Skip to Main Content
In contrast to traditional networks, Wireless Sensor networks (WSN) are more vulnerable to attacks such as DOS, eavesdropping, tampering, node compromise, wormhole and Sibyl. To give security protection for WSN, many attack prevention and detection methods were proposed. As the second line of defence, