International Journal of Engineering Trends and Technology (IJETT) – Volume 14 Number 5 – Aug 2014 Self-Organizing Clustering Methods for EnergyEfficient Data Gathering in Sensor Networks Sumit Kumar Chauhan#1, Bhagirathi Pradhan#2 #1 Post Graduate Scholar, Department of Electronics and Communication, Dehradun Institute of Technology, Dehradun, India #2 Assitant Professor, Department of Electronics and Communication, Dehradun Institute of Technology, Dehradun, India By deploying wireless sensor nodes and composing a sensor network. One can remotely obtain information about the behaviour, conditions and positions of entities in a region. Since sensor nodes operate on batteries, energy-efficient mechanisms for gathering sensor data are indispensable to prolong the lifetime of a sensor network as long as possible. A sensor node consumes energy in observing its surroundings, transmitting data and receiving data. Especially, energy consumption in data transmission scales proportionally to the nth power of the radius of the radio signal. Therefore, cluster-based data gathering mechanisms effectively save energy. In cluster-based data gathering, since each node can save transmission power and the number of collisions is also reduced, sensor networks can live for longer period. In clustering, however, we need to consider that a cluster-head consumes more energy than the other nodes in receiving data from cluster members, fusing data to reduce the size and sending the aggregated data to base station. In this paper, we synthesis existing clustering algorithms [1] in Wireless Sensor Networks and compare them in terms of their stable operation period(SOP) and highlights the challenges in clustering. Abstract— Various techniques had been proposed to make network more energy-efficient. In this paper, we focuses on balancing energy consumption within the network and also avoid network wide broadcasting of control packets. The rest of the paper is organized as follow: The rate of energy consumption of nodes in a particular region is approximately equal with this the lifetime of nodes in a particular region same and is denoted by Ʈ(i). EC algorithm targets to equalize the lifetime of all regions and to maintain equal energy levels at all regions throughout the lifetime of Wireless Sensor Networks. II. RELATED WORK A. Analysis of Energy consumption in Cluster-based Datagathering To prolong the lifetime of a sensor network, Cluster radii must be carefully determined. For example, if a radius is large a cluster-head consumes much energy in receiving sensor data from its members and sending the aggregated data to the next hoop node. In addition, cluster members consume much energy in sending their data to the distant cluster-head. However, at the same time, since the number of nodes in a cluster increases, a sensor node becomes a cluster-head less Keywords— WSN, Self-Organizing Clusters, LEACH, HEED, frequently. On the contrary, if a radius is small, the amount of MRPUC, EEUC, EC algorithms. energy consumed in data-gathering becomes small at the sacrifice of frequent rotation of the role of cluster-head. In I. INTRODUCTION addition to intra-cluster communications, the distance to a A Wireless Sensor Network (WSN) consists of autonomous, base station also affects the energy consumption of a cluster. self-organizing, lightweight sensor nodes, which can monitor If a cluster is close to a base-station, a cluster-head has to physical or environmental conditions. Wireless Sensor Nodes relay more sensor data from its outside region in multi-hop are the nodes, which can sense, compute and communicate the communication among cluster-heads. In this section, for each data. It is possible only due to miniaturization of various cluster-head to independently determine an appropriate radius component, which is made possible by MEMS technology. A of its cluster, we analytically investigate the relationship sensor node consists of microcontroller, battery, analog to among the energy consumption, cluster radius and the digital converter, sensing device. All these components have distance of a cluster-head to the base-station. their own function. There is some features of WSN which is B. Energy Consumption Model in Cluster-based Datamakes it more reliable for various applications, these factors gathering include fault tolerance, scalability, production cost, hardware To generalize the problem, we consider energy consumed constraint, sensor network topology, environment, in gathering data from cluster members to a cluster-head and transmission media and power consumption. WSN have sending aggregated data to the base station by multi-hop various application like military, environmental, health and transmission among cluster-heads. Therefore, since we do not home applications. The main challenge for WSN is energy consider how clusters are organized, results in this section can consumption. There is lot of energy consumes while be applied to other cluster-based data-gathering methods. We transmitting the data but sensor nodes have limited energy. ignore energy consumed