Efficient Sensors Network Management: Global and Local Clustering based on Genetic Algorithms and Fuzzy Logic

Abstract

Energy consumption is one of the major factors to be considered in wireless sensors networks (WSNs). In fact, routing protocols in WSNs are different than those applied in regular ad hoc networks since energy consumption is rarely considered in the latter. Sensors‟ clustering is a class of protocols where sensors are gathered into classes and only a cluster head is responsible for transmission to the base station, thus, reducing energy consumption in regular sensors. The LEACH protocol is a clustering protocol that selects cluster head sensors based on their residual energy. LEACH-GA is a modified version of LEACH where the base station determines sensors‟ clustering based on a genetic algorithm method. In this thesis, we introduce a protocol (GAGLCP) that exchanges roles between regular nodes and cluster heads in a round robin manner following the token ring methodology. The equi-distribution of cluster head burden over all sensors in the same cluster reduces the need of expensive periodic reclustering. In addition, an initial division of the WSN into domains of limited number of sensors is performed in order to reduce the GA search time. The domain memberships of edge sensors are handled through fuzzy logic based on the residual energy. We conducted several simulations of the GAGLCP and the LEACH protocols to test the residual energy and the number of life nods over rounds of clustering. The simulation results showed a better performance of GAGLCP over LEACH in terms of extending the lifetime of a WSN network.