Intrusion detection based on traffic analysis in wireless sensor networks

This paper proposes a lightweight, fast, and efficient traffic intensity-based intrusion detection method for wireless sensor networks (WSNs), containing a large number of nodes which transmit data periodically. Intrusion detection method is based on analysis of neighbor's behavior and thresholding technique, applied to selected parameters: the number of received packets in a time window of a given length and inter-arrival time of packets. It does not require installation of any additional hardware and extra communication costs; its computational requirements are negligible. The proposed technique ma be used in both hierarchical and flat WSNs. The paper contains recommendations to the choice of threshold value for each traffic parameter. The results of evaluation, which was performed in Castalia simulator for WSNs, demonstrate that the proposed intrusion detection technique has better accuracy in comparison to the common algorithms.