in MAC layer processing in carrier sense, collision detection and retransmission. The energy ISSN: 2231-5381 http://www.ijettjournal.org Page 255 International Journal of Engineering Trends and Technology (IJETT) – Volume 14 Number 5 – Aug 2014 consumed in transmitting and receiving a k bit message at d, m is given in equation 2.1 through 2.3 Etransmit(k,d) = k*(Eelec+efs*d2), if d<do 2.2.1 Etransmit(k,d) = k*(Eelec+emp*d4), if d≥do 2.2.2 Erecevice(k) = k*Eelec 2.2.3 A sensor node consumes Eelec(nJ/bit) in transmitter or receiver circuitry and e(pJ/bit/m2) in transmitter amplifier. The threshold do is introduced to take into account the effect of multipath fading. The total energy Ecluster consumed in one cluster in a single data collection round(DCR) is given by equation 2.4 Ecluster = Eall m-h + Ehead1 + Ehead2 + Ehead3 + Eproc 2.2.4 Eall m-h corresponds to the total amount of energy consumed by cluster members in sending their sensor data to a clusterhead. Eproc is energy consumed in data processing at clusterhead. Ehead1 is the energy consumed by a cluster-head in receiving sensor data from other cluster-heads, which it has to forward toward the base-station. Finally, Ehead3 stands for the energy consumed by a cluster-head in sending the aggregated sensor data to the next hop cluster-head or the base-station. The amount of energy consumed per sensor node, Enode averaged over multiple rounds, where the role of cluster-head is rotated, is given by the following equation. Enode = Ecluster /n 2.2.5 Where n denotes the number of sensor nodes in a cluster. When we assume the uniform distribution of sensor nodes, the following equation hold. n = ρ * Scluster 2.2.6 where ρ is the density of the sensor nodes and Scluster corresponds to the area from which a cluster-head gathers sensor data. In this , we consider a rectangular monitoring region of width W and length X. Base-station is located outside the network. We assume that sensor nodes are uniformly distributed in the monitoring region with density ρ. C. A Multi-hop Data Collection Protocol for WSNs In this section, an energy-efficient multi-hop data routing solution for WSNs organized as clusters is briefly outlined. We will use this routing protocol for clustering solutions we are going to compare. The routing algorithms is based on two ideas : First, Reducing the overhead in route discovery, and Second, Balancing energy consumption among all CHs. To achieve these goals a simple scheme is used and is known as reactive routing algorithms. The network region is divided into small size regions as shown in figure below. Where nodes at different hop distances to the sink are denoted by different symbols. Figure 1 Hop distance to the sink and rectangular regions The area in which nodes of a particular hop distance i reside can approximately be represented by a rectangular ISSN: 2231-5381 region Ri. The widths of these regions may not be equivalent and are random variables depending on the node locations and sensor communication range. However, we can denote the average region width by a. A CH node in region R i chooses its next hop towards the sink in the neighbour rectangular region Ri-1. The CH transmits a route request packet with a range of (W2+4a2)1/2, sufficiently large to cover R i-1. Each receiving CH in Ri-1 generates a reply packet and starts a route reply timer with an expiration time inversely proportional to its residual energy level. The first node that has an expired timer actually makes the transmission of a route reply packet back to the requester CH in Ri, while the rest quietly cancel their timers upon hearing this reply. This guarantees that a single reply packet is sent and thus prevents excessive message overhead. Furthermore, by considering the residual energy levels, priority is given to nodes with higher resources, a policy towards balancing energy consumption in the entire network. III. CLUSTERING ALGORITHMS FOR WSNS A. Low-Energy Adaptive Clustering Hierarchy(LEACH) LEACH[2] is a protocol architecture for sensor networks that combines the ideas of energy-efficient cluster-based routing and media access together with application specific data aggregation to achieve good performance in terms of system lifetime, latency and application perceived quality. The operation of LEACH is divided into rounds. Each round begins with a set-up phase when the clusters are organized, followed by a steady-state phase when data are transferred from the nodes to the cluster head and on to the BS, as shown in fig 2. Set-up Steady-state Frame Round time Figure 2 LEACH cluster head selection and data transmission process 1. Cluster dead selection process of LEACH : First of all, each node will calculate its CH probability for the current round r +1 and is given by and this probability is so chosen so that the average number of cluster heads for this round is equal to k. where the value of k can be determined analytically or through simulation. E[number of cluster heads] = *1=k 3.1.1 Where N is the total number of nodes in the network. Where, k= Where, is the distance between CH