Research Article An Energy-Efficient and Relay Hop...
Transcript of Research Article An Energy-Efficient and Relay Hop...
Research ArticleAn Energy-Efficient and Relay Hop Bounded Mobile DataGathering Algorithm in Wireless Sensor Networks
Ling Chen1 Jianxin Wang12 Xiaoqing Peng1 and Xiaoyan Kui1
1School of Information Science and Engineering Central South University Changsha 410083 China2Hunan Engineering Center for Currency Recognition and Self-Service Changsha 410083 China
Correspondence should be addressed to Jianxin Wang jxwangmailcsueducn
Received 17 July 2014 Accepted 12 September 2014
Academic Editor Shigeng Zhang
Copyright copy 2015 Ling Chen et al This is an open access article distributed under the Creative Commons Attribution Licensewhich permits unrestricted use distribution and reproduction in any medium provided the original work is properly cited
Recent studies reveal that great benefit can be achieved by employing mobile collectors to gather data in wireless sensor networksSince the mobile collector can traverse the transmission range of each sensor the energy of nodes may be saved near maximallyHowever for directly receiving data packet from every node the length of mobile collector route should be very long Hence it maysignificantly increase the data gathering latency To solve this problem several algorithms have been proposed One of them calledBRH-MDG found that data gathering latency can be effectively shortened by performing proper local aggregation via multihoptransmissions and then uploading the aggregated data to the mobile collector But the BRH-MDG algorithm did not carefullyanalyze and optimize the energy consumption of the entire network In this paper we propose a mathematical model for theenergy consumption of the LNs and present a new algorithm called EEBRHMThe simulation results show that under the premiseof bounded relay hop compared with BRH-MDG EEBRHM can prolong the networks lifetime by 730
1 Introduction
Recent years with better foreground of practical applicationresearch work on wireless sensor networks (WSNs) hasattractedmore andmore attention In general aWSN ismadeup of low-cost and low-energy sensor nodes that can commu-nicate with each other by wireless links Since sensor nodesare often deployed in remote or inaccessible environmentsrecharging sensorsrsquo battery is usually impossible If the batteryof one sensor node runs out the node will die Thus howto save sensor energy and maximize network lifetime is amain challenge of WSNs Existing research works indicatethat the communication cost is themain part of sensor nodesrsquoenergy consumption And most of communication cost isproduced in the process of gathering sensing data from sensornodesThus how to design an energy-efficient data gatheringprotocol has become a hot research issue [1]
Some data gather protocols which focus on sensor nodesthemselves [2ndash13] which can be divided into three categories(1) cluster-based protocols [2ndash5] (2) chain-based protocols[6ndash8] (3) tree-based protocols [9ndash13] In these protocols
sensing data will be sent to static data sink through oneand more relay nodes However the multihop transmissionwill consume plenty of energy and the nodes near arounddata sink will consume much more energy than others Asa result these nodes become the bottlenecks of the networkOnce these bottleneck-nodes fail other sensors cannot sendsensing data to sink and that means the lifetime of thewhole network is over although other nodesrsquo batteries are stillenoughTherefore all these protocols may cause nonuniformenergy consumption across the WSN
Particularly in a large-scale data-centric WSN it will bedifficult to obtain an ideal lifetime of the whole networkif only use static data sink to gather data from all sensornodes [14 15] Thus recent research work focuses on mobiledata gathering which sets one or more mobile collectorswith powerful transceivers and batteries to traverse the wholesensing area In general to collect data from sensors in short-range communications amobile collectorwill stop for a shorttime at some anchor points on its moving path Because themobility of the collector can effectively decrease the relayhops of data packet energy consumption of each sensor node
Hindawi Publishing CorporationInternational Journal of Distributed Sensor NetworksVolume 2015 Article ID 680301 9 pageshttpdxdoiorg1011552015680301
2 International Journal of Distributed Sensor Networks
would be greatly reduced Furthermore to maximize energysaving the mobile collector could traverse the transmissionrange of each sensor node so that each sensing data packetcan be transmitted to the mobile collector in one hop Butbecause of low speed of the mobile collector and long lengthof moving path the latency of WSN may be very longin the process of data gathering which will not meet therequirement of delay in some time-sensitive applications [16]
In fact the latency of multihop data gathering is muchshorter than mobile data gathering Thus in the time-sensitive applications we have to make a trade-off betweenthe energy saving and the data gathering latency A generalway in some papers is to shorten the latency of mobiledata gathering by setting local data aggregation nodes (LNs)Specifically a part of sensor nodes will be selected as the LNswhich can aggregate the local sensing data from its affiliatedsensor nodes within a limited hop count These LNs couldtemporarily cache the sensing data and upload them to themobile collector when it arrives [16] Although the protocolsof setting LNs can effectively reduce the latency ofmobile datagathering they do not consider the energy consumption ofthe whole network which will have an adverse impact on thelifetime of network
The main contributions of this paper can be summarizedas follows
(1) By modeling and analyzing the energy consumptionofwhole networkwe characterize the energy-efficientbounded relay hop mobile data gathering as anoptimization problem
(2) We propose an efficient algorithm which is energy-efficient and can achieve a longer lifetime of networkthan these existing mobile data gathering protocolswith LNs
(3) The performance of the proposed algorithm is eval-uated by comparing with other three existing mobiledata gathering schemes Simulation results illustratethat the proposed algorithms achieve distinctive per-formance
The rest of the paper is organized as follows Section 2reviews the related work on mobile data gathering Section 3models and analyzes the energy consumption of wholenetwork and then formulates the problem Section 4 givesdetailed description on our algorithm and analyzes the per-formance of it The performance evaluation is demonstratedin Section 5 Finally Section 6 concludes the paper
2 Related Work
In this section let us briefly review some recent work onmobile data gathering in WSNs The mobile data gatheringprotocols can be divided into two categories according to themobility pattern
In the first category the mobility is uncontrollable andthe moving path of the mobile collector is either randomor determinate For example [17ndash19] In [17] Shah et alproposed a scheme in which a special type of mobile nodesis used as forwarding agents to facilitate connectivity among
static sensors and transport data with random mobilityTo improve the work in [17] Jain et al [18] presented ananalytical model to understand the key performance metricsof the WSNs with mobile collector such as data transferlatency of data gathering and power consumptions Jea etal [19] set some fixed straight moving paths and the mobilecollector gathered sensing data from the nodes near thesepaths A common advantage of these protocols is that theycan get high stability and reliability and it is easy for networkmaintaining But they lack agility and cannot apply to highlydynamic environment
The mobility of the second category is controlled inwhich mobile collector can freely move to any positionof the WSN and its path can be constructed for specificpurpose such as [20ndash25] Furthermore in this categorythe approaches can be classified into three subclasses Inthe first subclass the mobile collector visits each sensornode or traverses the transmission range of each node andgathers sensing data in one hop [20 21] To ensure nodata loss Somasundara et al [20] studied the scheduling ofmobile collector Ma and Yang [21] proposed tour planningalgorithm for getting a short moving path and ensuringall data gathering to be completed in one hop and it canachieve perfect uniformity of energy consumption Althoughthese protocols can minimize the energy cost by completelyavoiding multihop relays they may make a long latencyof data gathering especially in a large-scale WSN In thesecond subclass mobile collector gathers sensing data fromthe sensor nodes nearmoving path bymultihop transmissionMa and Yang [22] proposed a moving trajectory planningalgorithm via finding some turning points which is adaptiveto the sensor nodes distribution and can effectively avertbarriers on the trajectory In this approach the sensornodes along each moving line segment forward packets tothe mobile collector by a multihop fashion For improvingrouting reliability Kusy et al [23] constructed an algorithmby introducing the mobility graph which can encode theknowledge of likely mobility patterns within the networkWe can extract the mobility graph from training data andpredict future relay nodes by applying it on the mobilecollector to ensure uninterrupted data streams Luo andHubaux [24] thought that sensing data should be gatheredvia multihop relays while the mobile collector moves alongthe perimeter of theWSN which is considered as the optimalpath for themobile collectorThese protocols could effectivelyreduce or limit the length of moving path to a certain levelHowever they do not set any limitation on the hop countthat will make network lifetime (or a certain level of energyefficiency) uncertain In the third subclass data transmissionpatterns and moving path planning are both considered Forexample Zhao et al [25] constructed a mobile data gatheringalgorithm by considering the full utilization of concurrentdata uploading and the minimization of path length In thealgorithm multiple sensor nodes can upload sensing datato the mobile collector in one hop at the same time whicheffectively reduces the time of data uploading Zhao and Yang[16] proposed an algorithm called BRH-MDG (boundedrelay hop mobile data gathering) in which the length ofmoving path is minimized and the local data aggregation is
International Journal of Distributed Sensor Networks 3
guaranteed in bounded-hop count In this paper our workfalls into the third subclass Via modeling and analyzing theenergy consumption of whole network we construct a newalgorithm called energy-efficient bounded relay hop mobiledata gathering (EEBRHM) which can achieve much longernetwork lifetime than others and only increases a little delaycompared with the BRH-MDG
3 Preliminaries
31 Network Model Assume that V1 V2 V
119899are sensor
nodes in a sensing field and V0is the sink These 119899 nodes are
randomly deployed in a119872 times119872 field and form a WSN ThisWSN can be represented by a connective undirected graph119866(119881 119864) where 119881 is the set of sensor nodes and 119864 is the setof edges If the distance between nodes V
119894and V
119895is shorter
than themaximum communication distance there is an edge(V119894 V119895) isin 119864 |119881| = 119899 + 1 is the number of sensor nodes and
|119864| = 119898 is the number of edgesThere are some characteristicsin the WSN
(1) The sensor nodes are stationary after deployment(2) In the process of data gathering for a relay node
both the received data and its sensing data will beaggregated into a certain length packet Then thepacket will be sent to the next node
(3) After deployment the energy of sensor nodes cannotbe recharged and the initial energy of each sensornodes is the same
(4) There has been a time synchronization scheme in theWSN And the time accuracy can be in milliseconds[26]
In general we can assume that the amount of energyrequired to transmittingreceiving 1 bit of data is a fixed valueLet 119864119905denote energy required by transmitting and 119864
119903denote
energy required by receiving At the same time we assumethat there are some excellent congestion control strategies inthe WSN which can avoid congestion and retransmission inthe process of data gathering
32 Definitions In the protocols of mobile data gatheringwith bounded-hop LNs such as BRH-MDG algorithm theWSN is divided into some subtrees whose roots are the LNsBecause mobile collector can be recharged the lifetime ofnetwork is determined by the life cycle of subtrees Accordingto paper [13] we give several definitions which will be used inthis paper
Definition 1 A round is defined as the process of gatheringthe sensing data from sensor nodes to the sink in one cycleby mobile collector which could be equal to the time spent atthe moving path by mobile collector
Definition 2 The lifetime of a node V119894is defined as the
number of rounds in the node survival period We can judgewhether one sensor node is survival by a formula 119864(V
119894) gt 119890
when the residual energy of one node is lower than it the
node is dead Specifically119864(V119894) represents the residual energy
of one node and 119890 is a threshold In this paper each sensornode can aggregate received data and itself sensing data intoa 119897 bits packet Formula (1) is used to count the lifetime of anode
119871node (V119894) = lfloor119864 (V119894)
119897 (119864119905minus 119864119903) + 119863 (V
119894) 119897119864119903
rfloor (1)
where119863(V119894) represents the degree of a node in the subtree
Definition 3 The lifetime of a subtree is the lifetime of the firstdead node in the subtree defined as the following formula
119871 tree (119879119873) = min 119871node (V119894) V119894isin 119879119873 (2)
where 119879119873
represents one subtree and V119894is a node in the
subtree
Definition 4 The lifetime of the whole network is the shortestlifetime of the subtrees defined as the following formula
119871119873(119866) = min
119879119873isin119879119875
119871 tree (119879119873) (3)
where119879119873represents one subtree and TP is the set of subtrees
33 Problem Statement As shown above the subtrees withthe shortest lifetime determine the lifetime of the whole net-work Thus if we want to optimize the energy consumptionof the whole network we should maximize the lifetime of allsubtrees
In Formula (1) 119897 119864119903 and 119864
119905are constant and 119864(V
119894) is
initialized as a constant Only 119863(V119894) is a variable so it is
the main optimization objective To extract the constant 119864119903
119864119905from the denominator and use Formula (1) to substitute
119871node(119881119894) in formula (2) formula (2) can be transformed intoan equivalent form
119871 tree (119879119873) = minlfloor119864 (V119894)
119863 (V119894) + 119888
rfloor V119894isin 119879119873 (4)
where 119888 = (119864119905minus 119864119903)119864119903 Based on formulas (3) and (4) the
maximum lifetime of the whole network can be calculated bythe following formula
max 119871119873(119866) = max min
119879119873isin119879119875
lfloor
119864 (V119894)
119863 (V119894) + 119888
rfloor V119894isin 119879119873 (5)
In formula (5) to maximize the lifetime of subtrees 119863(V119894) of
each node of subtrees should be minimized The problem ofhow to obtain a spanning tree withminimumnode degree onthe premise of hop bounded has been proved to be NP-hardin [13]
4 Design of EEBRHM
41 Algorithm Description Base on formula (5) our problemof maximizing the lifetime of subtrees can be transformedto minimize the degree of each node of subtrees To solve
4 International Journal of Distributed Sensor Networks
(a) An example of BRH-MDG algorithm (b) After balancing the degree of each node
Figure 1 An example to show the effect of balancing the degree of each node
this problem a general idea is to balance degrees of nodesin subtrees such as BRH-MDG algorithm As shown inFigure 1(a) the BRH-MDG algorithm divides the networkinto three subtrees Since the process of generating subtreesis random the distribution of nodesrsquo degrees is not uniformIn Figure 1(a) the dotted lines represent the moving paththe black nodes represent the nodes which have maximumdegrees in subtrees and the maximum degree among theseblack nodes is 5 By applying a method to balance the degreesof nodes in Figure 1(a) a new set of subtrees can be obtainedas shown in Figure 1(b) and the maximum degree is reducedto 2 According to Definitions 3 and 4 the lifetime of networkin Figure 1(b) is twice more than that in Figure 1(a)
The example demonstrates that the key point to optimizethe energy consumption of the whole WSN is to balance thedistribution of nodesrsquo degree in their own subtrees In thispaper we propose a new algorithm called EEBRHM whichuses the number of children of nodes in the initial tree ascontrol metric to adjust the degree of each node The detailof EEBRHM algorithm is shown in Algorithm 1 and Table 1
First of all EEBRHM algorithm will construct shortestpath trees (SPTs) which should cover all sensor nodes withminimum hops in the WSN (see Algorithm 1 line (1))Because a large WSN may not be fully connected it maybe constituted by several subnetworks and thus it needsmore than one SPT to cover all sensor nodes In EEBRHMalgorithm generally the root of SPT is the data sink For thenodes from other subnetworks which are far from the datasink we can set virtual roots in the center of the subnetworksThus the new algorithm can be applied to not only connectedWSNs but also disconnectedWSNs which is one of themainadvantages for the mobile data gathering Moreover beforeall sensor nodes being deployed each node will be assigneda unique integer ID which will be used to allocate separatedelay time slot for each sensor nodeThen EEBRHMwill usethe number of sensor nodesrsquo children as metric to balancethe degrees of sensor nodes (see Algorithm 1 lines (2)ndash(18))Specifically the energy of root is assumed to be infinite so itsdegree does not need to be balanced For a sensor node if itsparent node is not root and there exists a neighbor node withthe minimum number of children and at the same height asits parent in its own SPT (to effectively balance theminimum
number of childrenmust be less than original value at least 2)this neighbor node will be the new parent of the node Afterupdating the new parent the old parent and the new parentnode will renew the numbers of children and inform all theirown neighbor nodes
The next task of EEBRHM is to determine the LNs Sincethe LNs are more close to the data sink and the length ofmoving path119880 is shorter than other sensor nodes the sensornodes which is more close to the data sink will more likelyto be LNs To determine LNs firstly all sensor nodes areassigned to one of three separate large time slots accordingto the distance between sensor nodes and the data sink Theformula (119879
119899+IDlowast1199051015840) 119899 = 0 1 2 ensures that each nodewill be
assigned to a separate small time slot so that the sensor nodeswill not interfere with each other in the process of algorithmMoreover 119879
119899is measured in seconds and 1199051015840 is measured in
milliseconds A node which is more close to the data sinkwill be assigned to an earlier large time slot and would havepriority to be a LN At last TSP algorithm is used to constructan approximate shortest tour 119880 visiting V
0and all the LNs
42 Performance Analysis To evaluate the performance ofEEBRHM firstly we will analyze the time complexity ofEEBRHM Assume a WSN has 119870 disconnected subnetworksand 119873 is the total number of sensor nodes in the WSN (1 ⩽119870 ⩽ 119873) For a subnetwork 119896 (119896 = 1 2 119870) it would take119874((119873119896)
2) time to find a SPT [27] where 119873
119896represents the
number of sensor nodes in subnetwork 119896 In our algorithm itruns in119874(119873) time to balance the nodesrsquo degree and constructLNs And it runs in 119874(1198732) time to find tour 119880 at most byusing TSP algorithm [28] Thus the total time complexity ofEEBRHM is119874((119873
119896)
2) +119874(119873)+119874(119873
