A Robust Cluster Head Selection Method Based on K-Medoids Algorithm To Maximize Network Life Time and Energy Efficiency For Large WSNs

WSNs are utilized in number of fields like healthcare, weather, military, etc. These sensor nodes are battery running devices. Hence, the energy becomes the major issue. A number of researches are being conducted to increase their battery life. Sensing process is the primary process of Sensor nodes. A number of WSN processes are having great possibilities of improvement: 1) Clustering, 2) Routing 3) Routing Data Management. In this paper, we are improving an existing clustering algorithm which is based on k-medoids clustering algorithm. In this paper, we are proposing an effective and efficient cluster head selection method using kMedoids to solve the stated problem, especially for large WSNs consisting of thousands or millions of nodes. k-Medoids is more efficient and correct than k-Means for large clusters. Both of K-Means and k-Medoids utilize expectation maximization (EM) strategy to converge to a minimum error condition. While k-Medoids require the cluster centers to be centroids, in k-Means the centers could be anywhere in the sample space. kMedoid is more robust to outliners than k-Means therefore results in more quality clustering. It is also computationally more complex. Computer simulation will be performed in the NS2 environment and the proposed approach will be compared with LEAH and HEED. INTRODUCTION In the last half century, the computers have exponentially increased in processing power and at the same time decreased in both size. These rapid developments led to large market in which computer based devices are being used more and more in our society’s daily activities. In the last years, computers are reduced in size and becoming so small and so cheap, that single-purpose computers with embedded sensors are almost practical from both economical and theoretical points of view. Wireless sensor networks are the computers which are beginning to become a reality, and therefore some of the long overlooked limitations have become an important area of research. FIGURE 1: Wireless Sensor Network Architecture [LINK:http://www.google.com/url?q=http://www.fi.muni.cz /usr/staudek/vyuka/PA151/07_wpan_zb.ps.pdf&ust=139944 0775533106&usg=AFQjCNHj2reTSqQ0xQ7lDoxx2rvc-