and base station. LEACH selection of cluster heads for round r+1 depends on the information from most recent (r mod ( )) rounds , where nodes which have become cluster heads in the most recent (r mod ( )) rounds are not eligible to become cluster head for round r+1 in this way the LEACH can uniformly distribute the load among all the nodes in the network uniformly. Here we are using the decision function which will decide that node can become cluster head for the http://www.ijettjournal.org Page 256 International Journal of Engineering Trends and Technology (IJETT) – Volume 14 Number 5 – Aug 2014 current round or not. If the value of it means the node have become the cluster head in the most recent (r mod ( )) and is not eligible for this round as given in equation 3.1.2. 3.1.2 By using probabilities of becoming cluster head of nodes given by equation 3.1.2 leach ensures that each node can become cluster head al least once in rounds. Where represents the total number of nodes in any cluster. After rounds all nodes will become eligible to become cluster heads. The total number of nodes which are eligible to become cluster head for round r+1 are : E[ ] = N-k* 3.1.3 The average number of cluster heads generated for round r+1 is given by the following expression. E[no. of CHs] = *1 = N-k* =k 3.1.4 The expression for probability of becoming cluster head in equation 3.1.2 assumes that all nodes have equal energy when the network is deployed but this is not always the case. The probability of becoming cluster head when nodes start with different level of energies can be given as. 3.1.5 Where, is the residual energy of node and the total energy in the network. is 3.1.6 Now, the average number of cluster heads can be calculated as: E [no. of CHs]= *1= 3.1.7 2. Cluster formation process for LEACH : Now in this process each non-CH node will associate with its closest cluster head based on received signal strength indicator. For this purpose each CH node first broadcast a CH-ADV packet throughout the network. Each non cluster head node will receive all CH-ADV packets and will associate with the cluster head which generates the highest received signal strength. To associate to cluster head the node will send a CHASSO packet to the cluster head to associate with this message contains the cluster head id and node id. After the formation of clusters data transmission takes place. B. A HYBRID OF ENERGY AND COMMUNICATION COST(HEED) HEED[3] protocol considers residual energy of nodes as well as intra-cluster communication cost in finding final cluster heads, so it generates more balanced and energy efficient clusters. ISSN: 2231-5381 1. Cluster heads selection in HEED: The cluster head selection process in HEED occurs in number of iterations. First every node elects itself to be cluster head with probability which is not allowed to fall below a predefined threshold and is given by equation 3.2.1.the value of this threshold is inversely proportional in the initial sensor energy. = × 3.2.1 Where, is the initial percentage of cluster heads to be selected as tentative cluster heads this value is choose to limit the initial number of CH –broadcasts and does not create any impact on the quality of final cluster heads. is the maximum energy of sensor node which is equal to fully charge battery. Now nodes which have greater then will become tentative cluster heads and will broadcast a CH-ADV packet in its cluster range. The cluster range is predefined and is optimally chosen through simulation. Each non-CH node as well as CH which have tentative status will receive this advertisement message if they fall in cluster range. It may be possible that a node may receive two or more CH-ADV packets in that case the node will associate to cluster head that results in minimum cost and become cluster member. Now all nodes will double their in the next iteration and again sent CH-ADV packet. A tentative cluster head node can become a member node if it finds a CH with minimum cost. The iterations will continue until we find final cluster heads and all non-CH nodes become covered. A tentative cluster head will be considered final cluster head if its reaches a value of 1.The communication cost in HEED depends on node degree and further depends on the type of clusters. C. AN UNEQUAL CLUSTERING PROTOCOL FOR WSN’S(MRPUC) This method of clustering generates clusters with unequal size to resolve the problem of HOT-SPOT which arises in areas which are close to sink. As nodes in these areas will void of Energy resources more quickly as compared to nodes which are far away from base station because these nodes have to relay their own traffic as well as the traffic coming from outer regions of the network. This method of clustering generates clusters with well distribution over the network. And cluster sizes will be smaller near the sink region. 1. Cluster head selection process of MRPUC[4],[8]: First of the base station will broadcast a BS-ADV packet which is received by all nodes in the network. Now based on the strength of this received BS-ADV message each node will calculate its approximate distance to base station. With the help of this distance each node will calculate its cluster range with the help of equation 3.3.1. 