A statistical-based algorithm for event region detection in Wireless Sensor Networks

In this paper a new method is proposed for classifying randomly deployed sensor nodes over an area of interest into distinctive categories. The problem of event region and event boundary detection in Wireless Sensor Networks (WSNs) is addressed. Particularly, analysis is provided for a scenario whereby an area of interest featuring two distinctive phenomena is being monitored with a randomly deployed network of wirelessly connected sensor nodes. Each sensor node in the network is asked to acknowledge whether or not it classifies itself as an event-region node based only on its own environment reading. The key decision factor employed in this approach is the statistical attributes of received signal distribution at each sensor node. Applying this algorithm results in reducing the required bandwidth for transmitting the environmental reading to the base station to be proportional to the size of the event-region. This is opposed to other approaches where the required bandwidth is proportional to the size of the entire network.