Policy controlled self-configuration in unattended wireless sensor networks

With the decreasing cost and size of sensors, their importance has grown rapidly in the field of unattended network systems, which have a significant and vital role in military and security applications where human access is difficult once the nodes are deployed. A major issue in wireless sensor networks (WSN) is the limited availability of energy supply. The nodes have to cooperate and self-organize to provide an active backbone, making optimal use of energy by putting off the radio of the nodes which are not required. We propose the Policy Controlled Self-Configuration for Unattended Wireless Sensor Networks (PCSSN) scheme for self-organization in WSN. The algorithm consists of two phases: topology discovery and topology maintenance. Based on the concepts of Markov Decision Processes (MDP) we design a policy to activate the optimum number of nodes such that the application fidelity is not affected. The policy is based on the distance between neighboring nodes, the residual energy, and the neighbor count, which are indicators of the hop count, the network lifetime, and the state of connectivity of the network, respectively. We also consider the case of node failures, since failures are quite common in harsh unattended environments. In this paper, we show that our proposed scheme increases the utility function ranging from about 7% to 50% as the number of nodes increases. Further, we show that despite the overhead involved in the proposed scheme, the energy spent is much less than the active case, in which all the nodes stay active forever, without going to sleep.