HEAD NODE FAILURE DETECTION AND RECOVERY IN WIRELESS SENSOR NETWORKS

Abstract

Wireless sensor network is an emerging field and has applications in the area of disaster management (like earthquake, land sliding, tsumani etc), military services, battle field investigation, security surveillance etc. Wireless sensor network consists of hundreds and even thousands sensor nodes distributed autonomously to monitor physical or environmental phenomena. Each sensor node has limited computation, storage, sensing, wireless communication and battery power. Sensor nodes are deployed in unattended and harsh environment, hence recharging or replacement of batteries is impossible. Thus, battery constraint is an important challenge in WSN. For energy conservation and prolonging network lifetime various types of network topologies like clustering, tree-based etc are used. Sensor nodes are prone to failure (like energy exhaustion and software or hardware failure) due to hostile and harsh environment. Node failure causes connectivity loss and sometimes partitioning of the network. In a clustered WSN, if a cluster head fails, all member nodes of that cluster get disconnected from rest of the network. It results in total data loss from the region under this cluster. It also causes holes in network topology. Therefore, for normal operation of a cluster and network connectivity, it is important to a recover the failure of a CH within appropriate time and that too with accuracy. In this thesis, we have focused on detection of faulty CH in autonomous WSN and to recovery from it. Our aim is to develop a protocol which detects CH failures with accuracy and performs recovery. For accuracy we have used an agreement protocol in distributed manner, so that each cluster member can make a certain level of decision about failure. Each node participates to reach at final agreement about failure. Once CH failure is confirmed through agreement, an energy efficient new CH election protocol is proposed. This protocol, among other parameters takes residual energy as an essential component to finalize the selection of CH. Our protocol provides a robust and self healing approach that performs well for small to large scale WSNs