Prolonging network lifetime by electing suitable cluster head by dynamic weight adjustment for weighted clustering algorithm in MANET
Vijayanand Kumar, Rajesh Kumar Yadav · International Conference on Computing for Sustainable Global Development · 2016
Mobile Ad hoc network has taken the lead in the field of wireless communication. MANET has supported mobility, scalability and extendibility of the network over the air. Connectivity in wireless medium has several costs for communication. Costs such as life of connections, packets routing, information delay, security over the air and trusted source and receivers needs extra care in communication. Flat topology was not able to support scalability of the mobile nodes in the wireless network. To overcome this, hierarchy topology has been proposed which overcome the scalability problem. One of the hierarchy topology can be consider as clustering. Clustering of mobile nodes solves basic three problems which are (a) expanding the network (b) communication stays within the cluster so that other neighboring cluster remains unawares of the communication and (c) maintenance of routing become much easier. Clustering has mainly two processes which are (a) cluster formation and (b) cluster maintenance. Weighted clustering is one of the clustering schemes and this paper has described the constraints on weights which is fixed and not varying with the dynamics [Degree Difference Nv, Sum of Distances Dv, Mobility (Mv) and Power (Pv) of the nodes] of the mobile node in the network. Weight plays major role to select the best stable cluster head as it defines and support the dynamics of the mobile nodes. Algorithm depends on the minimum weight among the mobile nodes, which become the cluster head. Choosing cluster head is an important task of the clustering which takes in cluster formation and also in maintenance phase as well. This paper describes the dynamic weight adjustments by using method of soft computing. Soft computing has non-deterministic algorithm such as fuzzy logic and neural networks. Weight based clustering algorithm shows the nature of artificial neural network with fuzzy behavior on their node dynamics. Since less computation and fast clustering is goal of any clustering based algorithm, so proposed work is intended to select the best cluster head by choosing the appropriate weights for mobile nodes with less computation overhead. Choosing and selecting intermediate weights impact overall performance in election of cluster head selection process and consequently impact the overall network stability. Weight correction model helps in prolonging network lifetime and reduce load on selected cluster head.