2) In the worst case the
time complexity of EEBRHM is 119874(1198732)Due to balancing degree of sensor nodes the length
of moving path 119880 of EEBRHM may be longer than BRH-MDG algorithm which will increase a bit latency of networkby EEBRHM But the performance of network lifetime ofEEBRHM is much better than BRH-MDG
EEBRHM is also one of mobile data gathering algorithmsand the data transmission between nodes and mobile sink isthe main part of communication overhead The mechanism
International Journal of Distributed Sensor Networks 5
Algorithm EEBRHMInput Network topology 119866(119881 119864) the relay hop bound 119889
and the static data sink V0
Output A set of LNs a set of sub-trees and the tour 119880visiting the LNs and the data sink
(1) Construct SPTs for 119866 that cover all the vertices in 119881(2) for each node V do (3) if119867(V) ⩾ 2 (4) cn = CN(Parent (V)) V119905 = V(5) for each node V1015840 in Nb(V)
(6) if119867(V1015840) minus 119867(V) == minus1 (7) if CN(V1015840) lt 119888119899 (8) cn = CN(V1015840) V119905 = V1015840(9)
(10)
(11)
(12) if cn + 2 lt CN(Parent (V)) (13) Parent(V) ⩽ vt(14) Broadcast ACKMessage(15) When received the ACKMessage
the related nodes will make response(16)
(17)
(18) (19) Set three large time slots 119879
0 1198791 1198792 1198790lt 1198791lt 1198792
and set a constant 1199051015840 1199051015840 ≪ 1198790
(20) for each node V do (21) Get the unique integer ID of V(22) if V in one hop of V
0
(23) Allocate a delay time slot (1198790+ ID lowast 1199051015840) to V
(24)
(25) if V is out of one hop of V0but it is in two hop
(26) Allocate a delay time slot (1198791+ ID lowast 1199051015840) to V
(27)
(28) if V is out of two hop of V0
(29) Allocate a delay time slot (1198792+ ID lowast 1199051015840) to V
(30)
(31) (32) In the delay time slot of each node V (33) if V is not belong to a LN (34) Let it be a LN and broadcast a probe packet
with 119889 hops of time to live(35)
(36) if other nodes received the probe packet (37) if it is not a LN (38) if it is the child of relay node (39) Let it be an affiliated node of the source
node of the probe packet(40)
(41)
(42)
(43) (44) Use the TSP algorithm to find an approximate shortest
tour 119880 visiting V0and all the LNs
Algorithm 1 EEBRHM algorithm
6 International Journal of Distributed Sensor Networks
Table 1 Notations used in EEBRHM algorithm
Functions119867(V) This function returns the height of V in its own SPTParent(V) This function returns the parent node of VCN(V) This function returns the number of children nodes of VNb(V) This function returns a set of neighbor nodes of V
Variablescn The variable is used to cache the number of children of node VV119905 The variable is used to cache the node V which is relevant to cn
0
1000
2000
3000
4000
5000
6000
7000
8000
100 200 300 400 500
Net
wor
k lif
etim
e (ro
unds
)
Node number
EEBRHMBRH-MDG
(a) Network lifetime comparison
500
600
700
800
900
1000
1100
1200
1300
100 200 300 400 500
Tour
leng
th (m
)
Node number
EEBRHMBRH-MDG
CMESHDG
(b) Length of tour comparison
Figure 2 Performance comparison with changing the density of sensor nodes
of constructing the path of mobile sink of EEBRHM isthe same as BRH-MDG Thus the communication cost ofEEBRHM is just the same as BRH-MDG
5 Simulation Results
In the simulation we assume that 119873 sensor nodes arerandomly distributed over an119872times119872 square area and the datasink is set in the center of the area Furthermore we compareour algorithm with BRH-MDG [16] SHDG [21] and CME[19] SHDG is one kind of one hop mobile data gatheringscheme This algorithm will choose some pausing locationsfor mobile collector from a set of candidate locations so thatmobile collector can gather sensor data fromnodes in a singlehop Although we can use TSP algorithm to construct anapproximate shortest tour 119880 for mobile collector to visit allpausing locations and the data sink the latency of networkof SHDG is too long CME generates some parallel straighttours in advance which across the WSN Then the mobilecollector will traverse along these tours while gathering datafrom the sensor nodes nearby the tours And the other sensornodes would send their own sensor data to the nodes nearby
tours by multihop relays The latency of network of CMEis acceptable but due to randomly choosing a relay nodethe network lifetime achieved by CME may be poor Thedescription of BRH-MDGhas beenmentioned in Section 31All simulations are performed 30 times and average values oftheir results are taken as final results
51 The Impact of Sensor Node Density To compare theperformance of different algorithms we change the densityof sensor nodes The simulation setting in the experiments isthat communication radius 119877 bounded-hop 119889 and the areaare set as 30m 2 and 200 lowast 200m2 respectively Figure 2(a)plots the performance of EEBRHM and BRH-MDG as afunction of density of sensor nodes in terms of networklifetime From Figure 2(a) we can find that when the numberof sensor nodes becomes larger the advantage of EEBRHMover BRH-MDG become more obvious The performance ofnetwork lifetime of EEBRHM is 47 times than BRH-MDGon average There are two reasons to explain this First thedegrees of sensor nodes in BRH-MDGare very unevenwhichleads to uneven node energy consumption and thereforemake the network lifetime short Our EEBRHM algorithm
International Journal of Distributed Sensor Networks 7
Net
wor
k lif
etim
e (ro
unds
)
0
2000
4000
6000
8000
10000
20 30 40 50 60Communication radius (m)
EEBRHM d = 2
BRH-MDG d = 2
EEBRHM d = 3
BRH-MDG d = 3
(a) Network lifetime comparison
0
300
600
900
1200
1500
20 30 40 50 60
Tour
leng
th (m
)
Communication radius (m)
EEBRHM d = 2
BRH-MDG d = 2
EEBRHM d = 3
BRH-MDG d = 3
(b) Length of tour comparison
Figure 3 Performance comparison with changing the bounded 119889
can balance the degree of each node thus its performance ofnetwork lifetime is much better than BRH-MDGThe secondreason is that since BRH-MDGdoes not balance the degree ofeach node when the density of sensor nodes becomes largerthe degree of nodes in the WSN may increase very quicklytherefore the network lifetime of BRH-MDG would becomeshorter and shorter
Figure 2(b) shows the performance of four algorithms asa function of density of sensor nodes in terms of tour lengthThe simulation setting is the same as above For SHDG BRH-MDG and EEBRHM TSP algorithm is used to generate anapproximate shortest tour and mobile collector will movealong it For CME the parallel straight tours traversing thesensor field are 100m apart from each other One of thesetours must go through the center of the fieldThemobile col-lector can change onto other tours bymoving along the sensorfield border Both SHDG and CME algorithms are imple-mented in a centralized fashion BRH-MDG and EEBRHMare implemented in a distributed fashion In Figure 2(b) itis easy to observe that the tour length of BRH-MDG andEEBRHMgradually increases at first and then stabilizes whenthe density of nodes becomes sufficiently large The reasonis that when sensor nodes become more densely dispersedthey will have higher probability to be affiliated with a LNwhich is close to the data sink EEBRHM will generate moreLNs than BRH-MDG by balancing the degree of each nodewhichmakes the average tour length of EEBRHM longer thanBRH-MDG In contrast the average tour length of SHDG islonger than BRH-MDG and EEBRHM Because SHDGmustvisit one hop range of each node its pausing locations ofmobile collector will become much more than BRH-MDGand EEBRHMwith the continuously increased density Sincethe mobile collector goes along the fixed tours in the sensorfield the tour length of CME is a constant which in generalis bigger than the average tour length of BRH-MDG andEEBRHM
52 The Impact of Communication Radius To evaluate theperformance of different algorithms with changing the com-munication radius the comparison is shown in Figure 3 Thesimulation setting of the experiments is that the number ofnodes 119873 and the area are fixed at 400 and 200 lowast 200m2respectively And the bounded-hop 119889 is set as 2 and 3 respec-tively Figure 3(a) illustrates the performance of EEBRHMand BRH-MDG as a function of communication radius 119877in terms of network lifetime As shown in Figure 3(a) thenetwork lifetime of EEBRHM quit outperforms that of BRH-MDG with 119889 = 2 or 3 and it becomes more obvious when 119877increases When 119877 is 60m the network lifetime of EEBRHMis 10 times longer than that of BRH-MDG The reason hasbeen explained in Section 51
In Figure 3(b) the performance of EEBRHM and BRH-MDG as a function of communication radius 119877 in terms oftour length is compared From Figure 3(b) we can know thatthe tour length of EEBRHM and BRH-MDG first graduallyreduces as 119877 increases and then stabilizes when 119877 becomessufficiently long For the same reason as above the averagetour length of EEBRHM in this case is longer than BRH-MDG Furthermore for BRH-MDG as 119889 increases the LNscan ownmore subordinate nodeswhichwillmake the averagetour length of BRH-MDG became shorter When the 119877 is60m and 119889 is 2 the tour length of BRH-MDG is near 0
6 Conclusion
In this paper we have studied the energy consumptionmodelof the WSN with mobile data gathering with bounded-hop LNs A new novel algorithm EEBRHM is proposedto optimize the network lifetime of WSN Extensive sim-ulations have been carried out to prove the efficiency ofthe protocol The results demonstrate that the proposedalgorithm can greatly prolong the network lifetime of theWSN with bounded relay hop and obtain about 73 times
8 International Journal of Distributed Sensor Networks
improvement on the network lifetime compared with BRH-MDG Moreover the tour length of EEBRHM is shorter thanSHDG and CME and is only longer than BRH-MDG by 35
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgments
This work is sponsored by the National Natural ScienceFoundation of China under Grant nos 61173169 and 61103203and the State Key Program of National Natural Science ofChina under Grant no 61232001F02
References
[1] Z Rezaei and S Mobininejad ldquoEnergy saving in wirelesssensor networksrdquo International Journal of Computer Science ampEngineering Survey vol 3 no 1 pp 23ndash37 2012
[2] W R Heinzelman A Chandrakasan and H Balakrish-nan ldquoEnergy-efficient communication protocol for wirelessmicrosensor networksrdquo in Proceedings of the 33rd AnnualHawaii International Conference on System Siences (HICSS rsquo33)pp 3005ndash3014 IEEE Computer SocietyWashington DC USAJanuary 2000
[3] O Younis and S Fahmy ldquoHEED a hybrid energy-efficientdistributed clustering approach for ad hoc sensor networksrdquoIEEE Transactions on Mobile Computing vol 3 no 4 pp 366ndash379 2004
[4] X Y Kui Y Sheng H K Du and J B Liang ldquoConstructinga CDS-based network backbone for data collection in wirelesssensor networksrdquo International Journal of Distributed SensorNetworks vol 2013 Article ID 258081 12 pages 2013
[5] X Ma J Gao W Wang and J Wang ldquoA virtual-ring-baseddata storage and retrieval scheme in wireless sensor networksrdquoInternational Journal of Distributed Sensor Networks vol 2012Article ID 763015 10 pages 2012
[6] S Lindsey and C S Raghavendra ldquoPEGASIS power-efficientgathering in sensor information systemsrdquo in Proceedings of theIEEE Aerospace Conference IEEE Computer Society pp 1125ndash1130 San Francisco Calif USA March 2002
[7] S-M Jung Y-J Han and T-M Chung ldquoThe concentricclustering scheme for efficient energy consumption in thePEGASISrdquo in Proceedings of the 9th International Conferenceon Advanced Communication Technology (ICACT rsquo07) pp 260ndash265 IEEE Computer Society Phoenix Park February 2007
[8] G J Wang T Wang W J Jia M Y Guo and J Li ldquoAdaptivelocation updates for mobile sinks in wireless sensor networksrdquoJournal of Supercomputing vol 47 no 2 pp 127ndash145 2009
[9] Q Zhang Z-P Xie B Ling W-W Sun and B-L Shi ldquoMax-imum lifetime data gathering algorithm for wireless sensornetworksrdquo Journal of Software vol 16 no 11 pp 1946ndash1957 2005(Chinese)
[10] W Liang and Y Liu ldquoOnline data gathering for maximizingnetwork lifetime in sensor networksrdquo IEEE Transactions onMobile Computing vol 6 no 1 pp 2ndash11 2007
[11] Z F Liao J X Wang S G Zhang J N Cao and G YMin ldquoMinimizing movement for target coverage and network
connectivity in mobile sensor networksrdquo IEEE Transactions onParallel and Distributed Systems 2014
[12] Y Wu S Fahmy and N B Shroff ldquoOn the construction of amaximum-lifetime data gathering tree in sensor networks NP-completeness and approximation algorithmrdquo in Proceedings ofthe 27th IEEE Communications Society Conference on ComputerCommunications (INFOCOM rsquo08) IEEE Computer Society pp356ndash360 Phoenix Ariz USA April 2008
[13] J-B Liang J-X Wang and J-E Chen ldquoOn the construction ofa delay-constrained maximum lifetime tree in wireless sensornetworksrdquo Acta Electronica Sinica vol 38 no 2 pp 345ndash3512010
[14] K Karenos and V Kalogeraki ldquoTraffic management in sensornetworks with amobile sinkrdquo IEEE Transactions on Parallel andDistributed Systems vol 21 no 10 pp 1515ndash1530 2010
[15] X Xu J Luo and Q Zhang ldquoDelay tolerant event collection insensor networks with mobile sinkrdquo in Proceedings of the IEEEINFOCOM March 2010
[16] MZhao andYYang ldquoBounded relay hopmobile data gatheringin wireless sensor networksrdquo IEEE Transactions on Computersvol 61 no 2 pp 265ndash277 2012
[17] R C Shah S Roy S Jain and W Brunette ldquoData MULEsmodeling and analysis of a three-tier architecture for sparsesensor networksrdquo Ad Hoc Networks vol 1 no 2-3 pp 215ndash2332003
[18] S Jain R C Shah W Brunette G Borriello and S RoyldquoExploiting mobility for energy efficient data collection inwireless sensor networksrdquo Mobile Networks and Applicationsvol 11 no 3 pp 327ndash339 2006
[19] D Jea A Somasundara andM Srivastava ldquoMultiple controlledmobile elements (data mules) for data collection in sensor net-worksrdquo in Proceedings of the 1st IEEE International Conferenceon Distributed Computing in Sensor Systems (DCOSS rsquo05) pp244ndash257 July 2005
[20] A A Somasundara A Ramamoorthy and M B SrivastavaldquoMobile element scheduling with dynamic deadlinesrdquo IEEETransactions on Mobile Computing vol 6 no 4 pp 395ndash4102007
[21] MMa and Y Yang ldquoData gathering in wireless sensor networkswith mobile collectorsrdquo in Proceedings of the 22nd IEEE Inter-national Parallel and Distributed Processing Symposium (IPDPSrsquo08) April 2008
[22] M Ma and Y Yang ldquoSenCar an energy-efficient data gatheringmechanism for large-scale multihop sensor networksrdquo IEEETransactions on Parallel and Distributed Systems vol 18 no 10pp 1476ndash1488 2007
[23] B Kusy H Lee M Wicke N Milosavljevic and L GuibasldquoPredictive QoS routing to mobile sinks in wireless sensornetworksrdquo in Proceedings of the International Conference onInformation Processing in Sensor Networks (IPSN rsquo09) pp 109ndash120 April 2009
[24] J Luo and J-P Hubaux ldquoJoint mobility and routing for lifetimeelongation in wireless sensor networksrdquo in Proceedings of theIEEE 24th Annual Joint Conference of the IEEE Computerand Communications Societies (INFOCOM rsquo05) pp 1735ndash1746March 2005
[25] M Zhao M Ma and Y Yang ldquoEfficient data gathering withmobile collectors and space-division multiple access techniquein wireless sensor networksrdquo IEEE Transactions on Computersvol 60 no 3 pp 400ndash417 2011
[26] S M Lasassmeh and J M Conrad ldquoTime synchronization inwireless sensor networks a surveyrdquo in Proceedings of the IEEE
International Journal of Distributed Sensor Networks 9
Conference Energizing Our Future (SoutheastCon rsquo10) pp 242ndash245 Concord NC USA March 2010
[27] T H Cormen C E Leiserson R L Rivest and C SteinIntroduction to Algorithms The MIT Press Cambridge MassUSA 2nd edition 2001
[28] A Chowdhury A Ghosh S Sinha and S Das ldquoA novel geneticalgorithm to solve travelling salesman problem and blockingflow shop scheduling problemrdquo International Journal of Bio-Inspired Computation vol 5 no 5 pp 303ndash314 2013
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
would be greatly reduced Furthermore to maximize energysaving the mobile collector could traverse the transmissionrange of each sensor node so that each sensing data packetcan be transmitted to the mobile collector in one hop Butbecause of low speed of the mobile collector and long lengthof moving path the latency of WSN may be very longin the process of data gathering which will not meet therequirement of delay in some time-sensitive applications [16]
In fact the latency of multihop data gathering is muchshorter than mobile data gathering Thus in the time-sensitive applications we have to make a trade-off betweenthe energy saving and the data gathering latency A generalway in some papers is to shorten the latency of mobiledata gathering by setting local data aggregation nodes (LNs)Specifically a part of sensor nodes will be selected as the LNswhich can aggregate the local sensing data from its affiliatedsensor nodes within a limited hop count These LNs couldtemporarily cache the sensing data and upload them to themobile collector when it arrives [16] Although the protocolsof setting LNs can effectively reduce the latency ofmobile datagathering they do not consider the energy consumption ofthe whole network which will have an adverse impact on thelifetime of network
The main contributions of this paper can be summarizedas follows
(1) By modeling and analyzing the energy consumptionofwhole networkwe characterize the energy-efficientbounded relay hop mobile data gathering as anoptimization problem
(2) We propose an efficient algorithm which is energy-efficient and can achieve a longer lifetime of networkthan these existing mobile data gathering protocolswith LNs
(3) The performance of the proposed algorithm is eval-uated by comparing with other three existing mobiledata gathering schemes Simulation results illustratethat the proposed algorithms achieve distinctive per-formance
The rest of the paper is organized as follows Section 2reviews the related work on mobile data gathering Section 3models and analyzes the energy consumption of wholenetwork and then formulates the problem Section 4 givesdetailed description on our algorithm and analyzes the per-formance of it The performance evaluation is demonstratedin Section 5 Finally Section 6 concludes the paper
2 Related Work
In this section let us briefly review some recent work onmobile data gathering in WSNs The mobile data gatheringprotocols can be divided into two categories according to themobility pattern
In the first category the mobility is uncontrollable andthe moving path of the mobile collector is either randomor determinate For example [17ndash19] In [17] Shah et alproposed a scheme in which a special type of mobile nodesis used as forwarding agents to facilitate connectivity among
static sensors and transport data with random mobilityTo improve the work in [17] Jain et al [18] presented ananalytical model to understand the key performance metricsof the WSNs with mobile collector such as data transferlatency of data gathering and power consumptions Jea etal [19] set some fixed straight moving paths and the mobilecollector gathered sensing data from the nodes near thesepaths A common advantage of these protocols is that theycan get high stability and reliability and it is easy for networkmaintaining But they lack agility and cannot apply to highlydynamic environment
