Your SlideShare is downloading. ×
Neural network based energy efficient clustering and routing
Neural network based energy efficient clustering and routing
Neural network based energy efficient clustering and routing
Neural network based energy efficient clustering and routing
Neural network based energy efficient clustering and routing
Neural network based energy efficient clustering and routing
Upcoming SlideShare
Loading in...5

Thanks for flagging this SlideShare!

Oops! An error has occurred.

Saving this for later? Get the SlideShare app to save on your phone or tablet. Read anywhere, anytime – even offline.
Text the download link to your phone
Standard text messaging rates apply

Neural network based energy efficient clustering and routing


Published on

Published in: Education, Technology
1 Comment
1 Like
  • nice one...
    but how to interconnect these with neural network matlab toolbox?
    Are you sure you want to  Yes  No
    Your message goes here
No Downloads
Total Views
On Slideshare
From Embeds
Number of Embeds
Embeds 0
No embeds

Report content
Flagged as inappropriate Flag as inappropriate
Flag as inappropriate

Select your reason for flagging this presentation as inappropriate.

No notes for slide


  • 1. 2009 First International Conference on Networks & Communications Neural Network Based Energy Efficient Clustering and Routing in Wireless Sensor Networks 1 Neeraj Kumar, 2 Manoj Kumar, 3R.B. Patel 1,2 School of computer Science and Engineering, SMVD University, Katra (J&K), India 3 Department of Computer Science and Engineering, MM University, Mullana (Ambala), Haryana E-mail:,, * , R.B. Patel Shri Mata Vaishno Devi University, Katra(J&K), India Abstract power radio transmission is employed, wireless Energy is a valuable resource in Wireless Sensor communication is far from being perfect [4,5,6]. Networks (WSNs). The status of energy consumption In this paper, we address the issue of energy-efficient should be continuously monitored after network clustering and routing in WSNs using neural networks deployment. The information about energy status can be with the objective of maximizing the network lifetime. used to early notify both sensor nodes and Network First, we propose an efficient neural network based Deployers about resource depletion in some parts of the clustering algorithm for WSNs. Secondly, we propose network. It can also be used to perform energy-efficient routing and data transmission algorithm for WSNs. We routing in WSNs. In this paper, we propose a neural define an efficient metric to be used in taking the network based clustering and energy efficient routing in selection of next hop in routing. The problem is WSN with the objective of maximizing the network lifetime. formulated as LP with specified constraints and routing In the proposed scheme, the problem is formulated as metric linear programming (LP) with specified constraints. Rest of the paper is organized as follows: Section 2 Cluster head selection is done using adaptive learning in discusses related work, section 3 defines the energy neural networks followed by routing and data model used along with the defined routing metric and transmission. The simulation results show that the problem formulation, Section 4 describe the proposed proposed scheme can be used in wide area of solution, Section 5 provides the simulation and results applications in WSNs. obtained, and finally Section 6 concludes the article. Keywords: Sensor networks; Energy Efficient Routing, Linear Programming, Neural Networks. 2. Related Work There are number of clustering protocols have been 1. Introduction proposed in literature e.g. LEACH [7], PEGASIS [8], Wireless Sensor Networks (WSNs) is a class of wireless HEED [9], EEUC [10], and FLOC [11]. The cluster ad hoc networks in which sensor nodes collect, process, formation overhead of the clustering protocols includes and communicate data acquired from the physical packet transmission cost of the advertisement, node environment to an external Base-Station (BS). But these joining and leaving, and scheduling messages from networks have several challenges such as sensor nodes in sensor nodes. All these protocols do not support adaptive WSNs are normally battery-powered, and hence energy multi-level clustering, in which the clustering level has to be carefully used in order to avoid early cannot be changed until the new configuration is not termination of sensors’ lifetimes [1]. As such, the made. Therefore, the