AN ENERGY EFFICIENT ELECTION PROTOCOL IN MULTI-LEVEL CLUSTERING FOR WIRELESS SENSOR NETWORKS

Abstract

A wireless network with large number of sensors has potential to monitor stimulus around it. Data sensed by these sensors is sent hop by hop to the sink node for processing using low power transceivers. Sensor networks have severe energy constraints, low data rate with high redundancy, and many-to-one flows. Thus, data centric 'mechanisms that perform in-network aggregation of data are needed in this setting for energy-efficient information flow. Clustering is one of the data centric mechanisms in which various cluster heads perform in-network aggregation of data. Thus, there is more load on cluster head than regular nodes. Therefore, for load balancing the role of cluster head should be rotated among other regular nodes. Various schemes propose load balancing mechanisms but concentrates upon probability functions to elect new cluster head and periodically call the election process. In this work "An Energy Efficient Election Protocol in Multi-level Clustering for Wireless Sensor Networks", we propose a dynamic and distributed new cluster head election algorithm based upon two level clustering scheme. This algorithm can be extended to multi-levels in the similar fashion. All sensors are assumed to be homogeneous in nature and working in dual mode i.e. low power mode and high power mode. In the proposed algorithm, if energy level of current cluster head falls below a certain limit, election process is started. Nodes having energy levels higher or equal to a threshold value are candidate nodes for new cluster head election. After each election, node with maximum residual energy is selected as cluster head among these candidate nodes. Algorithm runs locally within each cluster at random time. Moreover, threshold value for selecting candidate nodes get, updated with each election to take into consideration overall network energy depletion with time. Simulation results show significant energy savings when compared with other periodic clustering scheme like Energy Efficient Multi-level Clustering (EEMC