Group Head Selection in WSN using EM Algorithm

Abstract

Recently wireless sensor network has become attractive field of research. A wireless sensor network (WSN) contains plenty of sensor nodes (SNs) that continuously monitor the physical phenomenon. The sensed data is cooperatively transferred through the network towards sink. In large wireless sensor network numerous nodes are deployed together to perform a task. The nodes have sensing, computing and communication features. Since the nodes are constrained in power, storage and computation, so it is necessary to deal with these issues efficiently. Therefore, a greater challenge is involved in saving energy and enhancing the lifespan of the network. Connectivity among the sensor nodes is another critical challenge in WSN due to limited communication range of sensor nodes. As a result, there arises the problem of network disconnection and huge energy consumption. The issues have been addressed in this paper and the proposal contains two phase scheme. In first phase, SGP (Spectral Graph Partitioning) technique is applied for the identification of disconnected portion of the network. Second phase is based on the improved EM (Expectation Maximization) clustering scheme for efficient cluster head and group head selection to diminish the energy consumption in the network.