Split and Merge LEACH Based Routing Algorithm for ... - IJCNIS

3 downloads 0 Views 722KB Size Report
Keywords: LEACH, Split, Merge, WSN, Energy, Algorithm. 1. Introduction. Wireless ... network life time to LEACH-F. Experimental results show that our algorithm ...
155 International Journal of Communication Networks and Information Security (IJCNIS)

Vol. 10, No. 1, April 2018

Split and Merge LEACH Based Routing Algorithm for Wireless Sensor Networks M. Zayoud1, H. M. Abdulsalam2, A. Al-Yatama3, S. Kadry4* 1

Department of Engineering, American University of Middle East-Kuwait State, P.O. Box 220 Dasman, 15453 Kuwait. 2 Dapertment of Information Sciences, Kuwait University- Kuwait State, P. O. Box 5969, Safat 13060, Kuwait. 3 Department of Computer Engineering, Kuwait University- Kuwait State, P. O. Box 5969, Safat 13060, Kuwait. 4 Department of Mathematics and Computer Science, Beirut Arab University, Lebanon. *corresponding author

Abstract: Hierarchical routing and clustering mechanisms in Wireless Sensor Networks (WSNs) help to reduce both the energy consumption, and the overhead that is created when all the sensor nodes in the network are sending information to the central data collection point or base station. LEACH (Low Energy Adaptive Clustering Hierarchy) is one of the most well-known energy efficient clustering algorithms for WSNs. In this paper, we extend the LEACH protocol to (LEACH-SM) protocol by introducing a Split and Merge stage to improve the performance and robustness. We consider the following design aspects: non-uniform distribution of sensors, cluster re-formation by splitting or merging clusters conditionally, routes maintenance, and nodes mobility. OPNET, a well-known simulator tool, is used to simulate LEACH-SM in order to evaluate the performance of the proposed protocol. Simulation results and comparisons with existing protocols show the effectiveness and strength of the proposed protocol in terms of enhancing the lifetime of the whole sensor network, where sensors are either static or mobile with low speed. Keywords: LEACH, Split, Merge, WSN, Energy, Algorithm.

1. Introduction Wireless Sensor Network (WSN) is a network that consists of sensors, deployed over a geographic area either randomly or in pre-deterministic distribution. Sensors get data records that are related to any phenomena and forward those to a central unit called the base station for processing and analysis. Examples of data that can be collected are: temperature, humidity, light conditions, seismic activities, etc. WSNs enable many new and exciting applications in both military and civilian environments [2]. Routing in WSNs is very challenging due to the relatively large number of sensor nodes and limited computational power, memory, and battery power in the sensor [3]. This fosters large endeavors in industrial investments on this field, standardization process and research activities [4-23]. Scholars have provided in-depth discussion on different clustering protocols in wireless sensor networks that are made up by communicating sensor nodes to gather and elaborate information from real world in a distributed and coordinated way in order to deliver an intelligent support to human activities [18-39]. Holistic approaches to evaluate energy efficiency and improve the global energy productivity through the use of high-performance and energy-efficient networks, services and applications are needed. [24]. Designing a reliable WSN requires many efficient methods in routing, data aggregation and localization, and due to the failure of nodes may leave some areas uncovered and degrade the fidelity of the collected data. Therefore, establish

