Design and Performance Analysis of Energy Aware Routing Protocol for Delay Sensitive Applications...

NC

SC

International Journal of Computer Networks and Communications Security

VOL. 1, NO. 2, JULY 2013, 46–53Available online at: www.ijcncs.orgISSN 2308-9830

Design and Performance Analysis of Energy Aware RoutingProtocol for Delay Sensitive Applications for Wireless Sensor

Networks

NANDINI PRASAD K S1, PUTTAMADAPPA C2

1Dr. Ambedkar Institute of Technology, Department of ISE, Bangalore2Sapthagiri College of Engineering, Department of ECE, Bangalore

E-mail: [email protected], [email protected]

ABSTRACT

Wireless Sensor Networks (WSNs) are emerging network technology with innumerable applications. Butsecurity and energy constraints reduce its successful deployments. The nodes in network are greatlyinvolved in transmissions and other processing operations for maintenance other than establishing orhandling a call. Due to limited processing ability, storage capacity and most importantly the availablebattery power of the nodes, it is required to minimize the transmission power and the amount of datatransmitted, for efficient operation. This paper presents a power aware routing protocol designed forwireless sensor networks. The proposed routing protocol is an extended and enhanced version of DynamicSource Routing protocol. It adds energy awareness to the existing implementation of DSR protocol. Energymetric is considered during route selection process to choose an optimal path in terms of overall energy ofthe nodes along the path, and “low energy notification” method is used during route maintenance process toincrease the lifetime of the 'bridge' nodes to avoid network partitioning. The performance of DSR protocoland Energy Aware DSR (EADSR) protocol are compared through NS2 simulation under differentscenarios. In all the cases, it is seen that EADSR protocol out-performs DSR protocol by energy saving inefficient manner.

Keywords: Transmission, DSR, Energy aware, routing protocol, WSNs.

1 INTRODUCTION

The deployment of specialized category of sensornetworks is increasing drastically to monitor thephysical environments in variety of delay sensitiveapplications such as military applications,agriculture, medical transport, industry etc. Themost important application of wireless sensornetwork is to monitor the critical condition whereend to end delay becomes bottleneck [1]. Themonitoring applications involve the sensing ofinformation during emergency state from thephysical environment where the network of sensoris deployed. During critical conditions likeexplosions, fire and leaking of toxic gases, there isa need of system which should be fast enough andmust respond within a fraction of seconds. Theseleads to a big challenge for researchers to develop afast, reliable and fault tolerant channeled sensor

networks during emergency and critical conditionsto base station that receives the events [2].

Wireless Sensor Network (WSN) [3] has beenregarded as special category of Ad hoc Networkthat may be used for a specific targeted application.Wireless Sensor Networks mainly consists ofthousands of low cost, small size and batterypowered sensor nodes which has more potentialsthan others Ad hoc networks to be deployed inmany emerging areas. There is a need of designinga robust, energy efficient, minimum delay routingprotocols to route the packets in these kinds ofnetworks [4]. Out of so many routing protocolsDynamic Source Routing protocol realized to bepromising one. A significant design challenge ispresented by Routing to meet the various designchallenges to cater the real time applicationrequirements in wireless sensor networks (WSNs)communication. In this paper, we design and

47

Nandini prasad K S, Puttamadappa C / International Journal of Computer Networks and Communications Security, 1 (2), JULY 2013

develop a modified Dynamic Source Routingprotocol, termed as Energy Aware DSR (EADSR)which aims to minimize delay on multipath routing,end to end delay and consumes very less energy toprovide enhance quality of Service for delayapplications in Wireless Sensor Networks. We haveconsidered main effecting factors such as end toend delay and unreliable of wireless channel linksto routes derive a routing metric used by the routingprotocol to determine the cost linked with separateroutes. The route with minimum end to end delayfor transmissions is selected. Although this schemecan somewhat increase the latency of the datatransfer but it results in a significant power savingand long lasting routes. It maintains the existingperformance of DSR and adds to it the powerawareness to increase the network lifetime. Themost significant benefit by adding energyawareness is that it avoids, for a long time,partitioning of network into non-communicatingdisjoint sets.