The mobility of the second category is controlled inwhich mobile collector can freely move to any positionof the WSN and its path can be constructed for specificpurpose such as [20ndash25] Furthermore in this categorythe approaches can be classified into three subclasses Inthe first subclass the mobile collector visits each sensornode or traverses the transmission range of each node andgathers sensing data in one hop [20 21] To ensure nodata loss Somasundara et al [20] studied the scheduling ofmobile collector Ma and Yang [21] proposed tour planningalgorithm for getting a short moving path and ensuringall data gathering to be completed in one hop and it canachieve perfect uniformity of energy consumption Althoughthese protocols can minimize the energy cost by completelyavoiding multihop relays they may make a long latencyof data gathering especially in a large-scale WSN In thesecond subclass mobile collector gathers sensing data fromthe sensor nodes nearmoving path bymultihop transmissionMa and Yang [22] proposed a moving trajectory planningalgorithm via finding some turning points which is adaptiveto the sensor nodes distribution and can effectively avertbarriers on the trajectory In this approach the sensornodes along each moving line segment forward packets tothe mobile collector by a multihop fashion For improvingrouting reliability Kusy et al [23] constructed an algorithmby introducing the mobility graph which can encode theknowledge of likely mobility patterns within the networkWe can extract the mobility graph from training data andpredict future relay nodes by applying it on the mobilecollector to ensure uninterrupted data streams Luo andHubaux [24] thought that sensing data should be gatheredvia multihop relays while the mobile collector moves alongthe perimeter of theWSN which is considered as the optimalpath for themobile collectorThese protocols could effectivelyreduce or limit the length of moving path to a certain levelHowever they do not set any limitation on the hop countthat will make network lifetime (or a certain level of energyefficiency) uncertain In the third subclass data transmissionpatterns and moving path planning are both considered Forexample Zhao et al [25] constructed a mobile data gatheringalgorithm by considering the full utilization of concurrentdata uploading and the minimization of path length In thealgorithm multiple sensor nodes can upload sensing datato the mobile collector in one hop at the same time whicheffectively reduces the time of data uploading Zhao and Yang[16] proposed an algorithm called BRH-MDG (boundedrelay hop mobile data gathering) in which the length ofmoving path is minimized and the local data aggregation is
International Journal of Distributed Sensor Networks 3
guaranteed in bounded-hop count In this paper our workfalls into the third subclass Via modeling and analyzing theenergy consumption of whole network we construct a newalgorithm called energy-efficient bounded relay hop mobiledata gathering (EEBRHM) which can achieve much longernetwork lifetime than others and only increases a little delaycompared with the BRH-MDG
3 Preliminaries
31 Network Model Assume that V1 V2 V
119899are sensor
nodes in a sensing field and V0is the sink These 119899 nodes are
randomly deployed in a119872 times119872 field and form a WSN ThisWSN can be represented by a connective undirected graph119866(119881 119864) where 119881 is the set of sensor nodes and 119864 is the setof edges If the distance between nodes V
119894and V
119895is shorter
than themaximum communication distance there is an edge(V119894 V119895) isin 119864 |119881| = 119899 + 1 is the number of sensor nodes and
|119864| = 119898 is the number of edgesThere are some characteristicsin the WSN
(1) The sensor nodes are stationary after deployment(2) In the process of data gathering for a relay node
both the received data and its sensing data will beaggregated into a certain length packet Then thepacket will be sent to the next node
(3) After deployment the energy of sensor nodes cannotbe recharged and the initial energy of each sensornodes is the same
(4) There has been a time synchronization scheme in theWSN And the time accuracy can be in milliseconds[26]
In general we can assume that the amount of energyrequired to transmittingreceiving 1 bit of data is a fixed valueLet 119864119905denote energy required by transmitting and 119864
119903denote
energy required by receiving At the same time we assumethat there are some excellent congestion control strategies inthe WSN which can avoid congestion and retransmission inthe process of data gathering
32 Definitions In the protocols of mobile data gatheringwith bounded-hop LNs such as BRH-MDG algorithm theWSN is divided into some subtrees whose roots are the LNsBecause mobile collector can be recharged the lifetime ofnetwork is determined by the life cycle of subtrees Accordingto paper [13] we give several definitions which will be used inthis paper
Definition 1 A round is defined as the process of gatheringthe sensing data from sensor nodes to the sink in one cycleby mobile collector which could be equal to the time spent atthe moving path by mobile collector
Definition 2 The lifetime of a node V119894is defined as the
number of rounds in the node survival period We can judgewhether one sensor node is survival by a formula 119864(V
119894) gt 119890
when the residual energy of one node is lower than it the
node is dead Specifically119864(V119894) represents the residual energy
of one node and 119890 is a threshold In this paper each sensornode can aggregate received data and itself sensing data intoa 119897 bits packet Formula (1) is used to count the lifetime of anode
119871node (V119894) = lfloor119864 (V119894)
119897 (119864119905minus 119864119903) + 119863 (V
119894) 119897119864119903
rfloor (1)
where119863(V119894) represents the degree of a node in the subtree
Definition 3 The lifetime of a subtree is the lifetime of the firstdead node in the subtree defined as the following formula
119871 tree (119879119873) = min 119871node (V119894) V119894isin 119879119873 (2)
where 119879119873
represents one subtree and V119894is a node in the
subtree
Definition 4 The lifetime of the whole network is the shortestlifetime of the subtrees defined as the following formula
119871119873(119866) = min
119879119873isin119879119875
119871 tree (119879119873) (3)
where119879119873represents one subtree and TP is the set of subtrees
33 Problem Statement As shown above the subtrees withthe shortest lifetime determine the lifetime of the whole net-work Thus if we want to optimize the energy consumptionof the whole network we should maximize the lifetime of allsubtrees
In Formula (1) 119897 119864119903 and 119864
119905are constant and 119864(V
119894) is
initialized as a constant Only 119863(V119894) is a variable so it is
the main optimization objective To extract the constant 119864119903
119864119905from the denominator and use Formula (1) to substitute
119871node(119881119894) in formula (2) formula (2) can be transformed intoan equivalent form
119871 tree (119879119873) = minlfloor119864 (V119894)
119863 (V119894) + 119888
rfloor V119894isin 119879119873 (4)
where 119888 = (119864119905minus 119864119903)119864119903 Based on formulas (3) and (4) the
maximum lifetime of the whole network can be calculated bythe following formula
max 119871119873(119866) = max min
119879119873isin119879119875
lfloor
119864 (V119894)
119863 (V119894) + 119888
rfloor V119894isin 119879119873 (5)
In formula (5) to maximize the lifetime of subtrees 119863(V119894) of
each node of subtrees should be minimized The problem ofhow to obtain a spanning tree withminimumnode degree onthe premise of hop bounded has been proved to be NP-hardin [13]
4 Design of EEBRHM
41 Algorithm Description Base on formula (5) our problemof maximizing the lifetime of subtrees can be transformedto minimize the degree of each node of subtrees To solve
4 International Journal of Distributed Sensor Networks
(a) An example of BRH-MDG algorithm (b) After balancing the degree of each node
Figure 1 An example to show the effect of balancing the degree of each node
this problem a general idea is to balance degrees of nodesin subtrees such as BRH-MDG algorithm As shown inFigure 1(a) the BRH-MDG algorithm divides the networkinto three subtrees Since the process of generating subtreesis random the distribution of nodesrsquo degrees is not uniformIn Figure 1(a) the dotted lines represent the moving paththe black nodes represent the nodes which have maximumdegrees in subtrees and the maximum degree among theseblack nodes is 5 By applying a method to balance the degreesof nodes in Figure 1(a) a new set of subtrees can be obtainedas shown in Figure 1(b) and the maximum degree is reducedto 2 According to Definitions 3 and 4 the lifetime of networkin Figure 1(b) is twice more than that in Figure 1(a)
The example demonstrates that the key point to optimizethe energy consumption of the whole WSN is to balance thedistribution of nodesrsquo degree in their own subtrees In thispaper we propose a new algorithm called EEBRHM whichuses the number of children of nodes in the initial tree ascontrol metric to adjust the degree of each node The detailof EEBRHM algorithm is shown in Algorithm 1 and Table 1
First of all EEBRHM algorithm will construct shortestpath trees (SPTs) which should cover all sensor nodes withminimum hops in the WSN (see Algorithm 1 line (1))Because a large WSN may not be fully connected it maybe constituted by several subnetworks and thus it needsmore than one SPT to cover all sensor nodes In EEBRHMalgorithm generally the root of SPT is the data sink For thenodes from other subnetworks which are far from the datasink we can set virtual roots in the center of the subnetworksThus the new algorithm can be applied to not only connectedWSNs but also disconnectedWSNs which is one of themainadvantages for the mobile data gathering Moreover beforeall sensor nodes being deployed each node will be assigneda unique integer ID which will be used to allocate separatedelay time slot for each sensor nodeThen EEBRHMwill usethe number of sensor nodesrsquo children as metric to balancethe degrees of sensor nodes (see Algorithm 1 lines (2)ndash(18))Specifically the energy of root is assumed to be infinite so itsdegree does not need to be balanced For a sensor node if itsparent node is not root and there exists a neighbor node withthe minimum number of children and at the same height asits parent in its own SPT (to effectively balance theminimum
number of childrenmust be less than original value at least 2)this neighbor node will be the new parent of the node Afterupdating the new parent the old parent and the new parentnode will renew the numbers of children and inform all theirown neighbor nodes
The next task of EEBRHM is to determine the LNs Sincethe LNs are more close to the data sink and the length ofmoving path119880 is shorter than other sensor nodes the sensornodes which is more close to the data sink will more likelyto be LNs To determine LNs firstly all sensor nodes areassigned to one of three separate large time slots accordingto the distance between sensor nodes and the data sink Theformula (119879
119899+IDlowast1199051015840) 119899 = 0 1 2 ensures that each nodewill be
assigned to a separate small time slot so that the sensor nodeswill not interfere with each other in the process of algorithmMoreover 119879
119899is measured in seconds and 1199051015840 is measured in
milliseconds A node which is more close to the data sinkwill be assigned to an earlier large time slot and would havepriority to be a LN At last TSP algorithm is used to constructan approximate shortest tour 119880 visiting V
0and all the LNs
42 Performance Analysis To evaluate the performance ofEEBRHM firstly we will analyze the time complexity ofEEBRHM Assume a WSN has 119870 disconnected subnetworksand 119873 is the total number of sensor nodes in the WSN (1 ⩽119870 ⩽ 119873) For a subnetwork 119896 (119896 = 1 2 119870) it would take119874((119873119896)
2) time to find a SPT [27] where 119873
119896represents the
number of sensor nodes in subnetwork 119896 In our algorithm itruns in119874(119873) time to balance the nodesrsquo degree and constructLNs And it runs in 119874(1198732) time to find tour 119880 at most byusing TSP algorithm [28] Thus the total time complexity ofEEBRHM is119874((119873
119896)
2) +119874(119873)+119874(119873
2) In the worst case the
time complexity of EEBRHM is 119874(1198732)Due to balancing degree of sensor nodes the length
of moving path 119880 of EEBRHM may be longer than BRH-MDG algorithm which will increase a bit latency of networkby EEBRHM But the performance of network lifetime ofEEBRHM is much better than BRH-MDG
EEBRHM is also one of mobile data gathering algorithmsand the data transmission between nodes and mobile sink isthe main part of communication overhead The mechanism
International Journal of Distributed Sensor Networks 5
Algorithm EEBRHMInput Network topology 119866(119881 119864) the relay hop bound 119889
and the static data sink V0
Output A set of LNs a set of sub-trees and the tour 119880visiting the LNs and the data sink
(1) Construct SPTs for 119866 that cover all the vertices in 119881(2) for each node V do (3) if119867(V) ⩾ 2 (4) cn = CN(Parent (V)) V119905 = V(5) for each node V1015840 in Nb(V)
(6) if119867(V1015840) minus 119867(V) == minus1 (7) if CN(V1015840) lt 119888119899 (8) cn = CN(V1015840) V119905 = V1015840(9)
(10)
(11)
(12) if cn + 2 lt CN(Parent (V)) (13) Parent(V) ⩽ vt(14) Broadcast ACKMessage(15) When received the ACKMessage
the related nodes will make response(16)
(17)
(18) (19) Set three large time slots 119879
0 1198791 1198792 1198790lt 1198791lt 1198792
and set a constant 1199051015840 1199051015840 ≪ 1198790
(20) for each node V do (21) Get the unique integer ID of V(22) if V in one hop of V
0
(23) Allocate a delay time slot (1198790+ ID lowast 1199051015840) to V
(24)
(25) if V is out of one hop of V0but it is in two hop
(26) Allocate a delay time slot (1198791+ ID lowast 1199051015840) to V
(27)
(28) if V is out of two hop of V0
(29) Allocate a delay time slot (1198792+ ID lowast 1199051015840) to V
(30)
(31) (32) In the delay time slot of each node V (33) if V is not belong to a LN (34) Let it be a LN and broadcast a probe packet
with 119889 hops of time to live(35)
(36) if other nodes received the probe packet (37) if it is not a LN (38) if it is the child of relay node (39) Let it be an affiliated node of the source
node of the probe packet(40)
(41)
(42)
(43) (44) Use the TSP algorithm to find an approximate shortest
tour 119880 visiting V0and all the LNs
Algorithm 1 EEBRHM algorithm
6 International Journal of Distributed Sensor Networks
Table 1 Notations used in EEBRHM algorithm
Functions119867(V) This function returns the height of V in its own SPTParent(V) This function returns the parent node of VCN(V) This function returns the number of children nodes of VNb(V) This function returns a set of neighbor nodes of V
Variablescn The variable is used to cache the number of children of node VV119905 The variable is used to cache the node V which is relevant to cn
0
1000
2000
3000
4000
5000
6000
7000
8000
100 200 300 400 500
Net
wor
k lif
etim
e (ro
unds
)
Node number
EEBRHMBRH-MDG
(a) Network lifetime comparison
500
600
700
800
900
1000
1100
1200
1300
100 200 300 400 500
Tour
leng
th (m
)
Node number
EEBRHMBRH-MDG
CMESHDG
(b) Length of tour comparison
Figure 2 Performance comparison with changing the density of sensor nodes
of constructing the path of mobile sink of EEBRHM isthe same as BRH-MDG Thus the communication cost ofEEBRHM is just the same as BRH-MDG
5 Simulation Results
In the simulation we assume that 119873 sensor nodes arerandomly distributed over an119872times119872 square area and the datasink is set in the center of the area Furthermore we compareour algorithm with BRH-MDG [16] SHDG [21] and CME[19] SHDG is one kind of one hop mobile data gatheringscheme This algorithm will choose some pausing locationsfor mobile collector from a set of candidate locations so thatmobile collector can gather sensor data fromnodes in a singlehop Although we can use TSP algorithm to construct anapproximate shortest tour 119880 for mobile collector to visit allpausing locations and the data sink the latency of networkof SHDG is too long CME generates some parallel straighttours in advance which across the WSN Then the mobilecollector will traverse along these tours while gathering datafrom the sensor nodes nearby the tours And the other sensornodes would send their own sensor data to the nodes nearby
tours by multihop relays The latency of network of CMEis acceptable but due to randomly choosing a relay nodethe network lifetime achieved by CME may be poor Thedescription of BRH-MDGhas beenmentioned in Section 31All simulations are performed 30 times and average values oftheir results are taken as final results
51 The Impact of Sensor Node Density To compare theperformance of different algorithms we change the densityof sensor nodes The simulation setting in the experiments isthat communication radius 119877 bounded-hop 119889 and the areaare set as 30m 2 and 200 lowast 200m2 respectively Figure 2(a)plots the performance of EEBRHM and BRH-MDG as afunction of density of sensor nodes in terms of networklifetime From Figure 2(a) we can find that when the numberof sensor nodes becomes larger the advantage of EEBRHMover BRH-MDG become more obvious The performance ofnetwork lifetime of EEBRHM is 47 times than BRH-MDGon average There are two reasons to explain this First thedegrees of sensor nodes in BRH-MDGare very unevenwhichleads to uneven node energy consumption and thereforemake the network lifetime short Our EEBRHM algorithm
International Journal of Distributed Sensor Networks 7
Net
wor
k lif
etim
e (ro
unds
)
0
2000
4000
6000
8000
10000
20 30 40 50 60Communication radius (m)
EEBRHM d = 2
BRH-MDG d = 2
EEBRHM d = 3
BRH-MDG d = 3
(a) Network lifetime comparison
0
300
600
900
1200
1500
20 30 40 50 60
Tour
leng
th (m
)
Communication radius (m)
EEBRHM d = 2
BRH-MDG d = 2
EEBRHM d = 3
BRH-MDG d = 3
(b) Length of tour comparison
Figure 3 Performance comparison with changing the bounded 119889
can balance the degree of each node thus its performance ofnetwork lifetime is much better than BRH-MDGThe secondreason is that since BRH-MDGdoes not balance the degree ofeach node when the density of sensor nodes becomes largerthe degree of nodes in the WSN may increase very quicklytherefore the network lifetime of BRH-MDG would becomeshorter and shorter
Figure 2(b) shows the performance of four algorithms asa function of density of sensor nodes in terms of tour lengthThe simulation setting is the same as above For SHDG BRH-MDG and EEBRHM TSP algorithm is used to generate anapproximate shortest tour and mobile collector will movealong it For CME the parallel straight tours traversing thesensor field are 100m apart from each other One of thesetours must go through the center of the fieldThemobile col-lector can change onto other tours bymoving along the sensorfield border Both SHDG and CME algorithms are imple-mented in a centralized fashion BRH-MDG and EEBRHMare implemented in a distributed fashion In Figure 2(b) itis easy to observe that the tour length of BRH-MDG andEEBRHMgradually increases at first and then stabilizes whenthe density of nodes becomes sufficiently large The reasonis that when sensor nodes become more densely dispersedthey will have higher probability to be affiliated with a LNwhich is close to the data sink EEBRHM will generate moreLNs than BRH-MDG by balancing the degree of each nodewhichmakes the average tour length of EEBRHM longer thanBRH-MDG In contrast the average tour length of SHDG islonger than BRH-MDG and EEBRHM Because SHDGmustvisit one hop range of each node its pausing locations ofmobile collector will become much more than BRH-MDGand EEBRHMwith the continuously increased density Sincethe mobile collector goes along the fixed tours in the sensorfield the tour length of CME is a constant which in generalis bigger than the average tour length of BRH-MDG andEEBRHM
52 The Impact of Communication Radius To evaluate theperformance of different algorithms with changing the com-munication radius the comparison is shown in Figure 3 Thesimulation setting of the experiments is that the number ofnodes 119873 and the area are fixed at 400 and 200 lowast 200m2respectively And the bounded-hop 119889 is set as 2 and 3 respec-tively Figure 3(a) illustrates the performance of EEBRHMand BRH-MDG as a function of communication radius 119877in terms of network lifetime As shown in Figure 3(a) thenetwork lifetime of EEBRHM quit outperforms that of BRH-MDG with 119889 = 2 or 3 and it becomes more obvious when 119877increases When 119877 is 60m the network lifetime of EEBRHMis 10 times longer than that of BRH-MDG The reason hasbeen explained in Section 51