a fault-tolerant mechanism is very crucial [40]. In this work, we only consider the problem of data routing using clusters to improve the network lifetime. Cluster-based routing algorithms for WSNs are algorithms by which nearby sensors are grouped together to form a number of groups (clusters). Each cluster is represented by one sensor called Cluster Head (CH). CHs collect data records from other sensors in their clusters and send them to the base station. Energy is, therefore, saved since not all sensors communicate with the base station for further processing. LEACH (Low Energy Adaptive Clustering Hierarchy) by Heinzelman [5] is one of the most famous, efficient and widely used clustering algorithm for WSNs utilizing homogeneous randomly deployed sensor nodes. It is a distributed cluster formation algorithm. All nodes in LEACH have a chance to become cluster heads (CH) at some point, in order to balance the energy that is consumed in one round. Extensions of LEACH that are found in the literature include LEACH-F [8], LEACH-C [9], LEACH-GA [10], LEACH-M [11], K-LEACH [12], Q-LEACH, S- LEACH [13], MultiHop LEACH [14], and W-LEACH [15, 16]. Aslam et at. [17] present a survey of some of the extended LEACH Protocols. We consider one particular derivation of LEACH, namely LEACH-F [8]. In LEACH-F, only the cluster heads are rotated, while the other cluster members are computed at the initial operation of the network and remain fixed. We introduce LEACH with Split and Merge (LEACH-SM), a new cluster-based routing algorithm for WSNs, to handle mobility and extend the network lifetime. We base our work on LEACH-F. We mainly target to address the shortcoming of LEACH-F, such that LEACH-SM is able to handle WSNs of un-balanced distribution of the sensor nodes either at deployment stage, or because of the change of the density in some clusters. It is also able to handle mobility of sensors. The key idea of LEACH-SM is that LEACH-SM is based on LEACH-F in defining initial fixed clusters, then in the case of the need to change the clusters structure, the algorithm only splits or merges the existing clusters to define the new clusters’ configuration instead of redefining the whole clusters structure to overcome the overhead of structure formation at each round as in LEACH. We simulate our algorithm using OPNET and compare the network life time to LEACH-F. Experimental results show that our algorithm performs well when applied on both nonuniform and evolving networks. The remainder of the paper is organized as follows: Section 2 shows some background and related work, section 3 gives an overview of the LEACH-SM routing protocol. And detailed

156 International Journal of Communication Networks and Information Security (IJCNIS)

characteristics of LEACH-SM. Section 4 states our experimental settings. Section 5 presents numerical results, and Section 6 concludes our work. Finally, future work and references.

2. Background and Related Work 2.1 LEACH algorithm Low Energy Adaptive Clustering Hierarchy Aggregation (LEACH) algorithm by Heinzelman [5] is a data aggregation algorithm based on cluster routing. The algorithm works in rounds such that each round has two phases namely, a setup phase and a steady state phase. In the setup phase, p% of n sensors are uniformly randomly chosen to be cluster heads (CHs) based on a threshold

Vol. 10, No. 1, April 2018

the fixed clusters structure. 2.4 Simulated Annealing algorithm Simulated Annealing algorithm (SA) is a probabilistic metaheuristic, that aims to find a global optimum in a large search space. It searches for the optimal solution by transiting from a current solution x to a neighborhood solution y using the following acceptance probability: (2) Where Ci is a controlling parameter. In SA algorithm, there is a decrease observed in the value of control parameter from an initial large value to a small final value. As per the supposition, sequence of Ci can be written (C1, C2… Ci, Cn), with n being the total iteration number of the algorithm. 2.5 Clusters formations

(1) where p is the desired number of CHs, t is the current round, and G is the set of nodes that have not been CHs in the last 1/p rounds. This ensures that a sensor that is chosen to be CH is not chosen in the next rounds until all other sensors in the network become CHs. This feature increases the lifetime for sensors since it ensures fair energy consumption. The algorithm chooses the CHs uniformly randomly, hence, it does not consider non-uniform networks. After all CHs are chosen, clusters are dynamically defined such that each nonCH becomes a member of the cluster with the nearest CH. In the steady state phase, each CH collects data from all sensors in its cluster based on Time Division Multiple Access (TDMA). CHs, then, compress the collected data and send it to the base station. 2.2 LEACH-C algorithm LEACH-C [9] works exactly like LEACH expect that it assumes centralized CH election, where each sensor sends information about its location to the base station at the beginning of each round, then the base station uses an optimization algorithm, such as Simulated Annealing (SA), to decide which clusters are to become CHs. The CHs are chosen based on their locations and their remaining energy such that clusters with more energy are candidates to be CHs. This gives a generally better distribution for CHs, however, it may reduce sensor lifetime due to the increase of communication between the sensors and the base station. 2.3 LEACH-F algorithm LEACH-F [8] is a centralized algorithm that assumes fixed clusters while only rotating the CHs for each cluster. This reduces the setup phase overhead since clusters are formed only once which means that there is no set-up overhead in the initial phase of each round. It, however, may force a node to stay in a cluster with a CH that is further than a CH of a nearby cluster. In order to initiate a cluster formation, LEACH-F also employs the same simulated annealing algorithm that is used in LEACH-C. However, LEACH-F is more energy efficient in contrast to LEACH-C. Yet, LEACHF, does not handle mobility, adjust its behavior when nodes are dying, or allow addition of new nodes in the system. Furthermore, in F-LEACH, the nodes might require a large amount of energy in order to communicate with their CHs in the case of having their CHs far away from them according to