The paper aims at discovering an efficient poweraware routing scheme for ad hoc network with lessor no mobility. Simulation result shows that theproposed scheme EAR outperforms other existingrouting protocols in terms of different energyrelated parameters. At the time route selection,EAR take crucial thing like battery status of thepath into consideration while also givingimportance to delay. With this main factor inconsideration, EAR always select less congestedand more stable route for data delivery. In thispaper, an attempt has been made to evaluate theperformance of DSR routing protocol using somesimulation network models, to investigate how wellthis protocol performs on WSNs using NS-2simulator. The performance study will focus on theimpact of the network size, network density (up to500 nodes), and the number of sources (dataconnections). The performance metrics used inthis work are average end-to-end delay, packetdelivery fraction and average energy consumptionper delivered packet. Simulation results revealbetter performance for the modified DSR protocol,most importantly when used for routing inunreliable wireless channel link conditions withhigh packet error rates. The remaining sections areorganized as follows: In section 2, the related workis discussed. Section 3 explains the features ofDSR protocol. Section 4 covers the implementationpart followed for DSR and EADSR. Section 5 dealswith the simulation carried out in theimplementation. Section 6 describes the results andperformance analysis of DSR and EADSR. Lastsection 7, deals the conclusions and future work tobe carried out.

2 RELATED WORK

Power aware routing (PAR) protocols have beenproposed in response to the energy conservationrequirement at the network layer. The energy hererefers to the energy utilized by the nodes to transmitand receive packets, i.e. the energy utilized by thecommunication subsystem. The energy required fordata processing and other auxiliary processing isnot considered here because the processing energydepends on the mission of each node and is usuallynegligible when compared to the power required bythe communication interface. An early goal of PARwas to select the best path such that the total energyconsumed by the network is minimized. The basicapproach is to minimize the average energyconsumed per packet (as it traverses the network)or per unit flow. As in the case of minimum-hoprouting, one serious drawback of this approach isthat nodes will have a wide difference in energyconsumption. Nodes on the minimum energy pathswill quickly drain out while the other nodes remainintact. This results in an early death of some nodes.

In scenarios where the nodes need to workcollaboratively, another objective of PAR isproposed: maximize the time taken by the firstnode/sensor to fail because it runs out of batterypower. This time is known as the system lifetime.Coyle et al. [5] proposed a set of power-awaremetrics based on battery power consumption atnodes. These metrics can be easily incorporatedinto existing routing protocols. One of the metrics,the minimum cost/packet metric, aims to maximizethe lifetime of all nodes in the network. Theminimum-cost routing algorithms using this metricachieved significant reduction in cost/packet overminimum-hop routing. Chen et al. [6] proposed anenergy-conserving routing protocol to maximize thesystem lifetime by balancing the energyconsumption among the nodes in proportion to theirenergy reserves. Norouzi et al. [7] investigated theminimum energy and maximum lifetime issuestogether and revealed that the two goals are notcompatible. As a trade-off, he proposed aconditional max–min battery capacity routingscheme which chooses the shortest path if all nodesin all possible routes have sufficient batterycapacity. When the battery capacity of some nodesfall below a predefined threshold, routes goingthrough these nodes will be avoided, and thereforethe time until the first node power-down isextended. These proposed schemes embed theenergy awareness (each node is aware of itsexisting battery reserves) into the routing protocoland were proposed for a sensor network, where allthe nodes are treated identical in terms of available

48


resources and functioning roles. In addition, thoseschemes are more suitable for static networksbecause the benefits come from the evendistribution of traffic among different nodes. Whilethe nodes are moving independently, the savingsprovided by these algorithms, if any, is negligiblebecause of the difficulty of real-timereconfiguration.