existing protocols are not adaptable concept of continuous monitoring of network resources to the various node distributions or the various sensing becomes a very important topic in WSNs. This same area. If the sensing area is changed by dynamic concept has been already investigated in many other circumstances of the networks, the fixed-level clustering environments, e.g., power plants [2], and in many protocols may operate inefficiently in terms of energy distributed systems [3]. consumption. Many recent experimental studies have shown that, Bandyopadhyay and Coyle [12] proposed the especially in the field of sensor networks where low randomized clustering algorithm to organize sensors into978-0-7695-3924-9/09 $26.00 © 2009 IEEE 34DOI 10.1109/NetCoM.2009.56 Authorized licensed use limited to: Bharat University. Downloaded on July 10,2010 at 18:41:30 UTC from IEEE Xplore. Restrictions apply.
  • 2. clusters in a wireless sensor network. Computation of the 3.2 Energy Modeloptimal probability of becoming a cluster head was To ascertain the amount of energy consumed by a radiopresented. Moscibroda and Wattenhofer [13] defined the transceiver, we apply the following energy model. Formaximum cluster-lifetime problem, and they proposed each packet transmitted by a sending node to one ordistributed, randomized algorithms that approximate the more receivers in its neighborhood, the energy isoptimal solution to maximize the lifetime of dominating calculated as according to [7]:sets on wireless sensor networks. Pemmaraju and e = et + ne r + ( N − n)e h r ………………………(1)Pirwani []14 considered the k-domatic partition problem, Where et and e r denote the amount of energy requiredand they proposed three deterministic, distributedalgorithms for finding large k-domatic partitions. Tan to send and receive, n the number of nodes which shouldand Korpeoglu [15] proposed two new algorithms under receive the packet, and N the total number of neighborsthe name PEDAP, which are near optimal minimum in the transmission range. e h r quantifies the amount ofspanning tree based wireless routing scheme. The energy required to decode only the packet headerperformance of the PEDAP was compared with LEACH According to model described in [7], et and e r areand PEGASIS, and showed a slightly better network defined aslifetime than PEGASIS. Yi et. al[16] presents a Powerefficient and adaptive clustering protocol PEACH et (d , k ) = (e elect + e amp * d ρ )8k ………………..(2) e r (k ) = e elect * 8k3 Models, Routing Metric and Problem Formulation for a distance d and a k byte message. We have3.1 Network ModelWe consider a network of homogeneous and energy- set e elect = 70nJ / bit , e amp = 120 pJ / bit / m 2 , d = 50m ,constrained sensor nodes that are randomly deployed in a ρ=4sensor field. Sensor nodes are initially powered bybatteries with full capacities. Each sensor collects data For a given header size n bytes, e h r would bewhich are typically correlated with other sensors in its accordingly calculated.vicinity, and then the correlated data is sent to the BS viaCluster Head (CH) for evaluation or decision making 3.2 Routing Metric and Problem Formulationpurposes. We assume periodic sensing with the same A proper routing metric has to be chosen that can be usedperiod for all sensors. To facilitate the operation of the to decide the next hop for data transmission. This metricnetwork, we apply a novel clustering scheme that results always ensure the best shortest route and incurs the leastin selection of cluster. Inside each fixed cluster, a node is energy to transmit the packet from source to destination.periodically elected to act as CH through which The cost of a link between two nodes S i and S j is equalcommunication to/from cluster takes place. to the energy spent by these nodes to transmit and to BS receive one data packet, successfully. Source The metric chosen is Routing cost is calculated as follows: ⎛ Ei D ⎞ R_C =⎜ ⎟ , ………..(3) ⎜ Et ( S i , S j ) + E r (S i , S j ) ⎟ ⎝ ⎠ Where E i D is energy associated with the delivery ratio of the packet originating from source node S i and correctly received at destination node, while E t ( S i , S j ) .is the energy used in transmitting from S i to S j and E r ( S i , S j ) is the energy used in receiving the packet. Data routing from every cluster head to the sink is done over multi-hop paths, which is given by Sensors CH BS minimizing equation (3)Fig.1: Data Transmission in typical Sensor Networks 4 Proposed Solution The solution for the energy aware routing problem is proposed using an LP formulation. The objective of the 35 Authorized licensed use limited to: Bharat University. Downloaded on July 10,2010 at 18:41:30 UTC from IEEE Xplore. Restrictions apply.
