Energy Efficient Intrusion Detection System for Wireless Sensor Network Using Neural Network

In wireless sensor network (WSN), the existing intrusion detection techniques rarely consider energy consumption for judging the node state. Also there is need for increased detection rate. Hence in this paper, we propose an energy efficient intrusion detection system for WSN. Initially the nodes deployed in the network are analyzed for malicious activity based on their residual energy. If the energy consumption of a node is greater than a predicted threshold, it is considered as malicious and the packets emerging from these nodes are said to be abnormal. The abnormal packets are passed through the Back Propagation Network (BPN) to analyze the data misinterpretation. In case, the output of BPN is found to be abnormal, then the attack detection is confirmed and attack type is estimated. By simulation results, we show that the proposed technique is more energy efficient.