3 FEATURES OF DYNAMIC SOURCEROUTING PROTOCOL

Dynamic source routing protocol(DSR) is an on-demand routing protocol designed to restrict thebandwidth consumed by control packets in ad hocwireless networks by eliminating the periodic table-update messages required in the table-drivenapproach [8]. The major difference between thisand other on-demand routing protocols is that it isbeacon-less and hence does not require periodichello packet (beacon) transmissions, which are usedby a node to inform its neighbors of its presence.The basic approach of this protocol (and all otheron-demand routing protocols) during the routeconstruction phase is to establish a route byflooding Route Request packets in the network. Thedestination node, on receiving a Route Requestpacket, responds by sending a Route Reply packetback to the source, which carries the route traversedby the Route Request packet received. Consider asource node that does not have a route to thedestination. When it has data packets to be sent tothat destination, it initiates a Route Request packet.This Route Request is flooded throughout thenetwork. Each node, upon receiving a RouteRequest packet, rebroadcasts the packet to itsneighbors if it has not forwarded it already,provided that the node is not the destination nodeand that the packet’s time to live (TTL) counter hasnot been exceeded. Each Route Request carries asequence number generated by the source node andthe path it has traversed. A node, upon receiving aRoute Request packet, checks the sequence numberon the packet before forwarding it. The packet isforwarded only if it is not a duplicate RouteRequest. The sequence number on the packet isused to prevent loop formations and to avoidmultiple transmissions of the same Route Requestby an intermediate node that receives it throughmultiple paths. Thus, all nodes except thedestination forward a Route Request packet duringthe route construction phase. A destination node,after receiving the first Route Request packet,replies to the source node through the reverse path

the Route Request packet had traversed. Nodes canalso learn about the neighboring routes traversed bydata packets if operated in the promiscuous mode(the mode of operation in which a node can receivethe packets that are neither broadcast nor addressedto itself). This route cache is also used during theroute construction phase. The important featuresinclude: Routes maintained only between nodeswho need to communicate, reduces overhead ofroute maintenance, Route caching can furtherreduce route discovery overhead, a single routediscovery may yield many routes to the destination,due to intermediate nodes replying from localcaches and its limitations are: Packet header sizegrows with route length due to source routing,Flood of route requests may potentially reach allnodes in the network, care must be taken to avoidcollisions between route requests propagated byneighboring nodes.

4 DESIGN OF EADSR

The DSR protocol is composed of twomechanisms that work together to allow thediscovery and maintenance of source routes in thead hoc network. They are route discovery and routemaintenance mechanisms. Modifications areproposed for each mechanism to add powerawareness to DSR and help in efficient usage ofenergy in the nodes [9]. Before discussing thechanges in design, some assumptions are made tosimplify the design process. It is assumed that allnodes wishing to communicate with other nodeswithin the sensor network are willing to participatefully in the protocols of the network. In particular,each node participating in the network should alsobe willing to forward packets for other nodes in thenetwork. Error in transmission medium is assumedto be zero. Nodes within the sensor network that thenetwork is static and wireless transmission range ofthe particular underlying network.

There have been two major modifications done toDSR algorithm: Change the routing algorithm DSRso that, given two nodes between which we mustestablish a multi-hop path. The path chosen amongall the possible ones is such that passing throughthe nodes along the path have a higher level ofenergy and less delay at that moment.

Modify the algorithm so that, when the energy ofa node which is forwarding data within a multi-hoppath reaches a level below or equal to a certainthreshold of the initial energy. This node willrequest the neighbors to look for another path, ifavailable. Thus it avoids consuming the residualenergy in a short time and also network partitions.

49


4.1 Modifications to route discovery phase

When a node receives a Route Request (RREQ)which is not a final receipt, before forwardingRREQ packets to neighboring nodes, it waits for apseudo random amount of time interval. This timeinterval is estimated from a uniform distribution ofprobability between 0 and a steady BroadcastJitter.The standard value of BroadcastJitter is 0.01seconds.