In Figure 3(b) the performance of EEBRHM and BRH-MDG as a function of communication radius 119877 in terms oftour length is compared From Figure 3(b) we can know thatthe tour length of EEBRHM and BRH-MDG first graduallyreduces as 119877 increases and then stabilizes when 119877 becomessufficiently long For the same reason as above the averagetour length of EEBRHM in this case is longer than BRH-MDG Furthermore for BRH-MDG as 119889 increases the LNscan ownmore subordinate nodeswhichwillmake the averagetour length of BRH-MDG became shorter When the 119877 is60m and 119889 is 2 the tour length of BRH-MDG is near 0
6 Conclusion
In this paper we have studied the energy consumptionmodelof the WSN with mobile data gathering with bounded-hop LNs A new novel algorithm EEBRHM is proposedto optimize the network lifetime of WSN Extensive sim-ulations have been carried out to prove the efficiency ofthe protocol The results demonstrate that the proposedalgorithm can greatly prolong the network lifetime of theWSN with bounded relay hop and obtain about 73 times
8 International Journal of Distributed Sensor Networks
improvement on the network lifetime compared with BRH-MDG Moreover the tour length of EEBRHM is shorter thanSHDG and CME and is only longer than BRH-MDG by 35
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgments
This work is sponsored by the National Natural ScienceFoundation of China under Grant nos 61173169 and 61103203and the State Key Program of National Natural Science ofChina under Grant no 61232001F02
References
[1] Z Rezaei and S Mobininejad ldquoEnergy saving in wirelesssensor networksrdquo International Journal of Computer Science ampEngineering Survey vol 3 no 1 pp 23ndash37 2012
[2] W R Heinzelman A Chandrakasan and H Balakrish-nan ldquoEnergy-efficient communication protocol for wirelessmicrosensor networksrdquo in Proceedings of the 33rd AnnualHawaii International Conference on System Siences (HICSS rsquo33)pp 3005ndash3014 IEEE Computer SocietyWashington DC USAJanuary 2000
[3] O Younis and S Fahmy ldquoHEED a hybrid energy-efficientdistributed clustering approach for ad hoc sensor networksrdquoIEEE Transactions on Mobile Computing vol 3 no 4 pp 366ndash379 2004
[4] X Y Kui Y Sheng H K Du and J B Liang ldquoConstructinga CDS-based network backbone for data collection in wirelesssensor networksrdquo International Journal of Distributed SensorNetworks vol 2013 Article ID 258081 12 pages 2013
[5] X Ma J Gao W Wang and J Wang ldquoA virtual-ring-baseddata storage and retrieval scheme in wireless sensor networksrdquoInternational Journal of Distributed Sensor Networks vol 2012Article ID 763015 10 pages 2012
[6] S Lindsey and C S Raghavendra ldquoPEGASIS power-efficientgathering in sensor information systemsrdquo in Proceedings of theIEEE Aerospace Conference IEEE Computer Society pp 1125ndash1130 San Francisco Calif USA March 2002
[7] S-M Jung Y-J Han and T-M Chung ldquoThe concentricclustering scheme for efficient energy consumption in thePEGASISrdquo in Proceedings of the 9th International Conferenceon Advanced Communication Technology (ICACT rsquo07) pp 260ndash265 IEEE Computer Society Phoenix Park February 2007
[8] G J Wang T Wang W J Jia M Y Guo and J Li ldquoAdaptivelocation updates for mobile sinks in wireless sensor networksrdquoJournal of Supercomputing vol 47 no 2 pp 127ndash145 2009
[9] Q Zhang Z-P Xie B Ling W-W Sun and B-L Shi ldquoMax-imum lifetime data gathering algorithm for wireless sensornetworksrdquo Journal of Software vol 16 no 11 pp 1946ndash1957 2005(Chinese)
[10] W Liang and Y Liu ldquoOnline data gathering for maximizingnetwork lifetime in sensor networksrdquo IEEE Transactions onMobile Computing vol 6 no 1 pp 2ndash11 2007
[11] Z F Liao J X Wang S G Zhang J N Cao and G YMin ldquoMinimizing movement for target coverage and network
connectivity in mobile sensor networksrdquo IEEE Transactions onParallel and Distributed Systems 2014
[12] Y Wu S Fahmy and N B Shroff ldquoOn the construction of amaximum-lifetime data gathering tree in sensor networks NP-completeness and approximation algorithmrdquo in Proceedings ofthe 27th IEEE Communications Society Conference on ComputerCommunications (INFOCOM rsquo08) IEEE Computer Society pp356ndash360 Phoenix Ariz USA April 2008
[13] J-B Liang J-X Wang and J-E Chen ldquoOn the construction ofa delay-constrained maximum lifetime tree in wireless sensornetworksrdquo Acta Electronica Sinica vol 38 no 2 pp 345ndash3512010
[14] K Karenos and V Kalogeraki ldquoTraffic management in sensornetworks with amobile sinkrdquo IEEE Transactions on Parallel andDistributed Systems vol 21 no 10 pp 1515ndash1530 2010
[15] X Xu J Luo and Q Zhang ldquoDelay tolerant event collection insensor networks with mobile sinkrdquo in Proceedings of the IEEEINFOCOM March 2010
[16] MZhao andYYang ldquoBounded relay hopmobile data gatheringin wireless sensor networksrdquo IEEE Transactions on Computersvol 61 no 2 pp 265ndash277 2012
[17] R C Shah S Roy S Jain and W Brunette ldquoData MULEsmodeling and analysis of a three-tier architecture for sparsesensor networksrdquo Ad Hoc Networks vol 1 no 2-3 pp 215ndash2332003
[18] S Jain R C Shah W Brunette G Borriello and S RoyldquoExploiting mobility for energy efficient data collection inwireless sensor networksrdquo Mobile Networks and Applicationsvol 11 no 3 pp 327ndash339 2006
[19] D Jea A Somasundara andM Srivastava ldquoMultiple controlledmobile elements (data mules) for data collection in sensor net-worksrdquo in Proceedings of the 1st IEEE International Conferenceon Distributed Computing in Sensor Systems (DCOSS rsquo05) pp244ndash257 July 2005
[20] A A Somasundara A Ramamoorthy and M B SrivastavaldquoMobile element scheduling with dynamic deadlinesrdquo IEEETransactions on Mobile Computing vol 6 no 4 pp 395ndash4102007
[21] MMa and Y Yang ldquoData gathering in wireless sensor networkswith mobile collectorsrdquo in Proceedings of the 22nd IEEE Inter-national Parallel and Distributed Processing Symposium (IPDPSrsquo08) April 2008
[22] M Ma and Y Yang ldquoSenCar an energy-efficient data gatheringmechanism for large-scale multihop sensor networksrdquo IEEETransactions on Parallel and Distributed Systems vol 18 no 10pp 1476ndash1488 2007
[23] B Kusy H Lee M Wicke N Milosavljevic and L GuibasldquoPredictive QoS routing to mobile sinks in wireless sensornetworksrdquo in Proceedings of the International Conference onInformation Processing in Sensor Networks (IPSN rsquo09) pp 109ndash120 April 2009
[24] J Luo and J-P Hubaux ldquoJoint mobility and routing for lifetimeelongation in wireless sensor networksrdquo in Proceedings of theIEEE 24th Annual Joint Conference of the IEEE Computerand Communications Societies (INFOCOM rsquo05) pp 1735ndash1746March 2005
[25] M Zhao M Ma and Y Yang ldquoEfficient data gathering withmobile collectors and space-division multiple access techniquein wireless sensor networksrdquo IEEE Transactions on Computersvol 60 no 3 pp 400ndash417 2011
[26] S M Lasassmeh and J M Conrad ldquoTime synchronization inwireless sensor networks a surveyrdquo in Proceedings of the IEEE
International Journal of Distributed Sensor Networks 9
Conference Energizing Our Future (SoutheastCon rsquo10) pp 242ndash245 Concord NC USA March 2010
[27] T H Cormen C E Leiserson R L Rivest and C SteinIntroduction to Algorithms The MIT Press Cambridge MassUSA 2nd edition 2001
[28] A Chowdhury A Ghosh S Sinha and S Das ldquoA novel geneticalgorithm to solve travelling salesman problem and blockingflow shop scheduling problemrdquo International Journal of Bio-Inspired Computation vol 5 no 5 pp 303ndash314 2013
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
guaranteed in bounded-hop count In this paper our workfalls into the third subclass Via modeling and analyzing theenergy consumption of whole network we construct a newalgorithm called energy-efficient bounded relay hop mobiledata gathering (EEBRHM) which can achieve much longernetwork lifetime than others and only increases a little delaycompared with the BRH-MDG
3 Preliminaries
31 Network Model Assume that V1 V2 V
119899are sensor
nodes in a sensing field and V0is the sink These 119899 nodes are
randomly deployed in a119872 times119872 field and form a WSN ThisWSN can be represented by a connective undirected graph119866(119881 119864) where 119881 is the set of sensor nodes and 119864 is the setof edges If the distance between nodes V
119894and V
119895is shorter
than themaximum communication distance there is an edge(V119894 V119895) isin 119864 |119881| = 119899 + 1 is the number of sensor nodes and
|119864| = 119898 is the number of edgesThere are some characteristicsin the WSN
(1) The sensor nodes are stationary after deployment(2) In the process of data gathering for a relay node
both the received data and its sensing data will beaggregated into a certain length packet Then thepacket will be sent to the next node
(3) After deployment the energy of sensor nodes cannotbe recharged and the initial energy of each sensornodes is the same
(4) There has been a time synchronization scheme in theWSN And the time accuracy can be in milliseconds[26]
In general we can assume that the amount of energyrequired to transmittingreceiving 1 bit of data is a fixed valueLet 119864119905denote energy required by transmitting and 119864
119903denote
energy required by receiving At the same time we assumethat there are some excellent congestion control strategies inthe WSN which can avoid congestion and retransmission inthe process of data gathering
32 Definitions In the protocols of mobile data gatheringwith bounded-hop LNs such as BRH-MDG algorithm theWSN is divided into some subtrees whose roots are the LNsBecause mobile collector can be recharged the lifetime ofnetwork is determined by the life cycle of subtrees Accordingto paper [13] we give several definitions which will be used inthis paper
Definition 1 A round is defined as the process of gatheringthe sensing data from sensor nodes to the sink in one cycleby mobile collector which could be equal to the time spent atthe moving path by mobile collector
Definition 2 The lifetime of a node V119894is defined as the
number of rounds in the node survival period We can judgewhether one sensor node is survival by a formula 119864(V
119894) gt 119890
when the residual energy of one node is lower than it the
node is dead Specifically119864(V119894) represents the residual energy
of one node and 119890 is a threshold In this paper each sensornode can aggregate received data and itself sensing data intoa 119897 bits packet Formula (1) is used to count the lifetime of anode
119871node (V119894) = lfloor119864 (V119894)
119897 (119864119905minus 119864119903) + 119863 (V
119894) 119897119864119903
rfloor (1)
where119863(V119894) represents the degree of a node in the subtree
Definition 3 The lifetime of a subtree is the lifetime of the firstdead node in the subtree defined as the following formula
119871 tree (119879119873) = min 119871node (V119894) V119894isin 119879119873 (2)
where 119879119873
represents one subtree and V119894is a node in the
subtree
Definition 4 The lifetime of the whole network is the shortestlifetime of the subtrees defined as the following formula
119871119873(119866) = min
119879119873isin119879119875
119871 tree (119879119873) (3)
where119879119873represents one subtree and TP is the set of subtrees
33 Problem Statement As shown above the subtrees withthe shortest lifetime determine the lifetime of the whole net-work Thus if we want to optimize the energy consumptionof the whole network we should maximize the lifetime of allsubtrees
In Formula (1) 119897 119864119903 and 119864
119905are constant and 119864(V
119894) is
initialized as a constant Only 119863(V119894) is a variable so it is
the main optimization objective To extract the constant 119864119903
119864119905from the denominator and use Formula (1) to substitute
119871node(119881119894) in formula (2) formula (2) can be transformed intoan equivalent form
119871 tree (119879119873) = minlfloor119864 (V119894)
119863 (V119894) + 119888
rfloor V119894isin 119879119873 (4)
where 119888 = (119864119905minus 119864119903)119864119903 Based on formulas (3) and (4) the
maximum lifetime of the whole network can be calculated bythe following formula
max 119871119873(119866) = max min
119879119873isin119879119875
lfloor
119864 (V119894)
119863 (V119894) + 119888
rfloor V119894isin 119879119873 (5)
In formula (5) to maximize the lifetime of subtrees 119863(V119894) of
each node of subtrees should be minimized The problem ofhow to obtain a spanning tree withminimumnode degree onthe premise of hop bounded has been proved to be NP-hardin [13]
4 Design of EEBRHM
41 Algorithm Description Base on formula (5) our problemof maximizing the lifetime of subtrees can be transformedto minimize the degree of each node of subtrees To solve
4 International Journal of Distributed Sensor Networks
(a) An example of BRH-MDG algorithm (b) After balancing the degree of each node
Figure 1 An example to show the effect of balancing the degree of each node
this problem a general idea is to balance degrees of nodesin subtrees such as BRH-MDG algorithm As shown inFigure 1(a) the BRH-MDG algorithm divides the networkinto three subtrees Since the process of generating subtreesis random the distribution of nodesrsquo degrees is not uniformIn Figure 1(a) the dotted lines represent the moving paththe black nodes represent the nodes which have maximumdegrees in subtrees and the maximum degree among theseblack nodes is 5 By applying a method to balance the degreesof nodes in Figure 1(a) a new set of subtrees can be obtainedas shown in Figure 1(b) and the maximum degree is reducedto 2 According to Definitions 3 and 4 the lifetime of networkin Figure 1(b) is twice more than that in Figure 1(a)
The example demonstrates that the key point to optimizethe energy consumption of the whole WSN is to balance thedistribution of nodesrsquo degree in their own subtrees In thispaper we propose a new algorithm called EEBRHM whichuses the number of children of nodes in the initial tree ascontrol metric to adjust the degree of each node The detailof EEBRHM algorithm is shown in Algorithm 1 and Table 1
First of all EEBRHM algorithm will construct shortestpath trees (SPTs) which should cover all sensor nodes withminimum hops in the WSN (see Algorithm 1 line (1))Because a large WSN may not be fully connected it maybe constituted by several subnetworks and thus it needsmore than one SPT to cover all sensor nodes In EEBRHMalgorithm generally the root of SPT is the data sink For thenodes from other subnetworks which are far from the datasink we can set virtual roots in the center of the subnetworksThus the new algorithm can be applied to not only connectedWSNs but also disconnectedWSNs which is one of themainadvantages for the mobile data gathering Moreover beforeall sensor nodes being deployed each node will be assigneda unique integer ID which will be used to allocate separatedelay time slot for each sensor nodeThen EEBRHMwill usethe number of sensor nodesrsquo children as metric to balancethe degrees of sensor nodes (see Algorithm 1 lines (2)ndash(18))Specifically the energy of root is assumed to be infinite so itsdegree does not need to be balanced For a sensor node if itsparent node is not root and there exists a neighbor node withthe minimum number of children and at the same height asits parent in its own SPT (to effectively balance theminimum
number of childrenmust be less than original value at least 2)this neighbor node will be the new parent of the node Afterupdating the new parent the old parent and the new parentnode will renew the numbers of children and inform all theirown neighbor nodes
The next task of EEBRHM is to determine the LNs Sincethe LNs are more close to the data sink and the length ofmoving path119880 is shorter than other sensor nodes the sensornodes which is more close to the data sink will more likelyto be LNs To determine LNs firstly all sensor nodes areassigned to one of three separate large time slots accordingto the distance between sensor nodes and the data sink Theformula (119879
119899+IDlowast1199051015840) 119899 = 0 1 2 ensures that each nodewill be
assigned to a separate small time slot so that the sensor nodeswill not interfere with each other in the process of algorithmMoreover 119879
119899is measured in seconds and 1199051015840 is measured in
milliseconds A node which is more close to the data sinkwill be assigned to an earlier large time slot and would havepriority to be a LN At last TSP algorithm is used to constructan approximate shortest tour 119880 visiting V
0and all the LNs
42 Performance Analysis To evaluate the performance ofEEBRHM firstly we will analyze the time complexity ofEEBRHM Assume a WSN has 119870 disconnected subnetworksand 119873 is the total number of sensor nodes in the WSN (1 ⩽119870 ⩽ 119873) For a subnetwork 119896 (119896 = 1 2 119870) it would take119874((119873119896)
2) time to find a SPT [27] where 119873
119896represents the
number of sensor nodes in subnetwork 119896 In our algorithm itruns in119874(119873) time to balance the nodesrsquo degree and constructLNs And it runs in 119874(1198732) time to find tour 119880 at most byusing TSP algorithm [28] Thus the total time complexity ofEEBRHM is119874((119873
119896)
2) +119874(119873)+119874(119873
2) In the worst case the
time complexity of EEBRHM is 119874(1198732)Due to balancing degree of sensor nodes the length
of moving path 119880 of EEBRHM may be longer than BRH-MDG algorithm which will increase a bit latency of networkby EEBRHM But the performance of network lifetime ofEEBRHM is much better than BRH-MDG
EEBRHM is also one of mobile data gathering algorithmsand the data transmission between nodes and mobile sink isthe main part of communication overhead The mechanism
International Journal of Distributed Sensor Networks 5
Algorithm EEBRHMInput Network topology 119866(119881 119864) the relay hop bound 119889
and the static data sink V0
Output A set of LNs a set of sub-trees and the tour 119880visiting the LNs and the data sink
(1) Construct SPTs for 119866 that cover all the vertices in 119881(2) for each node V do (3) if119867(V) ⩾ 2 (4) cn = CN(Parent (V)) V119905 = V(5) for each node V1015840 in Nb(V)
(6) if119867(V1015840) minus 119867(V) == minus1 (7) if CN(V1015840) lt 119888119899 (8) cn = CN(V1015840) V119905 = V1015840(9)
(10)
(11)
(12) if cn + 2 lt CN(Parent (V)) (13) Parent(V) ⩽ vt(14) Broadcast ACKMessage(15) When received the ACKMessage
the related nodes will make response(16)
(17)
(18) (19) Set three large time slots 119879
0 1198791 1198792 1198790lt 1198791lt 1198792
and set a constant 1199051015840 1199051015840 ≪ 1198790
(20) for each node V do (21) Get the unique integer ID of V(22) if V in one hop of V
0
(23) Allocate a delay time slot (1198790+ ID lowast 1199051015840) to V
(24)
(25) if V is out of one hop of V0but it is in two hop
(26) Allocate a delay time slot (1198791+ ID lowast 1199051015840) to V
(27)
(28) if V is out of two hop of V0
(29) Allocate a delay time slot (1198792+ ID lowast 1199051015840) to V
(30)
(31) (32) In the delay time slot of each node V (33) if V is not belong to a LN (34) Let it be a LN and broadcast a probe packet
with 119889 hops of time to live(35)
(36) if other nodes received the probe packet (37) if it is not a LN (38) if it is the child of relay node (39) Let it be an affiliated node of the source
node of the probe packet(40)
(41)
(42)
(43) (44) Use the TSP algorithm to find an approximate shortest
tour 119880 visiting V0and all the LNs
Algorithm 1 EEBRHM algorithm
6 International Journal of Distributed Sensor Networks
Table 1 Notations used in EEBRHM algorithm
Functions119867(V) This function returns the height of V in its own SPTParent(V) This function returns the parent node of VCN(V) This function returns the number of children nodes of VNb(V) This function returns a set of neighbor nodes of V
Variablescn The variable is used to cache the number of children of node VV119905 The variable is used to cache the node V which is relevant to cn
0
1000
2000
3000
4000
5000
6000
7000
8000
100 200 300 400 500
Net
wor
k lif
etim
e (ro
unds
)
Node number
EEBRHMBRH-MDG
(a) Network lifetime comparison
500
600
700
800
900
1000
1100
1200
1300
100 200 300 400 500