Cluster based routing is an efficient method for the provision of WSNs lifetime. Thakkar and Kotecha [29] described a solution in which Grid based method was implemented for the derivation of cluster formation. The technique emphasizes cluster head election method such that it is decentralized and uses Bollinger Bands. This pattern is referred to as a realistic topology, which is mainly a result of commonly practiced deployment methods [37]. Li, Qian, and Dai [34] propose a Code Dissemination Protocol. They use a topology that is grounded on LEACH algorithm. The main idea is to support remote code update technology. The findings show that the model is able to fulfill requirements of low energy consumption. ZORO-MSN [20] is a fixed zone-based partition scheme. Clusters here are presented as square zones, cluster head is presented as zone head (ZH), and the nodes are mobile. ZORO-WSN handles the mobility of the nodes but the zones or clusters are fixed, and they are in square forms rather than random. Power-Efficient Gathering in Sensor Information Systems (PEGASIS) [6] is a near optimal chain-based protocol. The underlying principle of this protocol is that in order to extend the lifetime of a network, it is crucial that the nodes converse only with the closest neighbors, while they take turns for communicating with the base stations.

3. LEACH-SM Routing Protocol Overview Similar to LEACH and LEACH-F, the basic idea of LEACHSM algorithm is to organize the network into clusters based on the distances between the nodes and the remaining energy of each node. However, nodes are homogeneous stationary or mobile. All nodes have a chance to become cluster heads at some point, in order to balance the energy spent per round by each sensor node. The cluster heads for each cluster are selected randomly and in a rotary scheme based on their energy load. Nodes join a cluster by depending on its location to ensure that communication with the cluster-head node requires the lowest amount of transmit power and to ensure minimum inter-cluster interference. LEACH-SM network operation has four phases: initial setup phase, split/merge phase, cluster-head election phase and data transmission phase, see Figure 1. The initial setup phase, the cluster-head election phase, and the data transmission phase are similar to LEACH-F. The main contribution of this

157 International Journal of Communication Networks and Information Security (IJCNIS)

paper is the introduction of split/merge phase that improves network lifetime and handles mobile nodes. Startup

Cluster Formation

CHs Election

Data Transmission

No

Need To reform cluster?

Yes

Split/Merge

End

Figure 1: LEACH-SM network operation phases In the initial setup phase, all nodes send energy and location information to the base station at the network startup. Based on this information, the base station can optimally form clusters since it has a global network view. For forming the clusters, the base station appoints a fixed number of nodes as cluster heads, and evenly distributes the number of nodes in each cluster. The base station uses the Simulated Annealing algorithm (SA) for forming the clusters, since SA finds the near optimal cluster formation. In the case of the need to change the clusters structure, the algorithm only splits or merges the existing clusters instead of redefining the whole clusters structure in the split/merge phase. This phase is explained in details in below sections. The cluster heads are selected in the cluster-head election phase in which the BS assesses the energy and the coordinates of the nodes then computes the current configurations score of the clusters which means the remaining energy level in the sensors assessed by the BS, and compares it to the previously computed score. When this score changes, this means we need to re-form the clusters, then the BS triggers to apply split/merge clusters to get a better cluster configuration. The remaining energy in each sensor indicates the score of the current cluster. If this random number is less than a threshold value, then the node with highest energy level becomes a cluster-head for the current round. The threshold value is calculated based on an equation that incorporates the desired percentage to become a cluster-head, the current round, and the set of nodes that have not been selected as a cluster-head in the last rounds. Finally, in the data transmission phase, the actual data is transferred to the base station. The duration of this later phase is longer than the duration of the combined other phases. Hence, the overhead of all earlier phases can be negligible.