The basic idea of the EADSR is that such a delay,instead of being random, should be inverselyproportional to the level of residual energy of thenode in that moment. In this way, the first RREQthat will come to destination node will bechanneled through the best route from the point ofview of "overall" energy. It is nothing but the sumof the energy levels of the intermediate nodes ismaximum compared to all other possible pathsfrom source to destination. Consequently, the totaldelay between the sending of RREQ by the sourceand the receipt by destination will be minimum.

4.2 Modifications to route maintenance phase

In the traditional DSR algorithm, once a certainpath is chosen for sending a stream of packets to acertain destination, it tends to be used until 'one ormore' nodes that compose it are no longer available(consume all their energy, moving outside of therange of neighboring nodes). The consequence isthat the nodes which consume more energy intransmission and reception tend to drain rapidlythan. This phenomenon may not be desirable, assome of these nodes may subsequently have thedata to be transmitted and in the absence of energy,it would not be able to do so. Even in the absenceof such a requirement, it may still be desirable tomaintain a balance in the energy consumption ofnodes, and the consequent breaks of the links,which sooner or later leads to the division of thenetwork into two or more non-communicatingpartitions.

To avoid the above consequence, the followingchanges are incorporated to DSR algorithm:

1. When the energy of an intermediate node X,forwarding data within a multi-hop pathreaches a level less than or equal to a certainthreshold of the initial energy. The node Xsends a special broadcast packet to itsneighbors Y, containing a flag in the header"low energy" set to 1, with which it impliesasks not to continue to forward packets to it,if there are other paths to the destinationnode.

2. Each neighbor Y which receives thisbroadcast packet of "low energy" or “lowenergy notification”, it eliminates the pathscontaining the link Y-> X from its RouteCache , which is considered to be "virtually"dead (although in fact it is still working, nothaving X completely exhausted of itsenergy). If there are packets that have yet tobe forwarded along the link Y-> X, they arestill sent, to avoid destabilizing the network.

3. If one of the neighbors Y receives a packetto be forwarded, containing in its path thelink Y-> X, Y will attempt to save (salvage)the pack et, or will generate a Route Error tobe sent in broadcast, so that it reaches thesource of this packet, which will attempt tore-send the package using a different path,not containing the link Y-> X.

4. If the source cannot find other paths in itsRoute Cache, and on sending of a RouteRequest, if it appears that the node X is stillactive, and it appears to be the only "bridge"node to a certain destination, it will be stillused to forward packets.

5 SIMULATION ENVIRONMENT

The simulation environment is created with annetwork simulator NS2 [17] in the area of 1000 m x1000 m. Simulation is carried out for the scenarioscontaining 10, 20 and 30 nodes with the simulationtime of 200 seconds and each node having theinitial energy of 100 joules.

Fig. 1. Simulation scenario with all the nodes havingequal energy

50


Simulation is done by considering the channel aswireless, propagation as two-ray ground, queue sizeat each node is set to 50 where the queue can holdmaximum of 50 packets at each node, antenna typeis omnidirectional where it can receive and transmitin all directions.

Fig. 2. Route request flooding

The constant bit rate (CBR) model [9] was usedfor transmission of data. In this model the datapacket is transmitted periodically with a packet sizeof 512 bytes from source to destination.

Fig. 3. Red nodes with energy below threshold

Analysis of these packet scan determine theperformance effects from parameter variation,

routing protocols and are performed using awkscripts. The nam file contains information on thetopology such as node movement traces and events.

6 RESULTS AND PERFORMANCEANALYSIS

The simulation is carried out in NS2 simulationenvironment for DSR and EADSR protocols toobtain certain parameters for comparison. Trace filesand NAM files are generated for each simulation.Trace files contain various values like packets sent,packets received, energy of nodes, and so on. Thesevalues represent the network behavior for theparticular topology. The required values areextracted from Trace files using AWK scripts.Graphs are plotted for the extracted values likeExhaustion time of first node, Network Lifetime,Packet Delivery Ratio, and Residual Energy at theend of the simulation.

Fig. 4. Comparison of exhaustion of first node, EADSRoutperforms.