Tour
leng
th (m
)
Node number
EEBRHMBRH-MDG
CMESHDG
(b) Length of tour comparison
Figure 2 Performance comparison with changing the density of sensor nodes
of constructing the path of mobile sink of EEBRHM isthe same as BRH-MDG Thus the communication cost ofEEBRHM is just the same as BRH-MDG
5 Simulation Results
In the simulation we assume that 119873 sensor nodes arerandomly distributed over an119872times119872 square area and the datasink is set in the center of the area Furthermore we compareour algorithm with BRH-MDG [16] SHDG [21] and CME[19] SHDG is one kind of one hop mobile data gatheringscheme This algorithm will choose some pausing locationsfor mobile collector from a set of candidate locations so thatmobile collector can gather sensor data fromnodes in a singlehop Although we can use TSP algorithm to construct anapproximate shortest tour 119880 for mobile collector to visit allpausing locations and the data sink the latency of networkof SHDG is too long CME generates some parallel straighttours in advance which across the WSN Then the mobilecollector will traverse along these tours while gathering datafrom the sensor nodes nearby the tours And the other sensornodes would send their own sensor data to the nodes nearby
tours by multihop relays The latency of network of CMEis acceptable but due to randomly choosing a relay nodethe network lifetime achieved by CME may be poor Thedescription of BRH-MDGhas beenmentioned in Section 31All simulations are performed 30 times and average values oftheir results are taken as final results
51 The Impact of Sensor Node Density To compare theperformance of different algorithms we change the densityof sensor nodes The simulation setting in the experiments isthat communication radius 119877 bounded-hop 119889 and the areaare set as 30m 2 and 200 lowast 200m2 respectively Figure 2(a)plots the performance of EEBRHM and BRH-MDG as afunction of density of sensor nodes in terms of networklifetime From Figure 2(a) we can find that when the numberof sensor nodes becomes larger the advantage of EEBRHMover BRH-MDG become more obvious The performance ofnetwork lifetime of EEBRHM is 47 times than BRH-MDGon average There are two reasons to explain this First thedegrees of sensor nodes in BRH-MDGare very unevenwhichleads to uneven node energy consumption and thereforemake the network lifetime short Our EEBRHM algorithm
International Journal of Distributed Sensor Networks 7
Net
wor
k lif
etim
e (ro
unds
)
0
2000
4000
6000
8000
10000
20 30 40 50 60Communication radius (m)
EEBRHM d = 2
BRH-MDG d = 2
EEBRHM d = 3
BRH-MDG d = 3
(a) Network lifetime comparison
0
300
600
900
1200
1500
20 30 40 50 60
Tour
leng
th (m
)
Communication radius (m)
EEBRHM d = 2
BRH-MDG d = 2
EEBRHM d = 3
BRH-MDG d = 3
(b) Length of tour comparison
Figure 3 Performance comparison with changing the bounded 119889
can balance the degree of each node thus its performance ofnetwork lifetime is much better than BRH-MDGThe secondreason is that since BRH-MDGdoes not balance the degree ofeach node when the density of sensor nodes becomes largerthe degree of nodes in the WSN may increase very quicklytherefore the network lifetime of BRH-MDG would becomeshorter and shorter
Figure 2(b) shows the performance of four algorithms asa function of density of sensor nodes in terms of tour lengthThe simulation setting is the same as above For SHDG BRH-MDG and EEBRHM TSP algorithm is used to generate anapproximate shortest tour and mobile collector will movealong it For CME the parallel straight tours traversing thesensor field are 100m apart from each other One of thesetours must go through the center of the fieldThemobile col-lector can change onto other tours bymoving along the sensorfield border Both SHDG and CME algorithms are imple-mented in a centralized fashion BRH-MDG and EEBRHMare implemented in a distributed fashion In Figure 2(b) itis easy to observe that the tour length of BRH-MDG andEEBRHMgradually increases at first and then stabilizes whenthe density of nodes becomes sufficiently large The reasonis that when sensor nodes become more densely dispersedthey will have higher probability to be affiliated with a LNwhich is close to the data sink EEBRHM will generate moreLNs than BRH-MDG by balancing the degree of each nodewhichmakes the average tour length of EEBRHM longer thanBRH-MDG In contrast the average tour length of SHDG islonger than BRH-MDG and EEBRHM Because SHDGmustvisit one hop range of each node its pausing locations ofmobile collector will become much more than BRH-MDGand EEBRHMwith the continuously increased density Sincethe mobile collector goes along the fixed tours in the sensorfield the tour length of CME is a constant which in generalis bigger than the average tour length of BRH-MDG andEEBRHM
52 The Impact of Communication Radius To evaluate theperformance of different algorithms with changing the com-munication radius the comparison is shown in Figure 3 Thesimulation setting of the experiments is that the number ofnodes 119873 and the area are fixed at 400 and 200 lowast 200m2respectively And the bounded-hop 119889 is set as 2 and 3 respec-tively Figure 3(a) illustrates the performance of EEBRHMand BRH-MDG as a function of communication radius 119877in terms of network lifetime As shown in Figure 3(a) thenetwork lifetime of EEBRHM quit outperforms that of BRH-MDG with 119889 = 2 or 3 and it becomes more obvious when 119877increases When 119877 is 60m the network lifetime of EEBRHMis 10 times longer than that of BRH-MDG The reason hasbeen explained in Section 51
In Figure 3(b) the performance of EEBRHM and BRH-MDG as a function of communication radius 119877 in terms oftour length is compared From Figure 3(b) we can know thatthe tour length of EEBRHM and BRH-MDG first graduallyreduces as 119877 increases and then stabilizes when 119877 becomessufficiently long For the same reason as above the averagetour length of EEBRHM in this case is longer than BRH-MDG Furthermore for BRH-MDG as 119889 increases the LNscan ownmore subordinate nodeswhichwillmake the averagetour length of BRH-MDG became shorter When the 119877 is60m and 119889 is 2 the tour length of BRH-MDG is near 0
6 Conclusion
In this paper we have studied the energy consumptionmodelof the WSN with mobile data gathering with bounded-hop LNs A new novel algorithm EEBRHM is proposedto optimize the network lifetime of WSN Extensive sim-ulations have been carried out to prove the efficiency ofthe protocol The results demonstrate that the proposedalgorithm can greatly prolong the network lifetime of theWSN with bounded relay hop and obtain about 73 times
8 International Journal of Distributed Sensor Networks
improvement on the network lifetime compared with BRH-MDG Moreover the tour length of EEBRHM is shorter thanSHDG and CME and is only longer than BRH-MDG by 35
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgments
This work is sponsored by the National Natural ScienceFoundation of China under Grant nos 61173169 and 61103203and the State Key Program of National Natural Science ofChina under Grant no 61232001F02
References
[1] Z Rezaei and S Mobininejad ldquoEnergy saving in wirelesssensor networksrdquo International Journal of Computer Science ampEngineering Survey vol 3 no 1 pp 23ndash37 2012
[2] W R Heinzelman A Chandrakasan and H Balakrish-nan ldquoEnergy-efficient communication protocol for wirelessmicrosensor networksrdquo in Proceedings of the 33rd AnnualHawaii International Conference on System Siences (HICSS rsquo33)pp 3005ndash3014 IEEE Computer SocietyWashington DC USAJanuary 2000
[3] O Younis and S Fahmy ldquoHEED a hybrid energy-efficientdistributed clustering approach for ad hoc sensor networksrdquoIEEE Transactions on Mobile Computing vol 3 no 4 pp 366ndash379 2004
[4] X Y Kui Y Sheng H K Du and J B Liang ldquoConstructinga CDS-based network backbone for data collection in wirelesssensor networksrdquo International Journal of Distributed SensorNetworks vol 2013 Article ID 258081 12 pages 2013
[5] X Ma J Gao W Wang and J Wang ldquoA virtual-ring-baseddata storage and retrieval scheme in wireless sensor networksrdquoInternational Journal of Distributed Sensor Networks vol 2012Article ID 763015 10 pages 2012
[6] S Lindsey and C S Raghavendra ldquoPEGASIS power-efficientgathering in sensor information systemsrdquo in Proceedings of theIEEE Aerospace Conference IEEE Computer Society pp 1125ndash1130 San Francisco Calif USA March 2002
[7] S-M Jung Y-J Han and T-M Chung ldquoThe concentricclustering scheme for efficient energy consumption in thePEGASISrdquo in Proceedings of the 9th International Conferenceon Advanced Communication Technology (ICACT rsquo07) pp 260ndash265 IEEE Computer Society Phoenix Park February 2007
[8] G J Wang T Wang W J Jia M Y Guo and J Li ldquoAdaptivelocation updates for mobile sinks in wireless sensor networksrdquoJournal of Supercomputing vol 47 no 2 pp 127ndash145 2009
[9] Q Zhang Z-P Xie B Ling W-W Sun and B-L Shi ldquoMax-imum lifetime data gathering algorithm for wireless sensornetworksrdquo Journal of Software vol 16 no 11 pp 1946ndash1957 2005(Chinese)
[10] W Liang and Y Liu ldquoOnline data gathering for maximizingnetwork lifetime in sensor networksrdquo IEEE Transactions onMobile Computing vol 6 no 1 pp 2ndash11 2007
[11] Z F Liao J X Wang S G Zhang J N Cao and G YMin ldquoMinimizing movement for target coverage and network
connectivity in mobile sensor networksrdquo IEEE Transactions onParallel and Distributed Systems 2014
[12] Y Wu S Fahmy and N B Shroff ldquoOn the construction of amaximum-lifetime data gathering tree in sensor networks NP-completeness and approximation algorithmrdquo in Proceedings ofthe 27th IEEE Communications Society Conference on ComputerCommunications (INFOCOM rsquo08) IEEE Computer Society pp356ndash360 Phoenix Ariz USA April 2008
[13] J-B Liang J-X Wang and J-E Chen ldquoOn the construction ofa delay-constrained maximum lifetime tree in wireless sensornetworksrdquo Acta Electronica Sinica vol 38 no 2 pp 345ndash3512010
[14] K Karenos and V Kalogeraki ldquoTraffic management in sensornetworks with amobile sinkrdquo IEEE Transactions on Parallel andDistributed Systems vol 21 no 10 pp 1515ndash1530 2010
[15] X Xu J Luo and Q Zhang ldquoDelay tolerant event collection insensor networks with mobile sinkrdquo in Proceedings of the IEEEINFOCOM March 2010
[16] MZhao andYYang ldquoBounded relay hopmobile data gatheringin wireless sensor networksrdquo IEEE Transactions on Computersvol 61 no 2 pp 265ndash277 2012
[17] R C Shah S Roy S Jain and W Brunette ldquoData MULEsmodeling and analysis of a three-tier architecture for sparsesensor networksrdquo Ad Hoc Networks vol 1 no 2-3 pp 215ndash2332003
[18] S Jain R C Shah W Brunette G Borriello and S RoyldquoExploiting mobility for energy efficient data collection inwireless sensor networksrdquo Mobile Networks and Applicationsvol 11 no 3 pp 327ndash339 2006
[19] D Jea A Somasundara andM Srivastava ldquoMultiple controlledmobile elements (data mules) for data collection in sensor net-worksrdquo in Proceedings of the 1st IEEE International Conferenceon Distributed Computing in Sensor Systems (DCOSS rsquo05) pp244ndash257 July 2005
[20] A A Somasundara A Ramamoorthy and M B SrivastavaldquoMobile element scheduling with dynamic deadlinesrdquo IEEETransactions on Mobile Computing vol 6 no 4 pp 395ndash4102007
[21] MMa and Y Yang ldquoData gathering in wireless sensor networkswith mobile collectorsrdquo in Proceedings of the 22nd IEEE Inter-national Parallel and Distributed Processing Symposium (IPDPSrsquo08) April 2008
[22] M Ma and Y Yang ldquoSenCar an energy-efficient data gatheringmechanism for large-scale multihop sensor networksrdquo IEEETransactions on Parallel and Distributed Systems vol 18 no 10pp 1476ndash1488 2007
[23] B Kusy H Lee M Wicke N Milosavljevic and L GuibasldquoPredictive QoS routing to mobile sinks in wireless sensornetworksrdquo in Proceedings of the International Conference onInformation Processing in Sensor Networks (IPSN rsquo09) pp 109ndash120 April 2009
[24] J Luo and J-P Hubaux ldquoJoint mobility and routing for lifetimeelongation in wireless sensor networksrdquo in Proceedings of theIEEE 24th Annual Joint Conference of the IEEE Computerand Communications Societies (INFOCOM rsquo05) pp 1735ndash1746March 2005
[25] M Zhao M Ma and Y Yang ldquoEfficient data gathering withmobile collectors and space-division multiple access techniquein wireless sensor networksrdquo IEEE Transactions on Computersvol 60 no 3 pp 400ndash417 2011
[26] S M Lasassmeh and J M Conrad ldquoTime synchronization inwireless sensor networks a surveyrdquo in Proceedings of the IEEE
International Journal of Distributed Sensor Networks 9
Conference Energizing Our Future (SoutheastCon rsquo10) pp 242ndash245 Concord NC USA March 2010
[27] T H Cormen C E Leiserson R L Rivest and C SteinIntroduction to Algorithms The MIT Press Cambridge MassUSA 2nd edition 2001
[28] A Chowdhury A Ghosh S Sinha and S Das ldquoA novel geneticalgorithm to solve travelling salesman problem and blockingflow shop scheduling problemrdquo International Journal of Bio-Inspired Computation vol 5 no 5 pp 303ndash314 2013
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
(a) An example of BRH-MDG algorithm (b) After balancing the degree of each node
Figure 1 An example to show the effect of balancing the degree of each node
this problem a general idea is to balance degrees of nodesin subtrees such as BRH-MDG algorithm As shown inFigure 1(a) the BRH-MDG algorithm divides the networkinto three subtrees Since the process of generating subtreesis random the distribution of nodesrsquo degrees is not uniformIn Figure 1(a) the dotted lines represent the moving paththe black nodes represent the nodes which have maximumdegrees in subtrees and the maximum degree among theseblack nodes is 5 By applying a method to balance the degreesof nodes in Figure 1(a) a new set of subtrees can be obtainedas shown in Figure 1(b) and the maximum degree is reducedto 2 According to Definitions 3 and 4 the lifetime of networkin Figure 1(b) is twice more than that in Figure 1(a)
The example demonstrates that the key point to optimizethe energy consumption of the whole WSN is to balance thedistribution of nodesrsquo degree in their own subtrees In thispaper we propose a new algorithm called EEBRHM whichuses the number of children of nodes in the initial tree ascontrol metric to adjust the degree of each node The detailof EEBRHM algorithm is shown in Algorithm 1 and Table 1
First of all EEBRHM algorithm will construct shortestpath trees (SPTs) which should cover all sensor nodes withminimum hops in the WSN (see Algorithm 1 line (1))Because a large WSN may not be fully connected it maybe constituted by several subnetworks and thus it needsmore than one SPT to cover all sensor nodes In EEBRHMalgorithm generally the root of SPT is the data sink For thenodes from other subnetworks which are far from the datasink we can set virtual roots in the center of the subnetworksThus the new algorithm can be applied to not only connectedWSNs but also disconnectedWSNs which is one of themainadvantages for the mobile data gathering Moreover beforeall sensor nodes being deployed each node will be assigneda unique integer ID which will be used to allocate separatedelay time slot for each sensor nodeThen EEBRHMwill usethe number of sensor nodesrsquo children as metric to balancethe degrees of sensor nodes (see Algorithm 1 lines (2)ndash(18))Specifically the energy of root is assumed to be infinite so itsdegree does not need to be balanced For a sensor node if itsparent node is not root and there exists a neighbor node withthe minimum number of children and at the same height asits parent in its own SPT (to effectively balance theminimum
number of childrenmust be less than original value at least 2)this neighbor node will be the new parent of the node Afterupdating the new parent the old parent and the new parentnode will renew the numbers of children and inform all theirown neighbor nodes
The next task of EEBRHM is to determine the LNs Sincethe LNs are more close to the data sink and the length ofmoving path119880 is shorter than other sensor nodes the sensornodes which is more close to the data sink will more likelyto be LNs To determine LNs firstly all sensor nodes areassigned to one of three separate large time slots accordingto the distance between sensor nodes and the data sink Theformula (119879
119899+IDlowast1199051015840) 119899 = 0 1 2 ensures that each nodewill be
assigned to a separate small time slot so that the sensor nodeswill not interfere with each other in the process of algorithmMoreover 119879
119899is measured in seconds and 1199051015840 is measured in
milliseconds A node which is more close to the data sinkwill be assigned to an earlier large time slot and would havepriority to be a LN At last TSP algorithm is used to constructan approximate shortest tour 119880 visiting V
0and all the LNs
42 Performance Analysis To evaluate the performance ofEEBRHM firstly we will analyze the time complexity ofEEBRHM Assume a WSN has 119870 disconnected subnetworksand 119873 is the total number of sensor nodes in the WSN (1 ⩽119870 ⩽ 119873) For a subnetwork 119896 (119896 = 1 2 119870) it would take119874((119873119896)
2) time to find a SPT [27] where 119873
119896represents the
number of sensor nodes in subnetwork 119896 In our algorithm itruns in119874(119873) time to balance the nodesrsquo degree and constructLNs And it runs in 119874(1198732) time to find tour 119880 at most byusing TSP algorithm [28] Thus the total time complexity ofEEBRHM is119874((119873
119896)
2) +119874(119873)+119874(119873
2) In the worst case the
time complexity of EEBRHM is 119874(1198732)Due to balancing degree of sensor nodes the length
of moving path 119880 of EEBRHM may be longer than BRH-MDG algorithm which will increase a bit latency of networkby EEBRHM But the performance of network lifetime ofEEBRHM is much better than BRH-MDG
EEBRHM is also one of mobile data gathering algorithmsand the data transmission between nodes and mobile sink isthe main part of communication overhead The mechanism
International Journal of Distributed Sensor Networks 5
Algorithm EEBRHMInput Network topology 119866(119881 119864) the relay hop bound 119889
and the static data sink V0
Output A set of LNs a set of sub-trees and the tour 119880visiting the LNs and the data sink
(1) Construct SPTs for 119866 that cover all the vertices in 119881(2) for each node V do (3) if119867(V) ⩾ 2 (4) cn = CN(Parent (V)) V119905 = V(5) for each node V1015840 in Nb(V)
(6) if119867(V1015840) minus 119867(V) == minus1 (7) if CN(V1015840) lt 119888119899 (8) cn = CN(V1015840) V119905 = V1015840(9)
(10)
(11)
(12) if cn + 2 lt CN(Parent (V)) (13) Parent(V) ⩽ vt(14) Broadcast ACKMessage(15) When received the ACKMessage
the related nodes will make response(16)
(17)
(18) (19) Set three large time slots 119879
0 1198791 1198792 1198790lt 1198791lt 1198792
and set a constant 1199051015840 1199051015840 ≪ 1198790
(20) for each node V do (21) Get the unique integer ID of V(22) if V in one hop of V
0
(23) Allocate a delay time slot (1198790+ ID lowast 1199051015840) to V
(24)
(25) if V is out of one hop of V0but it is in two hop
(26) Allocate a delay time slot (1198791+ ID lowast 1199051015840) to V
(27)
(28) if V is out of two hop of V0
(29) Allocate a delay time slot (1198792+ ID lowast 1199051015840) to V
(30)
(31) (32) In the delay time slot of each node V (33) if V is not belong to a LN (34) Let it be a LN and broadcast a probe packet
with 119889 hops of time to live(35)
(36) if other nodes received the probe packet (37) if it is not a LN (38) if it is the child of relay node (39) Let it be an affiliated node of the source
node of the probe packet(40)
(41)
(42)
(43) (44) Use the TSP algorithm to find an approximate shortest
tour 119880 visiting V0and all the LNs
Algorithm 1 EEBRHM algorithm
6 International Journal of Distributed Sensor Networks
Table 1 Notations used in EEBRHM algorithm
Functions119867(V) This function returns the height of V in its own SPTParent(V) This function returns the parent node of VCN(V) This function returns the number of children nodes of VNb(V) This function returns a set of neighbor nodes of V