3.3.1 Where and are the maximum and minimum cluster radii which are predefined and their optimal values can be calculated through simulations. is the distance of http://www.ijettjournal.org Page 257 International Journal of Engineering Trends and Technology (IJETT) – Volume 14 Number 5 – Aug 2014 node i to base station and is the maximum distance between sensor node and base station. Now after the calculation of these cluster ranges by every node each node will broadcast a node-ADV message this message contains node cluster range, node id and residual energy. The range of this node-ADV massage is same for every node and is equal to to ensure that each node must receive at least one node-ADV message. At the end of this broadcasting process each node will have a table known as neighbour information table as shown in table 1. Node ID Residual Energy Node State 1 1.2 UNKNOWN ….. ….. ….. Table 1 showing neighbour information Now each node will look at its table and will decide that it can become a cluster head for the current round or not. If node have residual energy which is higher than the residual energies of its neighbour which have entries in its table than the node can become cluster head for this round and sends a CH-ADV packet to all its neighbouring nodes the neighbouring nodes will quit the cluster head competition immediately and become normal nodes. 2. Creation member selection process : After the selection of cluster heads, each cluster head will transmit a CH-ADV packet in its cluster range, this packet contains residual energy of cluster head and its ID. each non-CH node will construct a table of CH’s as shown in table 3 if the node lies in the range of corresponding CH. node j will add cluster head i If d(i,j)< . Cluster Head ID Distance to it d Residual Energy E 1 70 1.9 ….. ….. ….. Table 2 list of candidate CH’s After the creation of CH’s table the node will decide to attach with the cluster head which will result in minimum cost. Where the cost function is given by equation 3.3.2. 3.3.2 Where Where is the cost to join cluster head with cluster range . is the distance between node j and cluster head k. maximum CH energy from candidate cluster head set and is the weighted factor which provides trade-off between residual energy of cluster head and distance from it and it must be optimally selected to generate good clustering hierarchy. To associate to the chosen cluster head the node will send a CH-ASSO packet to CH this packet contains node ID and CHID. Now at the end of cluster formation process nodes which are neither CH’s nor cluster members will choose a role of cluster head for themselves. After the cluster formation process the data transmission begins in a multi-hop fashion towards the sink. ISSN: 2231-5381 D. AN ENERGY-EFFICIENT UNEQUAL CLUSTERING PROTOCOL FOR WSN’s(EEUC) This method of clustering[7],[9] is quite similar to MRPUC the only difference is that it does not perform broadcasting by all nodes for cluster head selection that’s why this method of clustering is more energy efficient as compared to MRPUC. This method of clustering produces well distributed clusters as compared to HEED and LEACH and also targets to achieve energy equalization among sensor nodes. 1.Cluster heads selection in EEUC : As in MRPUC each node in EEUC will calculate its approximate distance from base station. After the calculation of distances an initial set of tentative cluster heads will be chosen and we take this initial percentage of tentative cluster heads to equal to T. the value of T is predefined and its value is chosen to limit the initial number of tentative cluster heads. After the selection of tentative cluster heads each node will calculate its cluster range through equation 3.4.1. 3.4.1 Where is the maximum competition radius its value is predefined and can be determined through simulation. is clustering parameter and its value varies between 0 and 1 the effect of on clustering. are maximum and minimum distance between sensor nodes and base station. After the selection of tentative cluster heads each tentative cluster head will broadcast a TCH-ADV packet in its competition range this TCH-ADV packet contains the TCH residual energy, cluster range and ID. Each TCH node will construct a table of its neighbour TCH’s. Two TCH’s are said to be neighbours if the lie in each other’s competition range. After the creation of neighbour tentative cluster heads table each TCH will look at its table if its energy is greater than all TCH in its table it will become CH for this round and send a CH-ADV packet. The TCH receive this CH-ADV packet will quit the CH-competition immediately and transmit a QUIT message to all its tentative CH neighbours. Upon receiving the QUIT message from any TCH the node will remove the entry of the corresponding TCH from its table. 