When a node dies, the other nodes that can bereached only through that node becomeunreachable. This is because network starts todestabilize with the energy exhaustion of first node.The figure 4 shows the exhaustion time of firstnode when using DSR is low compared to EADSR.This clearly indicates that EADSR maintains thenetwork stability for a long time compared to DSR.

51


Fig. 5. Residual Energy versus Number of sensor nodes

The total residual energy gives the total energyconsumed in the network before the end of thesimulation. This metric doesn't give muchinformation about how efficiently the energy isused but provides information about how muchenergy is saved during the routing process. Thefigure 5 shows that both DSR and EADSR performequally well in terms of residual energy.

Fig. 6. Packet Delivery Ratio versus Number of sensornodes

The ratio of number of data packets sent bysource through the network and data packetsreceived at the destination gives the packet deliveryratio. This ratio provides the information about howefficiently the routing is done and how successfulthe network is in data transmission. EADSRperforms equally well as DSR but that value isnegligible when compared with the energyefficiency gained.

Fig. 7. Time versus Number of sensor nodes

The time taken for 30% of the nodes in thenetwork to completely lose their energy is taken asthe Network lifetime. As soon as more nodes startto lose their energy, the network become unstableand paves way for Route Request Flooding in thenetwork. Many of the nodes become unreachabledue to death of 'bridge' nodes. Hence Networklifetime is one of the most important metrics tomeasure the network performance. The figure 7reveals that EADSR, compared to DSR, maintainsthe network stability for a long time, and avoids thepartition of network into non-communicatingdisjoint sets. The network will be available for datatransmission for a long time compared to DSR.Thus EADSR achieves the goal of efficiently usingthe energy.

7 CONCLUSION AND FUTURE WORK

Efficient usage of energy is one of the mostimportant concerns in sensor network, especiallywhile designing a routing protocol. This paper aimsat discovering an efficient power aware routingscheme for ad hoc network with less or no mobility.Simulation result shows that the proposed schemeEADSR outperforms other existing routingprotocols in terms of different energy relatedparameters. At the time route selection, EADSRtake crucial thing like battery status of the path intoconsideration while also giving importance todelay. With this main factor in consideration,EADSR always select less congested and morestable route for data delivery. Although this schemecan somewhat increase the latency of the datatransfer but it results in a significant power saving

52


and long lasting routes. It maintains the existingperformance of DSR and adds to it the powerawareness to increase the network lifetime. Themost significant benefit by adding energyawareness is that it avoids, for a long time,partitioning of network into non-communicatingdisjoint sets. Further the EADSR protocol may beused in real life scenario is straight forward but mayrequire little modifications. More wide research isnecessary to further increase the efficient usage ofenergy in sensor network.

8 REFERENCES

[1] I. F. Akyildiz and M. C. Vuran, “WirelessSensor Net- works,” 1st Edition, John Wiley &Sons, Ltd, Chichester, 2010.doi:10.1002/9780470515181

[2] Á. Lédeczi, A. Nádas, P. Völgyesi, G. Balogh,B. Kusy, J. Sallai, G. Pap, S. Dóra, K. Molnár,M. Maróti and G Simon, “CountersniperSystem for Urban Warfare,” ACMTransactions on Sensor Networks, Vol. 1, No.2, 2005, pp. 153-177.doi:10.1145/1105688.1105689

[3] T. He, S. Krishnamurthy, J. A. Stankovic, T.Abdelzaher, L. Luo, R. Stoleru, T. Yan, L. Gu,G. Zhou, J. Hui and B. Krogh, “VigilNet: AnIntegrated Sensor Network System for Energy-Efficient Surveillance,” ACM Transactions onSensor Networks, Vol. 2, No. 1, 2006, pp. 1-38.doi:10.1145/1138127.1138128

[4] Mourtaji, I., et al. "A New Technique forAdapting SIP Protocol to Ad Hoc Networks:VNSIP (Virtual Network for SIP) Illustrationand Evaluation of Performance" InternationalJournal of Computer Networks andCommunications Security 1.1 (2013).

