Extending Lifetime And Optimizing Energy Of Wireless Sensor Network Using Hybrid Clustering Algorithms

Abstract

The Wireless Sensor Network (WSN) is becoming increasingly popular as it enables sensor nodes to measure the surrounding environment, communicate and process measured data. WSN has been directed from military applications to various civil applications, especially in hostile areas. Medical, industrial and smart energy applications is still in need for extensive research due to various challenges encountered. Energy consumption is one of the vital challenges that face WSNs research. The problem is: nodes are supplied with batteries that cannot be recharged or replaced in the field of operation. Management of WSN’s energy helps increasing the network lifetime. Clustering is an efficient technique that is used for enhancing the energy consumed by WSN. However, the dynamic nature of the network made it inappropriate for applying traditional clustering techniques. In this thesis, we investigate the issue of applying Particle swarm optimization (PSO) as a powerful technique which can handle the WSN clustering problem providing a solution that can prolong the network lifetime. This thesis explores the advantages of hybrid clustering approaches to provide efficient and effective clustering technique that co-op with the dynamic nature of the network. Two problems have to be solved to cluster WSN. They are: the number of clusters to be produced and the cluster head (CH) for each cluster. Three approaches are presented. The first approach is a Hybrid K-means PSO clustering approach, ’KPSO’, that clusters the network into predefined number of clusters. K-means searches for the best number of clusters, and then groups the network into the selected clusters. PSO selects the best CH for each cluster. KPSO reduced the complexity on the way we are handling the problem and improved the network lifetime by an order of magnitude compared to the well-known Low Energy Adaptive Clustering Hierarchy protocol, LEACH. In the second approach, PSO task was to solve the whole clustering sub-problems. The second PSO Variable Clustering approach, PSO-VC, provides the optimum number of clusters as well as the best cluster layout. PSO-VC enhanced the network lifetime compared to LEACH and KPSO. The last approach, named KPSO-PSO, is an evolution of the first one. KPSO-PSO added a new PSO phase that perform clustering based on controlling the antenna power and thereby prolong the network lifetime. Experimental results showed that this approach can provide improved WSN lifetime over LEACH, KPSO and PSO-VC. Moreover to investigate the effectiveness of using PSO in the proposed clustering approaches, the same approaches are re-implemented using Genetic Algorithm (GA) instead of PSO. PSO proved to converge to better fitness values and resulted in an enhanced WSN lifetime over GA. Finally, We were able to develop a WSN Clustering Aided Toolbox (WSN-CAT) which can significantly help in simulating various WSN environments and helps exploring many tuning parameters for the proposed approaches.