Variablescn The variable is used to cache the number of children of node VV119905 The variable is used to cache the node V which is relevant to cn
0
1000
2000
3000
4000
5000
6000
7000
8000
100 200 300 400 500
Net
wor
k lif
etim
e (ro
unds
)
Node number
EEBRHMBRH-MDG
(a) Network lifetime comparison
500
600
700
800
900
1000
1100
1200
1300
100 200 300 400 500
Tour
leng
th (m
)
Node number
EEBRHMBRH-MDG
CMESHDG
(b) Length of tour comparison
Figure 2 Performance comparison with changing the density of sensor nodes
of constructing the path of mobile sink of EEBRHM isthe same as BRH-MDG Thus the communication cost ofEEBRHM is just the same as BRH-MDG
5 Simulation Results
In the simulation we assume that 119873 sensor nodes arerandomly distributed over an119872times119872 square area and the datasink is set in the center of the area Furthermore we compareour algorithm with BRH-MDG [16] SHDG [21] and CME[19] SHDG is one kind of one hop mobile data gatheringscheme This algorithm will choose some pausing locationsfor mobile collector from a set of candidate locations so thatmobile collector can gather sensor data fromnodes in a singlehop Although we can use TSP algorithm to construct anapproximate shortest tour 119880 for mobile collector to visit allpausing locations and the data sink the latency of networkof SHDG is too long CME generates some parallel straighttours in advance which across the WSN Then the mobilecollector will traverse along these tours while gathering datafrom the sensor nodes nearby the tours And the other sensornodes would send their own sensor data to the nodes nearby
tours by multihop relays The latency of network of CMEis acceptable but due to randomly choosing a relay nodethe network lifetime achieved by CME may be poor Thedescription of BRH-MDGhas beenmentioned in Section 31All simulations are performed 30 times and average values oftheir results are taken as final results
51 The Impact of Sensor Node Density To compare theperformance of different algorithms we change the densityof sensor nodes The simulation setting in the experiments isthat communication radius 119877 bounded-hop 119889 and the areaare set as 30m 2 and 200 lowast 200m2 respectively Figure 2(a)plots the performance of EEBRHM and BRH-MDG as afunction of density of sensor nodes in terms of networklifetime From Figure 2(a) we can find that when the numberof sensor nodes becomes larger the advantage of EEBRHMover BRH-MDG become more obvious The performance ofnetwork lifetime of EEBRHM is 47 times than BRH-MDGon average There are two reasons to explain this First thedegrees of sensor nodes in BRH-MDGare very unevenwhichleads to uneven node energy consumption and thereforemake the network lifetime short Our EEBRHM algorithm
International Journal of Distributed Sensor Networks 7
Net
wor
k lif
etim
e (ro
unds
)
0
2000
4000
6000
8000
10000
20 30 40 50 60Communication radius (m)
EEBRHM d = 2
BRH-MDG d = 2
EEBRHM d = 3
BRH-MDG d = 3
(a) Network lifetime comparison
0
300
600
900
1200
1500
20 30 40 50 60
Tour
leng
th (m
)
Communication radius (m)
EEBRHM d = 2
BRH-MDG d = 2
EEBRHM d = 3
BRH-MDG d = 3
(b) Length of tour comparison
Figure 3 Performance comparison with changing the bounded 119889
can balance the degree of each node thus its performance ofnetwork lifetime is much better than BRH-MDGThe secondreason is that since BRH-MDGdoes not balance the degree ofeach node when the density of sensor nodes becomes largerthe degree of nodes in the WSN may increase very quicklytherefore the network lifetime of BRH-MDG would becomeshorter and shorter
Figure 2(b) shows the performance of four algorithms asa function of density of sensor nodes in terms of tour lengthThe simulation setting is the same as above For SHDG BRH-MDG and EEBRHM TSP algorithm is used to generate anapproximate shortest tour and mobile collector will movealong it For CME the parallel straight tours traversing thesensor field are 100m apart from each other One of thesetours must go through the center of the fieldThemobile col-lector can change onto other tours bymoving along the sensorfield border Both SHDG and CME algorithms are imple-mented in a centralized fashion BRH-MDG and EEBRHMare implemented in a distributed fashion In Figure 2(b) itis easy to observe that the tour length of BRH-MDG andEEBRHMgradually increases at first and then stabilizes whenthe density of nodes becomes sufficiently large The reasonis that when sensor nodes become more densely dispersedthey will have higher probability to be affiliated with a LNwhich is close to the data sink EEBRHM will generate moreLNs than BRH-MDG by balancing the degree of each nodewhichmakes the average tour length of EEBRHM longer thanBRH-MDG In contrast the average tour length of SHDG islonger than BRH-MDG and EEBRHM Because SHDGmustvisit one hop range of each node its pausing locations ofmobile collector will become much more than BRH-MDGand EEBRHMwith the continuously increased density Sincethe mobile collector goes along the fixed tours in the sensorfield the tour length of CME is a constant which in generalis bigger than the average tour length of BRH-MDG andEEBRHM
52 The Impact of Communication Radius To evaluate theperformance of different algorithms with changing the com-munication radius the comparison is shown in Figure 3 Thesimulation setting of the experiments is that the number ofnodes 119873 and the area are fixed at 400 and 200 lowast 200m2respectively And the bounded-hop 119889 is set as 2 and 3 respec-tively Figure 3(a) illustrates the performance of EEBRHMand BRH-MDG as a function of communication radius 119877in terms of network lifetime As shown in Figure 3(a) thenetwork lifetime of EEBRHM quit outperforms that of BRH-MDG with 119889 = 2 or 3 and it becomes more obvious when 119877increases When 119877 is 60m the network lifetime of EEBRHMis 10 times longer than that of BRH-MDG The reason hasbeen explained in Section 51
In Figure 3(b) the performance of EEBRHM and BRH-MDG as a function of communication radius 119877 in terms oftour length is compared From Figure 3(b) we can know thatthe tour length of EEBRHM and BRH-MDG first graduallyreduces as 119877 increases and then stabilizes when 119877 becomessufficiently long For the same reason as above the averagetour length of EEBRHM in this case is longer than BRH-MDG Furthermore for BRH-MDG as 119889 increases the LNscan ownmore subordinate nodeswhichwillmake the averagetour length of BRH-MDG became shorter When the 119877 is60m and 119889 is 2 the tour length of BRH-MDG is near 0
6 Conclusion
In this paper we have studied the energy consumptionmodelof the WSN with mobile data gathering with bounded-hop LNs A new novel algorithm EEBRHM is proposedto optimize the network lifetime of WSN Extensive sim-ulations have been carried out to prove the efficiency ofthe protocol The results demonstrate that the proposedalgorithm can greatly prolong the network lifetime of theWSN with bounded relay hop and obtain about 73 times
8 International Journal of Distributed Sensor Networks
improvement on the network lifetime compared with BRH-MDG Moreover the tour length of EEBRHM is shorter thanSHDG and CME and is only longer than BRH-MDG by 35
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgments
This work is sponsored by the National Natural ScienceFoundation of China under Grant nos 61173169 and 61103203and the State Key Program of National Natural Science ofChina under Grant no 61232001F02
References
[1] Z Rezaei and S Mobininejad ldquoEnergy saving in wirelesssensor networksrdquo International Journal of Computer Science ampEngineering Survey vol 3 no 1 pp 23ndash37 2012
[2] W R Heinzelman A Chandrakasan and H Balakrish-nan ldquoEnergy-efficient communication protocol for wirelessmicrosensor networksrdquo in Proceedings of the 33rd AnnualHawaii International Conference on System Siences (HICSS rsquo33)pp 3005ndash3014 IEEE Computer SocietyWashington DC USAJanuary 2000
[3] O Younis and S Fahmy ldquoHEED a hybrid energy-efficientdistributed clustering approach for ad hoc sensor networksrdquoIEEE Transactions on Mobile Computing vol 3 no 4 pp 366ndash379 2004
[4] X Y Kui Y Sheng H K Du and J B Liang ldquoConstructinga CDS-based network backbone for data collection in wirelesssensor networksrdquo International Journal of Distributed SensorNetworks vol 2013 Article ID 258081 12 pages 2013
[5] X Ma J Gao W Wang and J Wang ldquoA virtual-ring-baseddata storage and retrieval scheme in wireless sensor networksrdquoInternational Journal of Distributed Sensor Networks vol 2012Article ID 763015 10 pages 2012
[6] S Lindsey and C S Raghavendra ldquoPEGASIS power-efficientgathering in sensor information systemsrdquo in Proceedings of theIEEE Aerospace Conference IEEE Computer Society pp 1125ndash1130 San Francisco Calif USA March 2002
[7] S-M Jung Y-J Han and T-M Chung ldquoThe concentricclustering scheme for efficient energy consumption in thePEGASISrdquo in Proceedings of the 9th International Conferenceon Advanced Communication Technology (ICACT rsquo07) pp 260ndash265 IEEE Computer Society Phoenix Park February 2007
[8] G J Wang T Wang W J Jia M Y Guo and J Li ldquoAdaptivelocation updates for mobile sinks in wireless sensor networksrdquoJournal of Supercomputing vol 47 no 2 pp 127ndash145 2009
[9] Q Zhang Z-P Xie B Ling W-W Sun and B-L Shi ldquoMax-imum lifetime data gathering algorithm for wireless sensornetworksrdquo Journal of Software vol 16 no 11 pp 1946ndash1957 2005(Chinese)
[10] W Liang and Y Liu ldquoOnline data gathering for maximizingnetwork lifetime in sensor networksrdquo IEEE Transactions onMobile Computing vol 6 no 1 pp 2ndash11 2007
[11] Z F Liao J X Wang S G Zhang J N Cao and G YMin ldquoMinimizing movement for target coverage and network
connectivity in mobile sensor networksrdquo IEEE Transactions onParallel and Distributed Systems 2014
[12] Y Wu S Fahmy and N B Shroff ldquoOn the construction of amaximum-lifetime data gathering tree in sensor networks NP-completeness and approximation algorithmrdquo in Proceedings ofthe 27th IEEE Communications Society Conference on ComputerCommunications (INFOCOM rsquo08) IEEE Computer Society pp356ndash360 Phoenix Ariz USA April 2008
[13] J-B Liang J-X Wang and J-E Chen ldquoOn the construction ofa delay-constrained maximum lifetime tree in wireless sensornetworksrdquo Acta Electronica Sinica vol 38 no 2 pp 345ndash3512010
[14] K Karenos and V Kalogeraki ldquoTraffic management in sensornetworks with amobile sinkrdquo IEEE Transactions on Parallel andDistributed Systems vol 21 no 10 pp 1515ndash1530 2010
[15] X Xu J Luo and Q Zhang ldquoDelay tolerant event collection insensor networks with mobile sinkrdquo in Proceedings of the IEEEINFOCOM March 2010
[16] MZhao andYYang ldquoBounded relay hopmobile data gatheringin wireless sensor networksrdquo IEEE Transactions on Computersvol 61 no 2 pp 265ndash277 2012
[17] R C Shah S Roy S Jain and W Brunette ldquoData MULEsmodeling and analysis of a three-tier architecture for sparsesensor networksrdquo Ad Hoc Networks vol 1 no 2-3 pp 215ndash2332003
[18] S Jain R C Shah W Brunette G Borriello and S RoyldquoExploiting mobility for energy efficient data collection inwireless sensor networksrdquo Mobile Networks and Applicationsvol 11 no 3 pp 327ndash339 2006
[19] D Jea A Somasundara andM Srivastava ldquoMultiple controlledmobile elements (data mules) for data collection in sensor net-worksrdquo in Proceedings of the 1st IEEE International Conferenceon Distributed Computing in Sensor Systems (DCOSS rsquo05) pp244ndash257 July 2005
[20] A A Somasundara A Ramamoorthy and M B SrivastavaldquoMobile element scheduling with dynamic deadlinesrdquo IEEETransactions on Mobile Computing vol 6 no 4 pp 395ndash4102007
[21] MMa and Y Yang ldquoData gathering in wireless sensor networkswith mobile collectorsrdquo in Proceedings of the 22nd IEEE Inter-national Parallel and Distributed Processing Symposium (IPDPSrsquo08) April 2008
[22] M Ma and Y Yang ldquoSenCar an energy-efficient data gatheringmechanism for large-scale multihop sensor networksrdquo IEEETransactions on Parallel and Distributed Systems vol 18 no 10pp 1476ndash1488 2007
[23] B Kusy H Lee M Wicke N Milosavljevic and L GuibasldquoPredictive QoS routing to mobile sinks in wireless sensornetworksrdquo in Proceedings of the International Conference onInformation Processing in Sensor Networks (IPSN rsquo09) pp 109ndash120 April 2009
[24] J Luo and J-P Hubaux ldquoJoint mobility and routing for lifetimeelongation in wireless sensor networksrdquo in Proceedings of theIEEE 24th Annual Joint Conference of the IEEE Computerand Communications Societies (INFOCOM rsquo05) pp 1735ndash1746March 2005
[25] M Zhao M Ma and Y Yang ldquoEfficient data gathering withmobile collectors and space-division multiple access techniquein wireless sensor networksrdquo IEEE Transactions on Computersvol 60 no 3 pp 400ndash417 2011
[26] S M Lasassmeh and J M Conrad ldquoTime synchronization inwireless sensor networks a surveyrdquo in Proceedings of the IEEE
International Journal of Distributed Sensor Networks 9
Conference Energizing Our Future (SoutheastCon rsquo10) pp 242ndash245 Concord NC USA March 2010
[27] T H Cormen C E Leiserson R L Rivest and C SteinIntroduction to Algorithms The MIT Press Cambridge MassUSA 2nd edition 2001
[28] A Chowdhury A Ghosh S Sinha and S Das ldquoA novel geneticalgorithm to solve travelling salesman problem and blockingflow shop scheduling problemrdquo International Journal of Bio-Inspired Computation vol 5 no 5 pp 303ndash314 2013
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
Algorithm EEBRHMInput Network topology 119866(119881 119864) the relay hop bound 119889
and the static data sink V0
Output A set of LNs a set of sub-trees and the tour 119880visiting the LNs and the data sink
(1) Construct SPTs for 119866 that cover all the vertices in 119881(2) for each node V do (3) if119867(V) ⩾ 2 (4) cn = CN(Parent (V)) V119905 = V(5) for each node V1015840 in Nb(V)
(6) if119867(V1015840) minus 119867(V) == minus1 (7) if CN(V1015840) lt 119888119899 (8) cn = CN(V1015840) V119905 = V1015840(9)
(10)
(11)
(12) if cn + 2 lt CN(Parent (V)) (13) Parent(V) ⩽ vt(14) Broadcast ACKMessage(15) When received the ACKMessage
the related nodes will make response(16)
(17)
(18) (19) Set three large time slots 119879
0 1198791 1198792 1198790lt 1198791lt 1198792
and set a constant 1199051015840 1199051015840 ≪ 1198790
(20) for each node V do (21) Get the unique integer ID of V(22) if V in one hop of V
0
(23) Allocate a delay time slot (1198790+ ID lowast 1199051015840) to V
(24)
(25) if V is out of one hop of V0but it is in two hop
(26) Allocate a delay time slot (1198791+ ID lowast 1199051015840) to V
(27)
(28) if V is out of two hop of V0
(29) Allocate a delay time slot (1198792+ ID lowast 1199051015840) to V
(30)
(31) (32) In the delay time slot of each node V (33) if V is not belong to a LN (34) Let it be a LN and broadcast a probe packet
with 119889 hops of time to live(35)
(36) if other nodes received the probe packet (37) if it is not a LN (38) if it is the child of relay node (39) Let it be an affiliated node of the source
node of the probe packet(40)
(41)
(42)
(43) (44) Use the TSP algorithm to find an approximate shortest
tour 119880 visiting V0and all the LNs
Algorithm 1 EEBRHM algorithm
6 International Journal of Distributed Sensor Networks
Table 1 Notations used in EEBRHM algorithm
Functions119867(V) This function returns the height of V in its own SPTParent(V) This function returns the parent node of VCN(V) This function returns the number of children nodes of VNb(V) This function returns a set of neighbor nodes of V
Variablescn The variable is used to cache the number of children of node VV119905 The variable is used to cache the node V which is relevant to cn
0
1000
2000
3000
4000
5000
6000
7000
8000
100 200 300 400 500
Net
wor
k lif
etim
e (ro
unds
)
Node number
EEBRHMBRH-MDG
(a) Network lifetime comparison
500
600
700
800
900
1000
1100
1200
1300
100 200 300 400 500
Tour
leng
th (m
)
Node number
EEBRHMBRH-MDG
CMESHDG
(b) Length of tour comparison
Figure 2 Performance comparison with changing the density of sensor nodes
of constructing the path of mobile sink of EEBRHM isthe same as BRH-MDG Thus the communication cost ofEEBRHM is just the same as BRH-MDG
5 Simulation Results
In the simulation we assume that 119873 sensor nodes arerandomly distributed over an119872times119872 square area and the datasink is set in the center of the area Furthermore we compareour algorithm with BRH-MDG [16] SHDG [21] and CME[19] SHDG is one kind of one hop mobile data gatheringscheme This algorithm will choose some pausing locationsfor mobile collector from a set of candidate locations so thatmobile collector can gather sensor data fromnodes in a singlehop Although we can use TSP algorithm to construct anapproximate shortest tour 119880 for mobile collector to visit allpausing locations and the data sink the latency of networkof SHDG is too long CME generates some parallel straighttours in advance which across the WSN Then the mobilecollector will traverse along these tours while gathering datafrom the sensor nodes nearby the tours And the other sensornodes would send their own sensor data to the nodes nearby
tours by multihop relays The latency of network of CMEis acceptable but due to randomly choosing a relay nodethe network lifetime achieved by CME may be poor Thedescription of BRH-MDGhas beenmentioned in Section 31All simulations are performed 30 times and average values oftheir results are taken as final results
51 The Impact of Sensor Node Density To compare theperformance of different algorithms we change the densityof sensor nodes The simulation setting in the experiments isthat communication radius 119877 bounded-hop 119889 and the areaare set as 30m 2 and 200 lowast 200m2 respectively Figure 2(a)plots the performance of EEBRHM and BRH-MDG as afunction of density of sensor nodes in terms of networklifetime From Figure 2(a) we can find that when the numberof sensor nodes becomes larger the advantage of EEBRHMover BRH-MDG become more obvious The performance ofnetwork lifetime of EEBRHM is 47 times than BRH-MDGon average There are two reasons to explain this First thedegrees of sensor nodes in BRH-MDGare very unevenwhichleads to uneven node energy consumption and thereforemake the network lifetime short Our EEBRHM algorithm
International Journal of Distributed Sensor Networks 7
Net
wor
k lif
etim
e (ro
unds
)
0
2000
4000
6000
8000
10000
20 30 40 50 60Communication radius (m)
EEBRHM d = 2
BRH-MDG d = 2
EEBRHM d = 3
BRH-MDG d = 3
(a) Network lifetime comparison
0
300
600
900
1200
1500
20 30 40 50 60
Tour
leng
th (m
)
Communication radius (m)
EEBRHM d = 2
BRH-MDG d = 2
EEBRHM d = 3
BRH-MDG d = 3
(b) Length of tour comparison
Figure 3 Performance comparison with changing the bounded 119889
can balance the degree of each node thus its performance ofnetwork lifetime is much better than BRH-MDGThe secondreason is that since BRH-MDGdoes not balance the degree ofeach node when the density of sensor nodes becomes largerthe degree of nodes in the WSN may increase very quicklytherefore the network lifetime of BRH-MDG would becomeshorter and shorter
Figure 2(b) shows the performance of four algorithms asa function of density of sensor nodes in terms of tour lengthThe simulation setting is the same as above For SHDG BRH-MDG and EEBRHM TSP algorithm is used to generate anapproximate shortest tour and mobile collector will movealong it For CME the parallel straight tours traversing thesensor field are 100m apart from each other One of thesetours must go through the center of the fieldThemobile col-lector can change onto other tours bymoving along the sensorfield border Both SHDG and CME algorithms are imple-mented in a centralized fashion BRH-MDG and EEBRHMare implemented in a distributed fashion In Figure 2(b) itis easy to observe that the tour length of BRH-MDG andEEBRHMgradually increases at first and then stabilizes whenthe density of nodes becomes sufficiently large The reasonis that when sensor nodes become more densely dispersedthey will have higher probability to be affiliated with a LNwhich is close to the data sink EEBRHM will generate moreLNs than BRH-MDG by balancing the degree of each nodewhichmakes the average tour length of EEBRHM longer thanBRH-MDG In contrast the average tour length of SHDG islonger than BRH-MDG and EEBRHM Because SHDGmustvisit one hop range of each node its pausing locations ofmobile collector will become much more than BRH-MDGand EEBRHMwith the continuously increased density Sincethe mobile collector goes along the fixed tours in the sensorfield the tour length of CME is a constant which in generalis bigger than the average tour length of BRH-MDG andEEBRHM
