Published on

IJERA (International journal of Engineering Research and Applications) is International online, ... peer reviewed journal. For more detail or submit your article, please visit

Published in: Technology
  • Be the first to comment

  • Be the first to like this

No Downloads
Total views
On SlideShare
From Embeds
Number of Embeds
Embeds 0
No embeds

No notes for slide


  1. 1. V.Chandrasekaran, S.Anitha / International Journal of Engineering Research and Applications (IJERA) ISSN: 2248-9622 Vol. 2, Issue5, September- October 2012, pp.2128-2131 A Novel Energy Efficient Distributed Clustering in Wireless Sensor Networks V.Chandrasekaran*, S.Anitha** *(Department of ECE, Velalar College of Engineering & Technology, Anna University, India) ** (Department of Information Technology, Kongu Engineering College, Anna University, India)ABSTRACTUnder increased level of traffic, Hot spots emerge to nodes through inter-cluster a wireless sensor network. This leads to thedepletion of network resources and disrupts thenetwork operation completely. This scenario ishigh in the case of clustering process as clusterheads perform more data processing than othersensor nodes. This increase in load on ClusterHead (CH) varies with the distance to the sink.To eliminate this problem, data aggregation canbe performed to reduce the processing andtransmission of information. This approachworks well in static environment but it iscomplex and expensive in the case of dynamicnetwork. So the role of Cluster head is rotatedamong the different nodes in the network, thusbalancing the load and conserves the importantresources. The size of the cluster based on thedistance or number of hops to the sink is alsotaken into consideration. This paper focuses on Fig.1 Energy Consumption Modeldistributed clustering with multi hop routing tominimize the energy consumption, average end- Dynamic allocation of heavy traffic andto-end delay and improving network lifetime other functions of CH to different nodes help tocompared LEACH. prevent failure of nodes in case of resource depletion. In clustered environment, Hot-spots areKeywords – Clustering, Cluster Head, Energy an important site to take into account as the data isEfficiency, Wireless Sensor Networks. highly congested near to sink. So in distributed with multi hop routing critical location must beI. INTRODUCTION concentrated by transmitting the data through With resource-constrained Wireless Sensor multiple high rate routes. Thus network lifetime isNetworks, effective use of these battery operated improved by reducing failures because of depletionresources is a challenging task nowadays in the of energy.developing IT field. Though many clustering Hot –spot is an important site where alltechniques like hierarchical exists, need is to find an data are collected. As the distance to sink decreases,optimal energy-efficient clustering mechanism that the amount of traffic on the nodes exaggerates. Thisuses the limited resources in an efficient manner. correlation is to be studied mathematically to build a The cluster Head acts as the local controller balanced network. For this, route establishment,in the scalable network for the process of data route maintenance and other procedures have to becollection, data forwarding and performs taken into consideration to know about energyaggregation functions. Each Cluster is formed based depletion of each node. In our proposed work, loadon the coherence between the set of nodes or having sharing, distributed, and energy-aware clusteringminimum distance between them. The one having with multi hop routing algorithm is used. Energythe high processing power is chosen as the cluster efficient clustering determines the cluster sizeshead and it performs tasks like grouping, secrecy based on the number of hops to the sink node. Afterand allocation of tasks. It also controls various choosing the cluster heads, this algorithm performsevents like failure recovery and node movements. the efficient clustering balancing both traffic andVarious functions such as load sharing, membership complexity. Performance is achieved with multi hopabilities and inter-cluster establishment are assigned network design reducing overhead and improving the network lifetime. 2128 | P a g e