  • 3. LP is to select a number of nodes with higher levels ofresidual energy to form an optimal route, while Output Layer Electedminimizing the total routing cost. Let us label the base- CHstation as node 0 and label the CH nodes as nodes 1 to n,where n is the total number of CH sensor nodes. So the Competitionproblem reduces to Layer Minimize ∑ R_C 1≤i ≤ nSubject to following constraints δ ( s, w 2 ) δ ( s, wm ) ∑Dij − 1≤ j ≤ n ∑ D ji = bi ………….(4) 1≤ j ≤ n δ ( s, w1 ) Dij ≥ 0 , 1 ≤ j ≤ n ………….….(5) E ≤ Pmax imum ……………………(6) Input LayerConstraint (4) specifies the amount of data transmitted bibetween two nodes S i and S j , Constraint (5) specifiesamount of data to be transmitted from two nodesS i and S j , Constraints (6) guarantees a minimum node Input Sensor node for election to be CHlifetime and limits the maximum power consumption ofany node in the network. Fig.2: Selection of CHThe proposed protocol is divided into two phases namely Neural networks have solved a wide range ofas: setting up phase and energy aware routing and data problems and have good learning capabilities. Theirtransmission phase. strengths include adaptation, ease of implementation, parallelization, speed, and flexibility. A two - layer feed4.1 Setup Phase forward neural network that implements the idea ofIn this part, the initial cluster head selection and cluster competitive learning is depicted in Figure 2 above. Theformation algorithm are introduced , followed by the nodes in the input layer admit input patterns of sensorenergy aware routing. nodes competing for CH and are fully connected to the output nodes in the competitive layer. Each output nodeCluster Head Election corresponds to a cluster and is associated withTo ensure balanced energy consumption among the weight W j , j = 1,2,...., m , where m is the number ofcluster head nodes throughout the network lifetime, clusters.many clustering protocols favor uniformly distributed The neurons in the competitive layer then competeclusters with stable average cluster sizes [7-11]. But we with each other, and only the one with the smallestpropose a new neural network based coverage aware E i D value becomes activated or fired. Each neuron in theclustering algorithm. The set of cluster head nodes canbe selected based on the cost metric defined in equation proposed algorithm for CH selection has an adaptive3. The densely populated parts of the network will be learning. The learning rate μ determines the adaptationovercrowded with cluster head nodes, while the scarcely of the vector towards the input pattern and is directlycovered areas will be left without any cluster head nodes. related to the convergence. If μ equals zero, there is noIn such a situation, it is likely that the high cost sensors learning. If μ is set to one, it will result in fast learning,from poorly covered areas will have to perform and the prototype vector is directly pointed to the inputexpensive data transmissions to distant cluster head pattern. For the other choices of μ , the new position ofnodes, further reducing their lifetime. There are threelayers in the proposed neural network: Input layer, the vector will be on the line between the old vectorCompetition layer and Output Layer. value and the input pattern. Generally, the learning rate could take a constant value or vary over time. 36 Authorized licensed use limited to: Bharat University. Downloaded on July 10,2010 at 18:41:30 UTC from IEEE Xplore. Restrictions apply.
  • 4. routes, the sink node first generates a Route Discovery 1. Initialize the Vector S = {S1 , S 2 ,...., S m } of message that is broadcasted throughout the network. sensor nodes competing for Cluster head. Upon receiving the broadcast message, each sensor node //Processing at Input Layer introduces a delay proportional to its cost before it 2. Choose a winner k from sensor nodes as CH forwards the Route Discovery message to nodes in range R . In this way a message arrives at each node whose E i D is minimum as follows along the desired minimum cost path. The cumulative k = arg min{E i D } // Competition Layer cost of the routing path from the sink to the node obtained in this phase is called the energy aware routing 3. Also E i D smallest Euclidean distance to BS i.e. cost of the node described in (3). Ei D = k ∑| S i =1, 2 ,...m i − BS | , where k is proponality 1. Sort the paths p1 , p 2 ,....... p m according to constant E i D as E i D1 < E i +1 D+11 < .......E m Dm 4.Update the value of weight vector as follows: 2. j = 1 // initialize the counter for available paths. w j (new) = w j (old ) + μ ( S i − w j (old )) , where 3. repeat and calculate E ≤ Pmax imum (Constraint 6) μ is learning rate of the neurons. 