[5] S. Bandyopadhyay and E. Coyle, “An EnergyEfficient Hierarchical Clustering Algorithm forWireless Sensor Networks,” Proceedings of the22nd Annual Joint Confer-ence of the IEEEComputer and CommunicationsSocieties(INFOCOM2003), San Francisco, 30March-3 April 2003, pp. 1713-1723.

[6] Y. Chen and Q. Zhao, “On the Lifetime ofWireless Sen-sor Networks,” IEEECommunications Letters, Vol. 9, No. 11, 2005,pp. 976-978.doi:10.1109/LCOMM.2005.11010

[7] A. Norouzi and A. Sertbas, “An IntegratedSurvey in Efficient Energy Management for

WSN Using Architec-ture Approach,”International Journal of Advanced Net-workingand Applications, Vol. 3, No. 1, 2011, pp. 968-977.

[8] A. Kanavalli, D. Sserubiri, P. D. Shenoy, K. R.Venugopal and L. M. Patnaik, “A Flat RoutingProtocol for Sensor Networks,” Proceeding ofInternational Conference on Methods andModels in Computer Science, Delhi, 14-15December 2009, pp. 1-5.

[9] Yi hua zhu, Wan-deng wu, Jian pan, Yi-pingtang, An energy efficient data gatheringalgorithm to prolong lifet ime of wirelesssensor networks”Computer Communications33(2010) 639-647.

[10] F.Bajaber, I.awan, “Centralized dynamicclustering for wireless sensor network”inInternational Conference on AdvancedInformation Networking and ApplicationsWorkshops, 2009, pp 193-198.

[11] Ameer Ahmed Abbasi, Mohamed Younis, “ASurvey on Clustering Algorith ms for WirelessSensor Networks”, Computer Communications30 (2007) 2826-2841.

[12] Kemal Akkaya, Mohamed Younis, “A Surveyon routing protocols for Wireless SensorNetworks”, Ad Hoc Networks 3 (2005) 325-349.

[13] Miau Yu, Jason H.Li and renato Levy,“Mobility Resistant Clustering in Multi-HopWireless Networks, Journal of Networks,Vol.1, No.1, May 2006.

[14] O. Younis, S. Fahmy, “HEED: AHybrid,Energy-Efficient, Distributed clusteringapproach for Ad Hoc sensor networks ,IEEETransactions on Mobile Computing 3 (4)(2004) 366–379.

[15] S. Lindsey, C. Raghavendra,"PEGASIS: PowerEfficient Gathering in Sensor InformationSystems,” IEEE Aerospace ConferenceProceedings, 2002, Vol. 3. No. 9-16, pp. 1125-1130.

[16] J. Kulik, W. R. Heinzelman, and H.Balakrishnan, "Negotiation-based protocols fordisseminating information in wireless sensornetworks," Wireless Networks, Volume: 8, pp.169-185, 2002.

[17] NETWORK SIMULATOR NS2 :WWW.ISI.EDU/NS/NSNAM/NS

53


AUTHOR PROFILES:

Prof. Nandini Prasad K Sreceived degree in computerscience & engineering fromPESIT, Bangalore and PGdegree from VTU, Belgaum.She is a research student of

Kuvempu University. She is pursuing Ph.D underthe guidance of Dr. Puttamadappa C. Currently; sheis an Associate Professor at Dr. AIT, Bangalore.

Dr. C Puttamadappa obtainedhis B.E degree from MysoreUniversity, PG degree fromBangalore University andDoctral degree from JadavpurUniversity, Kolkatta. Currently,he is working as Principal and

Professor at Sapthagiri college of Engineering,Bangalore. His research interest’s areas are Devices& Networks.

e ofAuthor

Design and Performance Analysis of Energy Aware Routing Protocol for Delay Sensitive Applications...

Engineering

Transcript of Design and Performance Analysis of Energy Aware Routing Protocol for Delay Sensitive Applications...