Clustering and Event Detection in Wireless Sensor Networks

Abstract

Clustering is a energy saving mechanism in WSN through which, number of group member can save their energy by communicating their information to cluster head only instead if transmitting to far located base station. Clustering is more or less hierarchical network where cluster heads can be cluster member for next leveled cluster heads [2]. The cluster heads perform data aggregation and information forwarding [6,11,12]. Clustering is having advantages of energy saving, more scalability, less load, more robustness, data aggregation/fusion, load balancing and, improved network lifetime and latency reduction [15]. Clustering protocols can be differentiated on number of parameters like some protocols may work in centralized way or some may be distributive, other criteria could be power base, location aware, multilevel and multi-hop inter-cluster communication etc. Centralized algorithms are those in which the base station allocates the cluster heads to sensors. Base station is having whole responsibility cluster heads rotation [13]. In Distributed algorithms cluster heads get choose on basis of mutual selection process among sensor nodes; no center authority ever exists. Power base clustering algorithms decide cluster heads on ground of residual battery life of sensor. Multi-hope inter cluster communication is feature in which information can communicated with the help of intermediate relay node. Some clustering protocol works with sensors which have Location awareness with the help of GPS. Multilevel clustering is representation of hierarchy of cluster heads. Table 1 is shown feature survey of existing clustering protocols. Section II in this paper will present gist of our previously proposed work MEDC and MEHEED along comparison with HEED protocol. This Section Present experimental results MEDC and MEHEED on different values parameters. Simulated result also present performance degradation in some specific cases. Section III will introduce change point detection and show experimental evaluation for fire detection application.