0 ≤ μ ≤1 4. repeat 5. Repeat Steps (2-4) iteratively. 5. 6. Neuron with smallest value of E i D is E C ,m = ∑ m i E max L // E C ,m is use to i =1, 2...m winner.// Output Layer store the minimal energy consumption per bit with m paths and is assigned maximum value initially. Fig.3: Algorithm for Cluster head selection 6. R _ C = 0 // initialize the value of routing cost4.2 Routing and Data Transmission 7. repeatAn algorithm for routing and data transmission is 8. Solve equation (3) and get the correspondingproposed in Figure 5. optimal energy distribution with respect toLet us denote by R the maximum number of routes that constraints defined in equations (4),(5),(6).exist between each source-destination pair, and l indicateby a route in R. Also, denote by pow( S i , l ) the power 9. Calculate E C ,m = Ei L∑ i =1, 2...mconsumed by node S i in transmitting to the next node on 10. Calculate the value of R _ C from equation (3)route l . For the sake of simplicity, we assume that this and R _ C updated = R _ C // Update value of routingparameter depends only on the distance between the costtransmitting and the receiving node. Then, we associate 11. Until | R − C updated − R − C |< δ 1 (predefinedwith each route l an energy cost routing metric definedin equation (3) above. The proposed algorithm scan all threshold)routes in R and determine the least expensive route to 12. Update the values of energy for each datareach the BS. A source will select the route that has the transmission as E C ,m Updated = E C , mleast energy consumption or the one that maximizes thenetwork lifetime. Until | E C ,m Updated − E C ,m |< δ 2 13. j = j + 1 // Update the counter of the paths 14. Until m > Destination _ node Source Destination 15. Compare all paths using R _ C metric and select the smallest one. 16. Send the data across the multiple paths defined. Fig. 4: Multipath Routing Fig. 5: Algorithm for Routing and Data TransmissionRoute UpdateThe cluster head nodes send their data over multi-hop Given m available paths, the overall energy consumptionpaths to the sink as shown in Figure 4. To obtain these per packet, E, can be written as 37 Authorized licensed use limited to: Bharat University. Downloaded on July 10,2010 at 18:41:30 UTC from IEEE Xplore. Restrictions apply.
  • 5. E= ∑ E L , where E i =1, 2...m i i is the energy consumption forone bit along path i and L is the packet length in bits. 100 Proposed 80 PEACH5. Simulation and Results Number of Alive NodesWe have considered a stationary WSN of size 400×400 60with a maximum transmission range of 50 m. Themessage length is assumed to be 48 bytes, including an 4012 byte packet header. The energy used to receive andtransmit data is modeled according to the energy model 20presented in Section 3. Other sources of energyconsumption like sensing, processing, and idle listening 0are neglected. MAC-layer behaviors such as contention, 1000 2000 3000 4000 5000duty cycles, or packet buffering are not addressed. We Number of Roundshave simulated the proposed scheme on ns-2[17]. Fig. 7: Number of alive nodes in PEACH and Proposed Figure 6 presents the energy consumption of the Schemeproposed scheme with well known PEACH clusteringprotocol [16] when the maximum transmission range is 100 Proposed60 m. The results demonstrate that the energy PEACHconsumption of proposed neural network based 90clustering is smaller than PEACH. Percentage of alive Nodes 80 0.10 70 Proposed PEACH 0.08 60Mean Residual Energy 50 0.06 40 0.04 50 100 150 200 250 300 Number of Nodes 0.02 Fig. 8: Percentage of alive nodes in PEACH and Proposed Scheme after 1500 rounds 0.00 1000 2000 The percentage of nodes alive and the mean of residual 3000 4000 5000 energy versus the number of nodes after 1500 rounds are Number of Rounds Fig.6: Mean residual Energy in PEACH and Proposed presented in Figures 8 and 9. Proposed scheme has Scheme highest percentage of residual energy compared with PEACH protocol. Also, the variation in the mean of Figure 7 presents the number of nodes alive when residual energy of proposed scheme is smaller thanusing clustering in proposed scheme and PEACH. This PEACHresult directly reflects the network lifetime of thewireless sensor networks. In the case of networks usingPEACH with the maximum transmission range r = 60 m,where a node runs out of energy occurs nearly after 4000rounds, while in propped scheme there is a slightimprovement and node runs out of energy in nearly4200 rounds. . 38 Authorized licensed use limited to: Bharat University. Downloaded on July 10,2010 at 18:41:30 UTC from IEEE Xplore. Restrictions apply.