52 The Impact of Communication Radius To evaluate theperformance of different algorithms with changing the com-munication radius the comparison is shown in Figure 3 Thesimulation setting of the experiments is that the number ofnodes 119873 and the area are fixed at 400 and 200 lowast 200m2respectively And the bounded-hop 119889 is set as 2 and 3 respec-tively Figure 3(a) illustrates the performance of EEBRHMand BRH-MDG as a function of communication radius 119877in terms of network lifetime As shown in Figure 3(a) thenetwork lifetime of EEBRHM quit outperforms that of BRH-MDG with 119889 = 2 or 3 and it becomes more obvious when 119877increases When 119877 is 60m the network lifetime of EEBRHMis 10 times longer than that of BRH-MDG The reason hasbeen explained in Section 51
In Figure 3(b) the performance of EEBRHM and BRH-MDG as a function of communication radius 119877 in terms oftour length is compared From Figure 3(b) we can know thatthe tour length of EEBRHM and BRH-MDG first graduallyreduces as 119877 increases and then stabilizes when 119877 becomessufficiently long For the same reason as above the averagetour length of EEBRHM in this case is longer than BRH-MDG Furthermore for BRH-MDG as 119889 increases the LNscan ownmore subordinate nodeswhichwillmake the averagetour length of BRH-MDG became shorter When the 119877 is60m and 119889 is 2 the tour length of BRH-MDG is near 0
6 Conclusion
In this paper we have studied the energy consumptionmodelof the WSN with mobile data gathering with bounded-hop LNs A new novel algorithm EEBRHM is proposedto optimize the network lifetime of WSN Extensive sim-ulations have been carried out to prove the efficiency ofthe protocol The results demonstrate that the proposedalgorithm can greatly prolong the network lifetime of theWSN with bounded relay hop and obtain about 73 times
8 International Journal of Distributed Sensor Networks
improvement on the network lifetime compared with BRH-MDG Moreover the tour length of EEBRHM is shorter thanSHDG and CME and is only longer than BRH-MDG by 35
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgments
This work is sponsored by the National Natural ScienceFoundation of China under Grant nos 61173169 and 61103203and the State Key Program of National Natural Science ofChina under Grant no 61232001F02
References
[1] Z Rezaei and S Mobininejad ldquoEnergy saving in wirelesssensor networksrdquo International Journal of Computer Science ampEngineering Survey vol 3 no 1 pp 23ndash37 2012
[2] W R Heinzelman A Chandrakasan and H Balakrish-nan ldquoEnergy-efficient communication protocol for wirelessmicrosensor networksrdquo in Proceedings of the 33rd AnnualHawaii International Conference on System Siences (HICSS rsquo33)pp 3005ndash3014 IEEE Computer SocietyWashington DC USAJanuary 2000
[3] O Younis and S Fahmy ldquoHEED a hybrid energy-efficientdistributed clustering approach for ad hoc sensor networksrdquoIEEE Transactions on Mobile Computing vol 3 no 4 pp 366ndash379 2004
[4] X Y Kui Y Sheng H K Du and J B Liang ldquoConstructinga CDS-based network backbone for data collection in wirelesssensor networksrdquo International Journal of Distributed SensorNetworks vol 2013 Article ID 258081 12 pages 2013
[5] X Ma J Gao W Wang and J Wang ldquoA virtual-ring-baseddata storage and retrieval scheme in wireless sensor networksrdquoInternational Journal of Distributed Sensor Networks vol 2012Article ID 763015 10 pages 2012
[6] S Lindsey and C S Raghavendra ldquoPEGASIS power-efficientgathering in sensor information systemsrdquo in Proceedings of theIEEE Aerospace Conference IEEE Computer Society pp 1125ndash1130 San Francisco Calif USA March 2002
[7] S-M Jung Y-J Han and T-M Chung ldquoThe concentricclustering scheme for efficient energy consumption in thePEGASISrdquo in Proceedings of the 9th International Conferenceon Advanced Communication Technology (ICACT rsquo07) pp 260ndash265 IEEE Computer Society Phoenix Park February 2007
[8] G J Wang T Wang W J Jia M Y Guo and J Li ldquoAdaptivelocation updates for mobile sinks in wireless sensor networksrdquoJournal of Supercomputing vol 47 no 2 pp 127ndash145 2009
[9] Q Zhang Z-P Xie B Ling W-W Sun and B-L Shi ldquoMax-imum lifetime data gathering algorithm for wireless sensornetworksrdquo Journal of Software vol 16 no 11 pp 1946ndash1957 2005(Chinese)
[10] W Liang and Y Liu ldquoOnline data gathering for maximizingnetwork lifetime in sensor networksrdquo IEEE Transactions onMobile Computing vol 6 no 1 pp 2ndash11 2007
[11] Z F Liao J X Wang S G Zhang J N Cao and G YMin ldquoMinimizing movement for target coverage and network
connectivity in mobile sensor networksrdquo IEEE Transactions onParallel and Distributed Systems 2014
[12] Y Wu S Fahmy and N B Shroff ldquoOn the construction of amaximum-lifetime data gathering tree in sensor networks NP-completeness and approximation algorithmrdquo in Proceedings ofthe 27th IEEE Communications Society Conference on ComputerCommunications (INFOCOM rsquo08) IEEE Computer Society pp356ndash360 Phoenix Ariz USA April 2008
[13] J-B Liang J-X Wang and J-E Chen ldquoOn the construction ofa delay-constrained maximum lifetime tree in wireless sensornetworksrdquo Acta Electronica Sinica vol 38 no 2 pp 345ndash3512010
[14] K Karenos and V Kalogeraki ldquoTraffic management in sensornetworks with amobile sinkrdquo IEEE Transactions on Parallel andDistributed Systems vol 21 no 10 pp 1515ndash1530 2010
[15] X Xu J Luo and Q Zhang ldquoDelay tolerant event collection insensor networks with mobile sinkrdquo in Proceedings of the IEEEINFOCOM March 2010
[16] MZhao andYYang ldquoBounded relay hopmobile data gatheringin wireless sensor networksrdquo IEEE Transactions on Computersvol 61 no 2 pp 265ndash277 2012
[17] R C Shah S Roy S Jain and W Brunette ldquoData MULEsmodeling and analysis of a three-tier architecture for sparsesensor networksrdquo Ad Hoc Networks vol 1 no 2-3 pp 215ndash2332003
[18] S Jain R C Shah W Brunette G Borriello and S RoyldquoExploiting mobility for energy efficient data collection inwireless sensor networksrdquo Mobile Networks and Applicationsvol 11 no 3 pp 327ndash339 2006
[19] D Jea A Somasundara andM Srivastava ldquoMultiple controlledmobile elements (data mules) for data collection in sensor net-worksrdquo in Proceedings of the 1st IEEE International Conferenceon Distributed Computing in Sensor Systems (DCOSS rsquo05) pp244ndash257 July 2005
[20] A A Somasundara A Ramamoorthy and M B SrivastavaldquoMobile element scheduling with dynamic deadlinesrdquo IEEETransactions on Mobile Computing vol 6 no 4 pp 395ndash4102007
[21] MMa and Y Yang ldquoData gathering in wireless sensor networkswith mobile collectorsrdquo in Proceedings of the 22nd IEEE Inter-national Parallel and Distributed Processing Symposium (IPDPSrsquo08) April 2008
[22] M Ma and Y Yang ldquoSenCar an energy-efficient data gatheringmechanism for large-scale multihop sensor networksrdquo IEEETransactions on Parallel and Distributed Systems vol 18 no 10pp 1476ndash1488 2007
[23] B Kusy H Lee M Wicke N Milosavljevic and L GuibasldquoPredictive QoS routing to mobile sinks in wireless sensornetworksrdquo in Proceedings of the International Conference onInformation Processing in Sensor Networks (IPSN rsquo09) pp 109ndash120 April 2009
[24] J Luo and J-P Hubaux ldquoJoint mobility and routing for lifetimeelongation in wireless sensor networksrdquo in Proceedings of theIEEE 24th Annual Joint Conference of the IEEE Computerand Communications Societies (INFOCOM rsquo05) pp 1735ndash1746March 2005
[25] M Zhao M Ma and Y Yang ldquoEfficient data gathering withmobile collectors and space-division multiple access techniquein wireless sensor networksrdquo IEEE Transactions on Computersvol 60 no 3 pp 400ndash417 2011
[26] S M Lasassmeh and J M Conrad ldquoTime synchronization inwireless sensor networks a surveyrdquo in Proceedings of the IEEE
International Journal of Distributed Sensor Networks 9
Conference Energizing Our Future (SoutheastCon rsquo10) pp 242ndash245 Concord NC USA March 2010
[27] T H Cormen C E Leiserson R L Rivest and C SteinIntroduction to Algorithms The MIT Press Cambridge MassUSA 2nd edition 2001
[28] A Chowdhury A Ghosh S Sinha and S Das ldquoA novel geneticalgorithm to solve travelling salesman problem and blockingflow shop scheduling problemrdquo International Journal of Bio-Inspired Computation vol 5 no 5 pp 303ndash314 2013
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
Table 1 Notations used in EEBRHM algorithm
Functions119867(V) This function returns the height of V in its own SPTParent(V) This function returns the parent node of VCN(V) This function returns the number of children nodes of VNb(V) This function returns a set of neighbor nodes of V
Variablescn The variable is used to cache the number of children of node VV119905 The variable is used to cache the node V which is relevant to cn
0
1000
2000
3000
4000
5000
6000
7000
8000
100 200 300 400 500
Net
wor
k lif
etim
e (ro
unds
)
Node number
EEBRHMBRH-MDG
(a) Network lifetime comparison
500
600
700
800
900
1000
1100
1200
1300
100 200 300 400 500
Tour
leng
th (m
)
Node number
EEBRHMBRH-MDG
CMESHDG
(b) Length of tour comparison
Figure 2 Performance comparison with changing the density of sensor nodes
of constructing the path of mobile sink of EEBRHM isthe same as BRH-MDG Thus the communication cost ofEEBRHM is just the same as BRH-MDG
5 Simulation Results
In the simulation we assume that 119873 sensor nodes arerandomly distributed over an119872times119872 square area and the datasink is set in the center of the area Furthermore we compareour algorithm with BRH-MDG [16] SHDG [21] and CME[19] SHDG is one kind of one hop mobile data gatheringscheme This algorithm will choose some pausing locationsfor mobile collector from a set of candidate locations so thatmobile collector can gather sensor data fromnodes in a singlehop Although we can use TSP algorithm to construct anapproximate shortest tour 119880 for mobile collector to visit allpausing locations and the data sink the latency of networkof SHDG is too long CME generates some parallel straighttours in advance which across the WSN Then the mobilecollector will traverse along these tours while gathering datafrom the sensor nodes nearby the tours And the other sensornodes would send their own sensor data to the nodes nearby
tours by multihop relays The latency of network of CMEis acceptable but due to randomly choosing a relay nodethe network lifetime achieved by CME may be poor Thedescription of BRH-MDGhas beenmentioned in Section 31All simulations are performed 30 times and average values oftheir results are taken as final results
51 The Impact of Sensor Node Density To compare theperformance of different algorithms we change the densityof sensor nodes The simulation setting in the experiments isthat communication radius 119877 bounded-hop 119889 and the areaare set as 30m 2 and 200 lowast 200m2 respectively Figure 2(a)plots the performance of EEBRHM and BRH-MDG as afunction of density of sensor nodes in terms of networklifetime From Figure 2(a) we can find that when the numberof sensor nodes becomes larger the advantage of EEBRHMover BRH-MDG become more obvious The performance ofnetwork lifetime of EEBRHM is 47 times than BRH-MDGon average There are two reasons to explain this First thedegrees of sensor nodes in BRH-MDGare very unevenwhichleads to uneven node energy consumption and thereforemake the network lifetime short Our EEBRHM algorithm
International Journal of Distributed Sensor Networks 7
Net
wor
k lif
etim
e (ro
unds
)
0
2000
4000
6000
8000
10000
20 30 40 50 60Communication radius (m)
EEBRHM d = 2
BRH-MDG d = 2
EEBRHM d = 3
BRH-MDG d = 3
(a) Network lifetime comparison
0
300
600
900
1200
1500
20 30 40 50 60
Tour
leng
th (m
)
Communication radius (m)
EEBRHM d = 2
BRH-MDG d = 2
EEBRHM d = 3
BRH-MDG d = 3
(b) Length of tour comparison
Figure 3 Performance comparison with changing the bounded 119889
can balance the degree of each node thus its performance ofnetwork lifetime is much better than BRH-MDGThe secondreason is that since BRH-MDGdoes not balance the degree ofeach node when the density of sensor nodes becomes largerthe degree of nodes in the WSN may increase very quicklytherefore the network lifetime of BRH-MDG would becomeshorter and shorter
Figure 2(b) shows the performance of four algorithms asa function of density of sensor nodes in terms of tour lengthThe simulation setting is the same as above For SHDG BRH-MDG and EEBRHM TSP algorithm is used to generate anapproximate shortest tour and mobile collector will movealong it For CME the parallel straight tours traversing thesensor field are 100m apart from each other One of thesetours must go through the center of the fieldThemobile col-lector can change onto other tours bymoving along the sensorfield border Both SHDG and CME algorithms are imple-mented in a centralized fashion BRH-MDG and EEBRHMare implemented in a distributed fashion In Figure 2(b) itis easy to observe that the tour length of BRH-MDG andEEBRHMgradually increases at first and then stabilizes whenthe density of nodes becomes sufficiently large The reasonis that when sensor nodes become more densely dispersedthey will have higher probability to be affiliated with a LNwhich is close to the data sink EEBRHM will generate moreLNs than BRH-MDG by balancing the degree of each nodewhichmakes the average tour length of EEBRHM longer thanBRH-MDG In contrast the average tour length of SHDG islonger than BRH-MDG and EEBRHM Because SHDGmustvisit one hop range of each node its pausing locations ofmobile collector will become much more than BRH-MDGand EEBRHMwith the continuously increased density Sincethe mobile collector goes along the fixed tours in the sensorfield the tour length of CME is a constant which in generalis bigger than the average tour length of BRH-MDG andEEBRHM
52 The Impact of Communication Radius To evaluate theperformance of different algorithms with changing the com-munication radius the comparison is shown in Figure 3 Thesimulation setting of the experiments is that the number ofnodes 119873 and the area are fixed at 400 and 200 lowast 200m2respectively And the bounded-hop 119889 is set as 2 and 3 respec-tively Figure 3(a) illustrates the performance of EEBRHMand BRH-MDG as a function of communication radius 119877in terms of network lifetime As shown in Figure 3(a) thenetwork lifetime of EEBRHM quit outperforms that of BRH-MDG with 119889 = 2 or 3 and it becomes more obvious when 119877increases When 119877 is 60m the network lifetime of EEBRHMis 10 times longer than that of BRH-MDG The reason hasbeen explained in Section 51
In Figure 3(b) the performance of EEBRHM and BRH-MDG as a function of communication radius 119877 in terms oftour length is compared From Figure 3(b) we can know thatthe tour length of EEBRHM and BRH-MDG first graduallyreduces as 119877 increases and then stabilizes when 119877 becomessufficiently long For the same reason as above the averagetour length of EEBRHM in this case is longer than BRH-MDG Furthermore for BRH-MDG as 119889 increases the LNscan ownmore subordinate nodeswhichwillmake the averagetour length of BRH-MDG became shorter When the 119877 is60m and 119889 is 2 the tour length of BRH-MDG is near 0
6 Conclusion
In this paper we have studied the energy consumptionmodelof the WSN with mobile data gathering with bounded-hop LNs A new novel algorithm EEBRHM is proposedto optimize the network lifetime of WSN Extensive sim-ulations have been carried out to prove the efficiency ofthe protocol The results demonstrate that the proposedalgorithm can greatly prolong the network lifetime of theWSN with bounded relay hop and obtain about 73 times
8 International Journal of Distributed Sensor Networks
improvement on the network lifetime compared with BRH-MDG Moreover the tour length of EEBRHM is shorter thanSHDG and CME and is only longer than BRH-MDG by 35
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgments
This work is sponsored by the National Natural ScienceFoundation of China under Grant nos 61173169 and 61103203and the State Key Program of National Natural Science ofChina under Grant no 61232001F02
References
[1] Z Rezaei and S Mobininejad ldquoEnergy saving in wirelesssensor networksrdquo International Journal of Computer Science ampEngineering Survey vol 3 no 1 pp 23ndash37 2012
[2] W R Heinzelman A Chandrakasan and H Balakrish-nan ldquoEnergy-efficient communication protocol for wirelessmicrosensor networksrdquo in Proceedings of the 33rd AnnualHawaii International Conference on System Siences (HICSS rsquo33)pp 3005ndash3014 IEEE Computer SocietyWashington DC USAJanuary 2000
[3] O Younis and S Fahmy ldquoHEED a hybrid energy-efficientdistributed clustering approach for ad hoc sensor networksrdquoIEEE Transactions on Mobile Computing vol 3 no 4 pp 366ndash379 2004
[4] X Y Kui Y Sheng H K Du and J B Liang ldquoConstructinga CDS-based network backbone for data collection in wirelesssensor networksrdquo International Journal of Distributed SensorNetworks vol 2013 Article ID 258081 12 pages 2013
[5] X Ma J Gao W Wang and J Wang ldquoA virtual-ring-baseddata storage and retrieval scheme in wireless sensor networksrdquoInternational Journal of Distributed Sensor Networks vol 2012Article ID 763015 10 pages 2012
[6] S Lindsey and C S Raghavendra ldquoPEGASIS power-efficientgathering in sensor information systemsrdquo in Proceedings of theIEEE Aerospace Conference IEEE Computer Society pp 1125ndash1130 San Francisco Calif USA March 2002
[7] S-M Jung Y-J Han and T-M Chung ldquoThe concentricclustering scheme for efficient energy consumption in thePEGASISrdquo in Proceedings of the 9th International Conferenceon Advanced Communication Technology (ICACT rsquo07) pp 260ndash265 IEEE Computer Society Phoenix Park February 2007
[8] G J Wang T Wang W J Jia M Y Guo and J Li ldquoAdaptivelocation updates for mobile sinks in wireless sensor networksrdquoJournal of Supercomputing vol 47 no 2 pp 127ndash145 2009
[9] Q Zhang Z-P Xie B Ling W-W Sun and B-L Shi ldquoMax-imum lifetime data gathering algorithm for wireless sensornetworksrdquo Journal of Software vol 16 no 11 pp 1946ndash1957 2005(Chinese)
[10] W Liang and Y Liu ldquoOnline data gathering for maximizingnetwork lifetime in sensor networksrdquo IEEE Transactions onMobile Computing vol 6 no 1 pp 2ndash11 2007
[11] Z F Liao J X Wang S G Zhang J N Cao and G YMin ldquoMinimizing movement for target coverage and network
connectivity in mobile sensor networksrdquo IEEE Transactions onParallel and Distributed Systems 2014
[12] Y Wu S Fahmy and N B Shroff ldquoOn the construction of amaximum-lifetime data gathering tree in sensor networks NP-completeness and approximation algorithmrdquo in Proceedings ofthe 27th IEEE Communications Society Conference on ComputerCommunications (INFOCOM rsquo08) IEEE Computer Society pp356ndash360 Phoenix Ariz USA April 2008
[13] J-B Liang J-X Wang and J-E Chen ldquoOn the construction ofa delay-constrained maximum lifetime tree in wireless sensornetworksrdquo Acta Electronica Sinica vol 38 no 2 pp 345ndash3512010
[14] K Karenos and V Kalogeraki ldquoTraffic management in sensornetworks with amobile sinkrdquo IEEE Transactions on Parallel andDistributed Systems vol 21 no 10 pp 1515ndash1530 2010
[15] X Xu J Luo and Q Zhang ldquoDelay tolerant event collection insensor networks with mobile sinkrdquo in Proceedings of the IEEEINFOCOM March 2010
[16] MZhao andYYang ldquoBounded relay hopmobile data gatheringin wireless sensor networksrdquo IEEE Transactions on Computersvol 61 no 2 pp 265ndash277 2012
[17] R C Shah S Roy S Jain and W Brunette ldquoData MULEsmodeling and analysis of a three-tier architecture for sparsesensor networksrdquo Ad Hoc Networks vol 1 no 2-3 pp 215ndash2332003
