Information Theory Based Intrusion Detection in Wireless Sensor Networks

Sensor networks are used for monitoring purposes in different environments. One of the biggest issues is to keep the network alive as long as possible. Another concern is to keep it safe from attacks. The limitations of sensor nodes make them particularly vulnerable to attacks from adversaries. The most damaging type of attack is Denial of Service (DoS) attack where parts of the network are overloaded with a flood of requests forcing them to deplete their power and die early. In this paper, we introduce a set of metrics by which intruders are identified among the other nodes. This approach is characterized by the fact that identification of intruders is based on the intrinsic behavior that is either harmful or not beneficial to the network. At the same time our approach saves the network power by taking advantage of network redundancy, and query minimum number of nodes without affecting the accuracy of the results. We tested different intruder detection metrics to see if we can accurately find intruders in the sensor network and how early to save the network from damage. Our results show the effectiveness of these metrics in detecting intruders with 100% accuracy and 0 error rate from some of them.