  • 6. [6]. J. Zhao, R. Govindan, Understanding packet 0.10 delivery performance in dense wireless sensor Proposed networks, in: Proceedings of the 1st ACM PEACH International Conference on Embedded Networked 0.08 Sensor Systems, SENSYS, ACM Press, Los Angeles, CA, USA, 2003.Mean Residual Energy 0.06 [7]. W.R. Heinzelman, A. Chandrakasan, H. Balakrishnan, Energyefficient communication 0.04 protocol for wireless microsensor networks, in: Hawaii International Conferenceon System 0.02 Sciences (HICSS), 2000. [8]. S. Lindsey, C. Raghavendra, K.M. Sivalingam, Data gathering algorithms in sensor networks 0.00 50 100 150 200 250 300 using energy metrics, IEEE Transactions on Number of Nodes Parallel and Distributed Systems, 13: 9, 924–935,Fig. 9: Mean residual Energy in PEACH and Proposed Scheme 2002. after 1500 rounds [9]. O. Younis, S. Fahmy, Heed: A hybrid, energy- efficient, distributed clustering approach for ad 6. Conclusions hoc sensor networks, IEEE Transactions onThis paper has proposed a neural network based energy Mobile Computing, 3: 4, 366–379, 2004.efficient routing and clustering protocol for WSNs. The [10]. C. Li, M. Ye, G. Chen, J. Wu, An energy-efficientselection of CH is done using adaptive learning unequal clustering mechanism for wireless sensormechanism. Simulations results show that it performs networks, in: Proceedings of the 2nd IEEEbetter than existing routing protocol PEACH in terms of International Conference on Mobile Ad-hoc andresidual energy and .number of alive nodes. So the Sensor Systems (MASS’05), 2005.proposed scheme can be used in wide areas of sensor [11]. M. Demirbas, A. Arora, V. Mittal, Floc: A fastnetworks where energy efficiency is a critical issue. local clustering service for wireless sensor networks, in: Orkshop on Dependability Issues in References Wireless Ad Hoc Networks and Sensor Networks ( [1]. J.N. Al-Karaki, A.E. Kamal, Routing techniques DIWANS/DSN), 2004. in wireless sensor networks: a survey, IEEE [12]. S. Bandyopadhyay, E.J. Coyle, An energy- Wireless Communications, 11:6, 6–28, 2004. efficient hierarchical clustering algorithm for [2]. P. Bates, Debugging heterogeneous distributed wireless sensor networks, in: IEEE INFOCOM, systems using eventbased models of behavior, 3,1713–1723, 2003. ACM Transactions on Computer Systems, 13: 1 [13]. T. Moscibroda, R. Wattenhofer, Maximizing the 1995. lifetime of dominating sets, in proc. of 19th [3]. C. Frei, Abstraction techniques for resource International Parallel and Distributed Processing allocation in communication networks, Ph.D. Symposium, 2005. dissertation, Swiss Federal Inst. Technol. (EPFL), [14]. S.V. Pemmaraju, I.A. Pirwani, Energy Lausanne, Switzerland, 2000. conservation in wireless sensor networks via [4]. Cerpa, N. Busek, D. Estrin, Scale: A tool for dominating partitions, in: MOBIHOC, 2006. simple connectivity assessment in lossy [15]. H.O. Tan, I. Korpeoglu, Power efficient data environments, Tech. Rep. 21, Center for gathering and aggregation in wireless sensor Embedded Networked Sensing, University of networks, Issue on Sensor Network Technology, California, Los Angeles, CA, USA, September SIGMOD Record. 2003. [16]. Sangho Yi , Junyoung Heo , Yookun Cho , Jiman [5]. M. Yarvis, W. Conner, L. Krishnamurthy, J. Hong, PEACH: Power-efficient and adaptive Chhabra, B. Elliott, A. Mainwaring, Real-world clustering hierarchy protocol for wireless sensor experiences with an interactive ad hoc sensor networks, Computer Communications, 30, 2842– network, in: Proceedings of the 31st IEEE 2852, 2007. International Conference on Parallel Processing [17]. K. Fall and K. Varadhan (editors), NS notes and Workshops, ICPPW, IEEE Computer Society, documentation, The VINT project, LBL, Feb Vancouver, BC, Canada, 2002. 2000, . 39 Authorized licensed use limited to: Bharat University. Downloaded on July 10,2010 at 18:41:30 UTC from IEEE Xplore. Restrictions apply.