedings of the4th International Conference on Fuzzy Systems and KnowledgeDiscovery (FSKD rsquo07) pp 577ndash581 August 2007

[25] Y Lee and K Kim ldquoOptimal migration path searching usingpath adjustment and reassignment for mobile agentrdquo in Pro-ceedings of the 4th International Conference on NetworkedComputing and Advanced InformationManagement (NCM rsquo08)pp 564ndash569 Gyeongju South Korea September 2008

[26] M Tentori M Rodrıguez and J Favela Vara ldquoAn agent-basedmiddleware for the design of activity-aware applicationsrdquo IEEEIntelligent Systems vol 26 no 3 pp 15ndash23 2011

International Journal of

AerospaceEngineeringHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

RoboticsJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Active and Passive Electronic Components

Control Scienceand Engineering

Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

International Journal of

RotatingMachinery

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporation httpwwwhindawicom

Journal ofEngineeringVolume 2014

Submit your manuscripts athttpwwwhindawicom

VLSI Design

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Shock and Vibration

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Civil EngineeringAdvances in

Acoustics and VibrationAdvances in

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Electrical and Computer Engineering

Journal of

Advances inOptoElectronics

Hindawi Publishing Corporation httpwwwhindawicom

Volume 2014

The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014

SensorsJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Modelling amp Simulation in EngineeringHindawi Publishing Corporation httpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Chemical EngineeringInternational Journal of Antennas and

Propagation

International Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Navigation and Observation

International Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

DistributedSensor Networks

International Journal of

International Journal of

AerospaceEngineeringHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

RoboticsJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Active and Passive Electronic Components

Control Scienceand Engineering

Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

International Journal of

RotatingMachinery

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporation httpwwwhindawicom

Journal ofEngineeringVolume 2014

Submit your manuscripts athttpwwwhindawicom

VLSI Design

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Shock and Vibration

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Civil EngineeringAdvances in

Acoustics and VibrationAdvances in

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Electrical and Computer Engineering

Journal of

Advances inOptoElectronics

Hindawi Publishing Corporation httpwwwhindawicom

Volume 2014

The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014

SensorsJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Modelling amp Simulation in EngineeringHindawi Publishing Corporation httpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Chemical EngineeringInternational Journal of Antennas and

Propagation

International Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Navigation and Observation

International Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

DistributedSensor Networks

International Journal of