  2. 2. V.Chandrasekaran, S.Anitha / International Journal of Engineering Research and Applications (IJERA) ISSN: 2248-9622 Vol. 2, Issue5, September- October 2012, pp.2128-2131In the remainder of this paper, reviews of the related head among the whole set of candidates that areworks are briefly discussed on clustered network grouped under each cluster and associating nodeand their drawbacks in Section II. Section III with the cluster head.presents the distributed clustering with multi hop Probability measure is used to determinerouting algorithm and the calculation of its energy the candidates for each cluster head for differentconsumption levels. Then, Section IV dealt with the clusters. Each node in the sensor network willperformance evaluation compared with LEACH. decide to become as the candidate. Normally scaledFinally, Section V concludes the discussion. down value S for probability P K is chosen as the ratio of initial energy Ei to the average initial energyII. REVIEW OF RELATED WORKS of the network Eiavg. Pk=S (Ei / Eiavg) for a node k in In WSNs, Clustering is an efficient region Rm. After the network parameterstechnique to analyze and control the information in initialization, this probability value is calculatedthe network. Of which heavy traffic on the cluster only once. Each node k selects a random value onhead is a critical concern. To balance the load, the interval 0 to 1 at the beginning of each dataCluster Head is frequently rotated between different collection process. The node is selected as the CHnodes in the network as indicated in LEACH. But it candidate when the node value is lesser than thecauses increase in transmission cost due to long calculated probability Pk. Thus each sensor node isdistance as all nodes transmits the information compared with the measure to elect as thedirectly to the sink. Due to this, the nodes that are candidates for the CH. To obtain accurate measure,far from the sink has high depletion rate than the residual energy can be used instead of initial energynodes nearer. For this problem, EECS selects less at the expense of increased message overhead ornumber of nodes been far to sink. Even though complexity of the network. By this approach, theenergy depletion is not minimized due to single hop chance of being selected as the candidate is directlytransmissions and scalability is not assured for large proportional to the initial energy levels. Thereforenetworks. In HEED , Cluster Head is selected based energy consumption is reduced by avoiding theon the maximum energy it possesses and multi hop message overheads for broadcasting and additionalinter cluster communication is adopted, thus circuits for monitoring the status of each node.reducing the transmission cost of longer distance. Each selected candidate sends theWith small cluster size, the traffic on individual broadcast packet indicating their residual energy tocluster head is reduced distributing traffic among the candidates residing within the radius of r m whichdifferent clusters near to sink. This approach too has is calculated from the clustering algorithm. Onissues like overhead due to more number of clusters receiving the broadcast packets, each candidateand control packets used for inter cluster leaves the selection process if the higher residualcommunication. This again results in heavy traffic energy is received compared to them. Thus the CHthan the original message traffic. Therefore, is chosen among the candidates.considering the variability in traffic at different partsof the network, an analytical discussion is to be 3.2 Cluster Formationcarried to remove the imbalance between inter and To have knowledge of the presence of CHintra clustering energy utilization. In UCR and by all other non-CH nodes, announcement packet isPEBECS , informing the formation of cluster transmitted by the node having high residual energythroughout the network cause wastage in the energy within the radius αr m. Region-wide broadcast isavailable and also limited to small networks. So done to confirm the reception of availability packetsthere requires a need to develop an energy-efficient by all other non-CH nodes. This causes additionalclustering with multi hop routing protocol to reduce transmission cost. In order to avoid this, α is chosenthe control overheads and to eliminate broadcasting to have high probability of association between CHto all the nodes in the network. This protocol is and its candidates. The number of CH nodes in eachdesigned in such a way to compromise between region can be modeled using poisson distribution isdistance from the cluster nodes to the sink and the given by exp (pmπα2r2m ) in region Rm . To have highsize of the clusters. association between the CH and non-CH nodes, probability value is chosen as 99%.When multipliedIII. PROPOSED SCHEME with the distribution probability, minimum value of In WSN, Energy efficient clustering with α is selected to reduce transmission cost.multi hop routing is implemented in distributed After the gathering all the announcementenvironment by data collection processes like packets from the all CH nodes by the candidateenergy level equalization, hop distance towards sink nodes, CH is chosen by candidates in each clusterand certain tradeoffs as discussed above. based on the highest residual energy that CH possess. Each cluster selects the CH which is nearest3.1 Cluster Head Selection to it within the transmission range. This association At the starting of the data collection happens by sending request and reply messages byprocess, cluster is formed by electing the cluster candidates and CH nodes respectively. 2129 | P a g e
