Energy-Efficient Data Mining Techniques for Emergency Detection in Wireless Sensor Networks

Event detection is an important part in many Wireless Sensor Network (WSN) applications such as forest fire, environmental pollution. In this kind of applications, the event must be detected early in order to reduce the threats, damages. In this paper, we propose a new approach for early forest fire detection, which is based on the integration of Data Mining techniques into sensor nodes. The idea is to partition the node set into clusters so that each node can individually detect fires using classification techniques. Once a fire is detected, the corresponding node will send an alert to its cluster-head. This alert will then be routed via gateways, other cluster-heads to the sink in order to inform the firefighters. The approach is validated using the CupCarbon simulator. The results show that our approach can provide a fast reaction to forest fires with efficient energy consumption.