2.Cluster member’s selection phase for EEUC : After the formation of clusters in EEUC cluster head will transmits a CH-ADV message its cluster range each non-CH node will join the cluster head based on its distance from the cluster head. A non- CH node will choose the closest CH to associate with by transmitting a CH-ASSO message. After cluster formation the data transmission begins in multi-hop fashion. E. ENERGY EFFICIENT CLUSTERING(EC) EC[6] is another unequal clustering algorithm. The algorithm begins with dividing the network into number of regions where the length of regions are random variables here we are considering the average length for each region and that is a as shown in fig 3. http://www.ijettjournal.org Page 258 International Journal of Engineering Trends and Technology (IJETT) – Volume 14 Number 5 – Aug 2014 Figure 3 EC divides network area into different regions 1.Cluster heads selection phase : The algorithm begins with selecting tentative cluster heads among all sensor nodes. The percentage of tentative cluster heads is predefined and is T in our case, further the probability of become tentative CH depends on the node residual energy and average energy in the network so the probability of becoming tentative cluster head can be calculated as =T for node j. The main aim of the algorithm is to equalize the lifetime of regions and maximize this lifetime. First, the algorithm calculates the probability of becoming cluster head for every region and is given by equation 3.5.1 Where is the number of cluster heads in region i. The probability of becoming cluster head can be calculated as by approximating the cluster regions with circles. 3.5.2 Where is the cluster range for region i. Now the lifetime of any region i can be calculated as = , where is the energy consumption in single data collection round in region . Now the aim of the algorithm is to calculate this lifetime value for every region and equalize them to a maximum value as shown in equation 3.5.3 and determine the corresponding probability values. 3.5.4 Now we have to calculate the probability values for each individual region with the help of equation 3.5.4. To calculate the probability values first of all we will calculate the probability value for the region where is the total no of regions. The probability value of region is independent of the probability values of all other regions in the network and hence, this region has to transmit only if own packet. So with the help of we can calculate .then with the help of we can calculate and so on. the value of lifetime is iteratively increased until we start to get the negative and imaginary values for . the iterations of increasing L is stopped at largest possible value where we get the positive and real values for the probability values at different node density settings is shown in fig 4. ISSN: 2231-5381 Figure 4 probability values for different regions at different node density settings After the calculation of values each node will calculate its cluster range and each tentative cluster head will transmit a CH-ADV message this packet contains the ID of cluster head and its residual energy in its cluster range. if A tentative cluster head receives a CH-ADV message which reports higher residual energy this tentative cluster head will become normal node. If a tentative cluster head does not reports any CH-ADV message which has higher residual energy then that node will become final CH. 2. Cluster member selection process for EC : Each final cluster head will broadcast a CH-ADV message in its region. Normal nodes will receive these broadcast messages and will join the closest cluster head by measuring the received signal strength and associate with the cluster head by transmitting a CH-ASSO message and upon reception of CH-CONF message from the corresponding cluster head. This unequal clustering algorithm is the most energy efficient as compared to EEUC AND MRPUC unequal approaches. Because EC does not assume region wide broadcasting which saves in energy in cluster formation process, further it focuses on energy equalization of regions and determine corresponding probability values for becoming CH’s. Now, the data transmission occurs in a multi-hop fashion in the network. IV. ADVANTAGES AND DRAWBACKS A. Advantages : With the help of Clustering we can- reduce energy consumption, prolong the network lifetime, cluster the whole network with selected CH, rotate CHs for energy distribution, Increase bandwidth reuse and thus increases capacity of the network, Increases scalability of the network and developers may benefits because these designs usually reduce the cost of site development and increase the market price of individual plots in comparison with traditional subdivisions. These design can benefit rural areas by reinforcing the policy of maintaining the local rural character that is included in many comprehensive land use plane. B. Drawbacks : Perhaps most important, local officials, developers and the community may be predisposed toward. Traditional development designs because they are familiar and well understood. An education effort may be necessary to help these groups understand the goals and advantages of cluster