3.1

Vol. 10, No. 1, April 2018

Detailed characteristics of LEACH-SM

The design of the LEACH-SM algorithm is based on specific characteristics, such that the following issues are handled:  The base station is assumed to continuously supervise the energy levels of the nodes and their coordinates.  After a period called UPDATE PERIOD (UPTR) the base station assesses the energy and the coordinates of the nodes. It then computes , where score is a numeric value that gives an indication of the optimal clusters configuration. The initial values of these scores are selected randomly by each sensor in the cluster.  When is greater than a threshold, then the base station is triggered to implement splitting or merging clusters to reach a better cluster configuration from the current one. The configuration of the clusters changes by splitting a cluster into two or more clusters or merging two or more clusters to become one cluster.  After the cluster configuration is established, the algorithm selects from all the nodes of the network some nodes to be assigned as cluster heads one by one, such that each node is assigned an integer value that refers to the score of the node of being a cluster head. The score is calculated based on the distance between the node and the base station and the energy level of the node. If the score values is 1, then this node is the best node to become a cluster head. Every increase in the integer value means that this node is less in score.  Cluster heads are then assigned, such that the node with the best score within each cluster is assigned to be the cluster head of this specific cluster.  The algorithm then continues to perform as any regular cluster-based routing algorithm by gathering data records at cluster heads and then sending them to the base station. 3.2 Clusters formation phase As mentioned earlier, LEACH-SM clusters’ formation is based on SA algorithm in order to compute the optimal cluster configuration. It aims to optimize the energy consumption of the nodes by forming clusters, in which the distances between their members are optimal. For clusters formation, the algorithm first selects a set of C distinct nodes as initial inputs for the SA algorithm. Then, the score of the specified nodes is calculated as shown in algorithm 1. By the means of SA, iterations are then carried out to find the optimum set of CHs based on the score values. 3.3 Split/Merge Phase In this phase, the base station monitors the coordinates and energy levels for all nodes in each round, and calculates the value of . If , then the base station explores if merging or splitting clusters can produce better score based on the remaining energy and distances between the nodes using SA algorithm to search for the best cluster configuration depending on the current existed clusters. When the computation is ready, reforming the sensor nodes into a new optimal clusters configuration takes

158 International Journal of Communication Networks and Information Security (IJCNIS)

place in the WSN. Figures 2 (a), (b), and (c) represent how LEACH-SM works in the field. The frequency of this phase is based on the value of UPTR. In Figure 2(a), cluster C5 on the left splits into C5 and C6, while in Figure 2(b) clusters C4 and C5 on the left merge into C4. Figure 2(c) show that cluster C3on the left splits into C3 and C4, while clusters C4 and C5 on the left merge into C5.

Vol. 10, No. 1, April 2018

Where p denotes the exchange parameter for partial results, and n denotes the number of nodes in the network. Compute optimal cluster configuration algorithm (Algorithm 1) Compute_optimal_cluster_configuration() { Int prev_CL_number; //defines the number of clusters computed previously Int previous_score; //defines the score of the clusters computed previously Int recent_score; //defines the score of the clusters computed recently // get the coordinates of the nodes of the network Get_all_the_coordinates_of_the_nodes() // get the coordinates of the nodes of the network Get_all_the_energy_levels_of_the_nodes() If(previous_score!=recent_score) { // get the optimal cluster configuration. It is the one having // the minimal score produced by the function // “compute_optimal_clusters_with_sim_ann” that use simulated annealing // produce a configuration with a given number of cluster (i in our case) for(i=1;i

Suggest Documents