[18] S Jain R C Shah W Brunette G Borriello and S RoyldquoExploiting mobility for energy efficient data collection inwireless sensor networksrdquo Mobile Networks and Applicationsvol 11 no 3 pp 327ndash339 2006
[19] D Jea A Somasundara andM Srivastava ldquoMultiple controlledmobile elements (data mules) for data collection in sensor net-worksrdquo in Proceedings of the 1st IEEE International Conferenceon Distributed Computing in Sensor Systems (DCOSS rsquo05) pp244ndash257 July 2005
[20] A A Somasundara A Ramamoorthy and M B SrivastavaldquoMobile element scheduling with dynamic deadlinesrdquo IEEETransactions on Mobile Computing vol 6 no 4 pp 395ndash4102007
[21] MMa and Y Yang ldquoData gathering in wireless sensor networkswith mobile collectorsrdquo in Proceedings of the 22nd IEEE Inter-national Parallel and Distributed Processing Symposium (IPDPSrsquo08) April 2008
[22] M Ma and Y Yang ldquoSenCar an energy-efficient data gatheringmechanism for large-scale multihop sensor networksrdquo IEEETransactions on Parallel and Distributed Systems vol 18 no 10pp 1476ndash1488 2007
[23] B Kusy H Lee M Wicke N Milosavljevic and L GuibasldquoPredictive QoS routing to mobile sinks in wireless sensornetworksrdquo in Proceedings of the International Conference onInformation Processing in Sensor Networks (IPSN rsquo09) pp 109ndash120 April 2009
[24] J Luo and J-P Hubaux ldquoJoint mobility and routing for lifetimeelongation in wireless sensor networksrdquo in Proceedings of theIEEE 24th Annual Joint Conference of the IEEE Computerand Communications Societies (INFOCOM rsquo05) pp 1735ndash1746March 2005
[25] M Zhao M Ma and Y Yang ldquoEfficient data gathering withmobile collectors and space-division multiple access techniquein wireless sensor networksrdquo IEEE Transactions on Computersvol 60 no 3 pp 400ndash417 2011
[26] S M Lasassmeh and J M Conrad ldquoTime synchronization inwireless sensor networks a surveyrdquo in Proceedings of the IEEE
International Journal of Distributed Sensor Networks 9
Conference Energizing Our Future (SoutheastCon rsquo10) pp 242ndash245 Concord NC USA March 2010
[27] T H Cormen C E Leiserson R L Rivest and C SteinIntroduction to Algorithms The MIT Press Cambridge MassUSA 2nd edition 2001
[28] A Chowdhury A Ghosh S Sinha and S Das ldquoA novel geneticalgorithm to solve travelling salesman problem and blockingflow shop scheduling problemrdquo International Journal of Bio-Inspired Computation vol 5 no 5 pp 303ndash314 2013
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
Net
wor
k lif
etim
e (ro
unds
)
0
2000
4000
6000
8000
10000
20 30 40 50 60Communication radius (m)
EEBRHM d = 2
BRH-MDG d = 2
EEBRHM d = 3
BRH-MDG d = 3
(a) Network lifetime comparison
0
300
600
900
1200
1500
20 30 40 50 60
Tour
leng
th (m
)
Communication radius (m)
EEBRHM d = 2
BRH-MDG d = 2
EEBRHM d = 3
BRH-MDG d = 3
(b) Length of tour comparison
Figure 3 Performance comparison with changing the bounded 119889
can balance the degree of each node thus its performance ofnetwork lifetime is much better than BRH-MDGThe secondreason is that since BRH-MDGdoes not balance the degree ofeach node when the density of sensor nodes becomes largerthe degree of nodes in the WSN may increase very quicklytherefore the network lifetime of BRH-MDG would becomeshorter and shorter
Figure 2(b) shows the performance of four algorithms asa function of density of sensor nodes in terms of tour lengthThe simulation setting is the same as above For SHDG BRH-MDG and EEBRHM TSP algorithm is used to generate anapproximate shortest tour and mobile collector will movealong it For CME the parallel straight tours traversing thesensor field are 100m apart from each other One of thesetours must go through the center of the fieldThemobile col-lector can change onto other tours bymoving along the sensorfield border Both SHDG and CME algorithms are imple-mented in a centralized fashion BRH-MDG and EEBRHMare implemented in a distributed fashion In Figure 2(b) itis easy to observe that the tour length of BRH-MDG andEEBRHMgradually increases at first and then stabilizes whenthe density of nodes becomes sufficiently large The reasonis that when sensor nodes become more densely dispersedthey will have higher probability to be affiliated with a LNwhich is close to the data sink EEBRHM will generate moreLNs than BRH-MDG by balancing the degree of each nodewhichmakes the average tour length of EEBRHM longer thanBRH-MDG In contrast the average tour length of SHDG islonger than BRH-MDG and EEBRHM Because SHDGmustvisit one hop range of each node its pausing locations ofmobile collector will become much more than BRH-MDGand EEBRHMwith the continuously increased density Sincethe mobile collector goes along the fixed tours in the sensorfield the tour length of CME is a constant which in generalis bigger than the average tour length of BRH-MDG andEEBRHM
52 The Impact of Communication Radius To evaluate theperformance of different algorithms with changing the com-munication radius the comparison is shown in Figure 3 Thesimulation setting of the experiments is that the number ofnodes 119873 and the area are fixed at 400 and 200 lowast 200m2respectively And the bounded-hop 119889 is set as 2 and 3 respec-tively Figure 3(a) illustrates the performance of EEBRHMand BRH-MDG as a function of communication radius 119877in terms of network lifetime As shown in Figure 3(a) thenetwork lifetime of EEBRHM quit outperforms that of BRH-MDG with 119889 = 2 or 3 and it becomes more obvious when 119877increases When 119877 is 60m the network lifetime of EEBRHMis 10 times longer than that of BRH-MDG The reason hasbeen explained in Section 51
In Figure 3(b) the performance of EEBRHM and BRH-MDG as a function of communication radius 119877 in terms oftour length is compared From Figure 3(b) we can know thatthe tour length of EEBRHM and BRH-MDG first graduallyreduces as 119877 increases and then stabilizes when 119877 becomessufficiently long For the same reason as above the averagetour length of EEBRHM in this case is longer than BRH-MDG Furthermore for BRH-MDG as 119889 increases the LNscan ownmore subordinate nodeswhichwillmake the averagetour length of BRH-MDG became shorter When the 119877 is60m and 119889 is 2 the tour length of BRH-MDG is near 0
6 Conclusion
In this paper we have studied the energy consumptionmodelof the WSN with mobile data gathering with bounded-hop LNs A new novel algorithm EEBRHM is proposedto optimize the network lifetime of WSN Extensive sim-ulations have been carried out to prove the efficiency ofthe protocol The results demonstrate that the proposedalgorithm can greatly prolong the network lifetime of theWSN with bounded relay hop and obtain about 73 times
8 International Journal of Distributed Sensor Networks
improvement on the network lifetime compared with BRH-MDG Moreover the tour length of EEBRHM is shorter thanSHDG and CME and is only longer than BRH-MDG by 35
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgments
This work is sponsored by the National Natural ScienceFoundation of China under Grant nos 61173169 and 61103203and the State Key Program of National Natural Science ofChina under Grant no 61232001F02
References
[1] Z Rezaei and S Mobininejad ldquoEnergy saving in wirelesssensor networksrdquo International Journal of Computer Science ampEngineering Survey vol 3 no 1 pp 23ndash37 2012
[2] W R Heinzelman A Chandrakasan and H Balakrish-nan ldquoEnergy-efficient communication protocol for wirelessmicrosensor networksrdquo in Proceedings of the 33rd AnnualHawaii International Conference on System Siences (HICSS rsquo33)pp 3005ndash3014 IEEE Computer SocietyWashington DC USAJanuary 2000
[3] O Younis and S Fahmy ldquoHEED a hybrid energy-efficientdistributed clustering approach for ad hoc sensor networksrdquoIEEE Transactions on Mobile Computing vol 3 no 4 pp 366ndash379 2004
[4] X Y Kui Y Sheng H K Du and J B Liang ldquoConstructinga CDS-based network backbone for data collection in wirelesssensor networksrdquo International Journal of Distributed SensorNetworks vol 2013 Article ID 258081 12 pages 2013
[5] X Ma J Gao W Wang and J Wang ldquoA virtual-ring-baseddata storage and retrieval scheme in wireless sensor networksrdquoInternational Journal of Distributed Sensor Networks vol 2012Article ID 763015 10 pages 2012
[6] S Lindsey and C S Raghavendra ldquoPEGASIS power-efficientgathering in sensor information systemsrdquo in Proceedings of theIEEE Aerospace Conference IEEE Computer Society pp 1125ndash1130 San Francisco Calif USA March 2002
[7] S-M Jung Y-J Han and T-M Chung ldquoThe concentricclustering scheme for efficient energy consumption in thePEGASISrdquo in Proceedings of the 9th International Conferenceon Advanced Communication Technology (ICACT rsquo07) pp 260ndash265 IEEE Computer Society Phoenix Park February 2007
[8] G J Wang T Wang W J Jia M Y Guo and J Li ldquoAdaptivelocation updates for mobile sinks in wireless sensor networksrdquoJournal of Supercomputing vol 47 no 2 pp 127ndash145 2009
[9] Q Zhang Z-P Xie B Ling W-W Sun and B-L Shi ldquoMax-imum lifetime data gathering algorithm for wireless sensornetworksrdquo Journal of Software vol 16 no 11 pp 1946ndash1957 2005(Chinese)
[10] W Liang and Y Liu ldquoOnline data gathering for maximizingnetwork lifetime in sensor networksrdquo IEEE Transactions onMobile Computing vol 6 no 1 pp 2ndash11 2007
[11] Z F Liao J X Wang S G Zhang J N Cao and G YMin ldquoMinimizing movement for target coverage and network
connectivity in mobile sensor networksrdquo IEEE Transactions onParallel and Distributed Systems 2014
[12] Y Wu S Fahmy and N B Shroff ldquoOn the construction of amaximum-lifetime data gathering tree in sensor networks NP-completeness and approximation algorithmrdquo in Proceedings ofthe 27th IEEE Communications Society Conference on ComputerCommunications (INFOCOM rsquo08) IEEE Computer Society pp356ndash360 Phoenix Ariz USA April 2008
[13] J-B Liang J-X Wang and J-E Chen ldquoOn the construction ofa delay-constrained maximum lifetime tree in wireless sensornetworksrdquo Acta Electronica Sinica vol 38 no 2 pp 345ndash3512010
[14] K Karenos and V Kalogeraki ldquoTraffic management in sensornetworks with amobile sinkrdquo IEEE Transactions on Parallel andDistributed Systems vol 21 no 10 pp 1515ndash1530 2010
[15] X Xu J Luo and Q Zhang ldquoDelay tolerant event collection insensor networks with mobile sinkrdquo in Proceedings of the IEEEINFOCOM March 2010
[16] MZhao andYYang ldquoBounded relay hopmobile data gatheringin wireless sensor networksrdquo IEEE Transactions on Computersvol 61 no 2 pp 265ndash277 2012
[17] R C Shah S Roy S Jain and W Brunette ldquoData MULEsmodeling and analysis of a three-tier architecture for sparsesensor networksrdquo Ad Hoc Networks vol 1 no 2-3 pp 215ndash2332003
[18] S Jain R C Shah W Brunette G Borriello and S RoyldquoExploiting mobility for energy efficient data collection inwireless sensor networksrdquo Mobile Networks and Applicationsvol 11 no 3 pp 327ndash339 2006
[19] D Jea A Somasundara andM Srivastava ldquoMultiple controlledmobile elements (data mules) for data collection in sensor net-worksrdquo in Proceedings of the 1st IEEE International Conferenceon Distributed Computing in Sensor Systems (DCOSS rsquo05) pp244ndash257 July 2005
[20] A A Somasundara A Ramamoorthy and M B SrivastavaldquoMobile element scheduling with dynamic deadlinesrdquo IEEETransactions on Mobile Computing vol 6 no 4 pp 395ndash4102007
[21] MMa and Y Yang ldquoData gathering in wireless sensor networkswith mobile collectorsrdquo in Proceedings of the 22nd IEEE Inter-national Parallel and Distributed Processing Symposium (IPDPSrsquo08) April 2008
[22] M Ma and Y Yang ldquoSenCar an energy-efficient data gatheringmechanism for large-scale multihop sensor networksrdquo IEEETransactions on Parallel and Distributed Systems vol 18 no 10pp 1476ndash1488 2007
[23] B Kusy H Lee M Wicke N Milosavljevic and L GuibasldquoPredictive QoS routing to mobile sinks in wireless sensornetworksrdquo in Proceedings of the International Conference onInformation Processing in Sensor Networks (IPSN rsquo09) pp 109ndash120 April 2009
[24] J Luo and J-P Hubaux ldquoJoint mobility and routing for lifetimeelongation in wireless sensor networksrdquo in Proceedings of theIEEE 24th Annual Joint Conference of the IEEE Computerand Communications Societies (INFOCOM rsquo05) pp 1735ndash1746March 2005
[25] M Zhao M Ma and Y Yang ldquoEfficient data gathering withmobile collectors and space-division multiple access techniquein wireless sensor networksrdquo IEEE Transactions on Computersvol 60 no 3 pp 400ndash417 2011
[26] S M Lasassmeh and J M Conrad ldquoTime synchronization inwireless sensor networks a surveyrdquo in Proceedings of the IEEE
International Journal of Distributed Sensor Networks 9
Conference Energizing Our Future (SoutheastCon rsquo10) pp 242ndash245 Concord NC USA March 2010
[27] T H Cormen C E Leiserson R L Rivest and C SteinIntroduction to Algorithms The MIT Press Cambridge MassUSA 2nd edition 2001
[28] A Chowdhury A Ghosh S Sinha and S Das ldquoA novel geneticalgorithm to solve travelling salesman problem and blockingflow shop scheduling problemrdquo International Journal of Bio-Inspired Computation vol 5 no 5 pp 303ndash314 2013
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
improvement on the network lifetime compared with BRH-MDG Moreover the tour length of EEBRHM is shorter thanSHDG and CME and is only longer than BRH-MDG by 35
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgments
This work is sponsored by the National Natural ScienceFoundation of China under Grant nos 61173169 and 61103203and the State Key Program of National Natural Science ofChina under Grant no 61232001F02
References
[1] Z Rezaei and S Mobininejad ldquoEnergy saving in wirelesssensor networksrdquo International Journal of Computer Science ampEngineering Survey vol 3 no 1 pp 23ndash37 2012
[2] W R Heinzelman A Chandrakasan and H Balakrish-nan ldquoEnergy-efficient communication protocol for wirelessmicrosensor networksrdquo in Proceedings of the 33rd AnnualHawaii International Conference on System Siences (HICSS rsquo33)pp 3005ndash3014 IEEE Computer SocietyWashington DC USAJanuary 2000
[3] O Younis and S Fahmy ldquoHEED a hybrid energy-efficientdistributed clustering approach for ad hoc sensor networksrdquoIEEE Transactions on Mobile Computing vol 3 no 4 pp 366ndash379 2004
[4] X Y Kui Y Sheng H K Du and J B Liang ldquoConstructinga CDS-based network backbone for data collection in wirelesssensor networksrdquo International Journal of Distributed SensorNetworks vol 2013 Article ID 258081 12 pages 2013
[5] X Ma J Gao W Wang and J Wang ldquoA virtual-ring-baseddata storage and retrieval scheme in wireless sensor networksrdquoInternational Journal of Distributed Sensor Networks vol 2012Article ID 763015 10 pages 2012
[6] S Lindsey and C S Raghavendra ldquoPEGASIS power-efficientgathering in sensor information systemsrdquo in Proceedings of theIEEE Aerospace Conference IEEE Computer Society pp 1125ndash1130 San Francisco Calif USA March 2002
[7] S-M Jung Y-J Han and T-M Chung ldquoThe concentricclustering scheme for efficient energy consumption in thePEGASISrdquo in Proceedings of the 9th International Conferenceon Advanced Communication Technology (ICACT rsquo07) pp 260ndash265 IEEE Computer Society Phoenix Park February 2007
[8] G J Wang T Wang W J Jia M Y Guo and J Li ldquoAdaptivelocation updates for mobile sinks in wireless sensor networksrdquoJournal of Supercomputing vol 47 no 2 pp 127ndash145 2009
[9] Q Zhang Z-P Xie B Ling W-W Sun and B-L Shi ldquoMax-imum lifetime data gathering algorithm for wireless sensornetworksrdquo Journal of Software vol 16 no 11 pp 1946ndash1957 2005(Chinese)
[10] W Liang and Y Liu ldquoOnline data gathering for maximizingnetwork lifetime in sensor networksrdquo IEEE Transactions onMobile Computing vol 6 no 1 pp 2ndash11 2007
[11] Z F Liao J X Wang S G Zhang J N Cao and G YMin ldquoMinimizing movement for target coverage and network
connectivity in mobile sensor networksrdquo IEEE Transactions onParallel and Distributed Systems 2014
[12] Y Wu S Fahmy and N B Shroff ldquoOn the construction of amaximum-lifetime data gathering tree in sensor networks NP-completeness and approximation algorithmrdquo in Proceedings ofthe 27th IEEE Communications Society Conference on ComputerCommunications (INFOCOM rsquo08) IEEE Computer Society pp356ndash360 Phoenix Ariz USA April 2008
[13] J-B Liang J-X Wang and J-E Chen ldquoOn the construction ofa delay-constrained maximum lifetime tree in wireless sensornetworksrdquo Acta Electronica Sinica vol 38 no 2 pp 345ndash3512010
[14] K Karenos and V Kalogeraki ldquoTraffic management in sensornetworks with amobile sinkrdquo IEEE Transactions on Parallel andDistributed Systems vol 21 no 10 pp 1515ndash1530 2010
[15] X Xu J Luo and Q Zhang ldquoDelay tolerant event collection insensor networks with mobile sinkrdquo in Proceedings of the IEEEINFOCOM March 2010
[16] MZhao andYYang ldquoBounded relay hopmobile data gatheringin wireless sensor networksrdquo IEEE Transactions on Computersvol 61 no 2 pp 265ndash277 2012
[17] R C Shah S Roy S Jain and W Brunette ldquoData MULEsmodeling and analysis of a three-tier architecture for sparsesensor networksrdquo Ad Hoc Networks vol 1 no 2-3 pp 215ndash2332003
[18] S Jain R C Shah W Brunette G Borriello and S RoyldquoExploiting mobility for energy efficient data collection inwireless sensor networksrdquo Mobile Networks and Applicationsvol 11 no 3 pp 327ndash339 2006
[19] D Jea A Somasundara andM Srivastava ldquoMultiple controlledmobile elements (data mules) for data collection in sensor net-worksrdquo in Proceedings of the 1st IEEE International Conferenceon Distributed Computing in Sensor Systems (DCOSS rsquo05) pp244ndash257 July 2005
[20] A A Somasundara A Ramamoorthy and M B SrivastavaldquoMobile element scheduling with dynamic deadlinesrdquo IEEETransactions on Mobile Computing vol 6 no 4 pp 395ndash4102007
[21] MMa and Y Yang ldquoData gathering in wireless sensor networkswith mobile collectorsrdquo in Proceedings of the 22nd IEEE Inter-national Parallel and Distributed Processing Symposium (IPDPSrsquo08) April 2008
[22] M Ma and Y Yang ldquoSenCar an energy-efficient data gatheringmechanism for large-scale multihop sensor networksrdquo IEEETransactions on Parallel and Distributed Systems vol 18 no 10pp 1476ndash1488 2007
[23] B Kusy H Lee M Wicke N Milosavljevic and L GuibasldquoPredictive QoS routing to mobile sinks in wireless sensornetworksrdquo in Proceedings of the International Conference onInformation Processing in Sensor Networks (IPSN rsquo09) pp 109ndash120 April 2009
[24] J Luo and J-P Hubaux ldquoJoint mobility and routing for lifetimeelongation in wireless sensor networksrdquo in Proceedings of theIEEE 24th Annual Joint Conference of the IEEE Computerand Communications Societies (INFOCOM rsquo05) pp 1735ndash1746March 2005
[25] M Zhao M Ma and Y Yang ldquoEfficient data gathering withmobile collectors and space-division multiple access techniquein wireless sensor networksrdquo IEEE Transactions on Computersvol 60 no 3 pp 400ndash417 2011
[26] S M Lasassmeh and J M Conrad ldquoTime synchronization inwireless sensor networks a surveyrdquo in Proceedings of the IEEE
International Journal of Distributed Sensor Networks 9
Conference Energizing Our Future (SoutheastCon rsquo10) pp 242ndash245 Concord NC USA March 2010
[27] T H Cormen C E Leiserson R L Rivest and C SteinIntroduction to Algorithms The MIT Press Cambridge MassUSA 2nd edition 2001
[28] A Chowdhury A Ghosh S Sinha and S Das ldquoA novel geneticalgorithm to solve travelling salesman problem and blockingflow shop scheduling problemrdquo International Journal of Bio-Inspired Computation vol 5 no 5 pp 303ndash314 2013
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
Conference Energizing Our Future (SoutheastCon rsquo10) pp 242ndash245 Concord NC USA March 2010
[27] T H Cormen C E Leiserson R L Rivest and C SteinIntroduction to Algorithms The MIT Press Cambridge MassUSA 2nd edition 2001
[28] A Chowdhury A Ghosh S Sinha and S Das ldquoA novel geneticalgorithm to solve travelling salesman problem and blockingflow shop scheduling problemrdquo International Journal of Bio-Inspired Computation vol 5 no 5 pp 303ndash314 2013
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