A Novel Intrusion Detection Method for WSN

Wireless sensor network (WSN), which combines the technology of sensor, embedded system and wireless communications, has become increasingly popular and important in our lives. Security is an important issue for WSN. In this paper, we propose a novel method to detect the attacks in WSN. Our method composes of two important stages: offline training and online testing. In the offline training stage, we collect enough training samples, extract the features and then train the models. In the online testing stage, we extract the features of the captured network packets and compare them with the trained models. The hierarchical system could dramatically reduce the amount of online training without sacrificing the detecting accuracy. We deploy the proposed approach in a wireless sensor network for forest monitoring to evaluate its performance. The experiments show that our method performs better compared to the traditional methods.