development. During the planning phases, lot and home layout may take extra work to ensure that while homes are – Cluster/conservation development land use planning. Local http://www.ijettjournal.org Page 259 International Journal of Engineering Trends and Technology (IJETT) – Volume 14 Number 5 – Aug 2014 community located closer together, they still take advantage of the open-space goals of the design. Methods to protect and maintain the open space must be carefully developed, implemented and monitored. Although not necessarily a restricting disadvantages the management of waste water must be carefully, designed for smaller lots. While these disadvantages should be acknowledges and addressed, none should preclude the use of cluster development. V. CONCLUSIONS Clustering is used for topology control in the network, where topology control is a mechanism to save energy and increasing scalability of the network. With the help of clustering we can make wireless sensor network energy efficient. In this paper we reviewed some existing clustering algorithms. Some of which are equal clustering algorithms and some are unequal clustering algorithms. We see from simulation results that unequal clustering algorithms shows better performance as compared to equal clustering algorithms in multi-hop data collection scenario. Equal size clustering algorithms suffers from HOT-SPOT problem of the network which is resolved by unequal size clustering. From simulation results we can see that EC out performs all clustering algorithms such as LEACH,MRPUC,EEUC and HEED and shows better performance in different network scenarios. In this paper we discussed some challenges faced by existing clustering algorithms. And these challenges motivates for the discovery of new clustering approaches. [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] G. Chen, C. Li, M. Ye, and J. Wu, “An unequal cluster-based routing protocol in wireless sensor networks,” Wireless Networks, pp. 193– 207, Apr. 2007. C. Cheng, Chi K. Tse, and C. M. Lau, “A Clustering Algorithm for Wireless Sensor Networks Based on Social Insect Colonies,” IEEE sensors journal., vol. 11, no. 3, pp. 711-721, 2011. S. Soro and W. B. Heinzelman, “Prolonging the lifetime of wireless sensor networks via unequal clustering,” in IPDPS, 2005. M. Perillo, Z. Cheng, and W. Heinzelman, “An analysis of strategies for mitigating the sensor network hot spot problem,” in Proc. Second Annual International Conference on Mobile and Ubiquitous Systems: Networking and Services, 2005, pp. 474–478. T. Rappaport, Wireless Communications: Principles & Practice. Englewood Cliffs, NJ: Prentice-Hall, 1996. I. F. Akyildiz, W. Su, Y. Sankarasubramaniam, and E. Cayirci, “A Survey on Sensor Networks”, IEEE Communications Magazine, vol. 40, no. 8, pp. 102-114, 2002. S. Lindsey and C.S. Raghavendra, “PEGASIS: Power-Efficient Gathering in Sensor Information Systems,” Proc. Int’l Conf. Comm. (ICC ’01), 2001. O. Younis, M. Krunz and S. Ramasubramanian, “Node Clustering in Wireless Sensor Networks: Recent Developments and Deployment Challenges”, IEEE Network Magazine, May, 2006. L. M.Arboleda C, and N. Nasser, "Comparison of Clustering Algorithms for Wireless Sensor Networks," Electrical and Computer Engineering, 2006. CCECE '06. Canadian Conference, 2006, pp. 1787 - 1792, May 2006. M. Ye, C. Li, G. Chen, and J. Wu, “EECS: an energy efficient clustering scheme in wireless sensor networks,” in Proc. IPCCC, Apr. 2005, pp. 535–540. ACKNOWLEDGMENT To present this paper I would like to thanks Assistant Prof. Bhagirathi Pradhan for his continuous motivation and guidance. REFERENCES [1] [2] [3] [4] [5] [6] [7] [8] A. A. Abbasi and M. Younis, “A survey on clustering algorithms for wireless sensor Networks,” Computer Commun., vol. 30, pp. 2826–2841,2007. W. B. Heinzelman, A. P. Chandrakasan, and H. Balakrishnan, “An application- specific protocol architecture for wireless microsensor networks,” IEEE Trans. Wireless Commun., vol. 1, no. 4, pp. 660– 670,Oct. 2002. O. Younis and S. Fahmy, “HEED: a hybrid, energy-efficient, distributed clustering approach for ad hoc sensor networks,” IEEE Trans. Mobile Comput., vol. 3, no. 4, pp. 366–379, 2004. B. Gong, L. Li, S. Wang, and X. Zhou, “Multihop routing protocol with unequal clustering for wireless sensor networks,” in Proc. CCCM, 2008, pp. 552–556. D. Wei, Y. Jin, S. Vural, K. Moessner, and R. Tafazolli, “EC supporting materials,” http://epubs.surrey.ac.uk. D. Wei, Y. Jin, S. Vural, K. Moessner, and R. Tafazolli, “An EnergyEfficient Clustering Solution for Wireless Sensor Networks,” IEEE Trans. Wireless communications., vol. 10, no. 11, pp. 3973- 3983, 2011. C. F. Li, M. Ye, G. Chen, and J. Wu, “An Energy-Efficient Unequal Clustering Mechanism for Wireless Sensor Networks,” IEEE International Conf. Mobile Adhoc and Sensor Systems, pp. 8, Nov. 2005. S. Lee, J. Lee, H. Sin, S. Yoo, S. Lee, J. Lee, Y. Lee, and S. Kim, “An energy-efficient distributed unequal clustering protocol for wireless sensor networks,” in Proc. PWASET, Dec. 2008, pp. 1274–1278. ISSN: 2231-5381 http://www.ijettjournal.org Page 260