  3. 3. V.Chandrasekaran, S.Anitha / International Journal of Engineering Research and Applications (IJERA) ISSN: 2248-9622 Vol. 2, Issue5, September- October 2012, pp.2128-2131There may be nodes which do not joined in any ofthe clusters due to missing of announcement packetsat the end of the cluster formation stage. To haveassociation with the closest CH, the algorithmextends the transmission range of those nodes torecover from the node isolation problem. Among nnumber of nodes in wireless sensor networks, forselection of candidates’ nS messages are needed,where S is the scaled down value of probability ofinitial energy of the nodes in the network. For theCH selection, totally K messages are needed, whereK is the number of CH sending the announcementmessages.3.3 Inter-cluster routing in Distributed clusteredenvironment The aim of the routing algorithm is to Fig.2 Comparison of Network Lifetimeminimize the message and circuit overheadsrequired for the selection of candidates, clusterheads ,cluster formation and their association beforethe actual message transmission begins andcompromise the energy wastage among the nodes inthe network or between the clusters. To solve these issues, CH in region Rmtake its next hop to the sink through the nodes inregion Rm-1 The received nodes make reply based onthe timer it set. The timeout for replying is inverselyproportional to the residual energy of individualnodes. Thus the node having highest residual energywill timeout sooner sends the reply back to the CHfirst. When hearing this reply packet, remainingnodes resets its timer. By this, both overhead andenergy wastage is reduced.IV. PERFORMANCE EVALUATION In this section, the performance of our Fig.3 Average Residual Energy vs No. of RoundsE2DC is compared with LEACH. The parameters ofsimulations are listed in Tabel.1, and the parametersof the radio model are the same as LEACH. V. CONCLUSION In this paper, we present a novelTabel.1 Simulation Parameters distributed, energy efficient clustering scheme applied for periodical data gathering. E2DCParameters Value produces a uniform distribution of cluster headsNetwork Size 100m x 100m across the network through localizedNumber of Nodes 100 communication with little overhead. The results areInitial energy 2J better compared to LEACH. Simulation resultsEelec 50 nJ/bit show that E2DC prolongs the network lifetime and 5 nJ the total energy is efficiently consumed. All of ourEDA 25 bytes contributions here are focused on the cluster set-up stage. There are still much space to improve thePacket size performance of data transmission. In the large scale sensor networks, multi-hop communication is a mainstream technique for energy saving. REFERENCES [1] Dali Wei ,Yichao Jin and Serdar Vural, ”An Energy-Efficient Clustering Solution for Wireless Sensor Networks”, IEEE Transactions on Wireless 2130 | P a g e
  4. 4. V.Chandrasekaran, S.Anitha / International Journal of Engineering Research and Applications (IJERA) ISSN: 2248-9622 Vol. 2, Issue5, September- October 2012, pp.2128-2131 Communications,vol. 10, no. 11, pp.3973- [6] S. Vural and E. Ekici, “On multihop 3983,2011. distances in wireless sensor networks with [2] W. Heinzelman, et. al., An application- random node locations,” IEEE Trans. specific protocol architecture for wireless Mobile Comput., vol. 9, no. 4, pp. 540–552, microsensor net- works," IEEE Apr. 2010. Transactions on Wireless Communications, [7] D. Wei, Y. Jin, S. Vural, K. Moessner, and 1(4):660-669, 2002. R. Tafazolli, “EC supporting materials,” [3] O. Younis and S. Fahmy, “HEED: a hybrid, energy-efficient, Distributed [8] O. J. Boxma and U. Yechiali, “Poisson clustering approach forad hoc sensor processes, ordinary and compound.” networks,” IEEE Trans. Mobile Comput., [9] B. Gong, L. Li, S. Wang, and X. Zhou, vol. 3, no. 4, pp. 366–379, 2004. “Multihop routing protocol with unequal [4] G. Chen, C. Li, M. Ye, and J. Wu, “An clustering for wireless sensor networks,” in unequal cluster-based routing protocol in Proc. CCCM, 2008,pp. 552–556. wireless sensor networks,” Wireless [10] D. Kumar, T. C. Aseri, and R. B. Patel, Networks, pp. 193–207, Apr. 2007. “EEHC: energy efficient heterogeneous [5] S. Soro and W. B. Heinzelman, clustered scheme for wireless sensor “Prolonging the lifetime of wireless sensor networks,” Computer Commun., vol. 32, networks via unequal clustering,” in no. 4, pp. 662–667, Mar. 2009. IPDPS, 2005. 2131 | P a g e