A Novel Detection Scheme Based on D-S Evidence Theory in Wireless Sensor Networks

Wireless sensor networks (WSNs) is a key technology which is deployed under the assumption that participating sensors are cooperative in forwarding each other´s packets. However, some nodes may behave selfishly by dropping other´s packets or refusing to provide services in an attempt to maximize its throughput with minimum expense. In this paper, we present a novel detection scheme based on Dempster-Shafer (D-S) evidence theory in WSN to detect and isolate misbehavior sensors. However, when the scheme is operated, counter-intuitive results may appear. To overcome this problem, this paper improves the original D-S evidence theory, which defines a new variable to modify the collected evidence before combination and then combines these evidences according to Dempster combination rule. Simulation results show that this scheme can detect and isolate misbehavior sensors effectively and accurately, suppress nodes collusion and improve network performance. Compared with other existing detection scheme, this scheme has more security, robustness and accuracy.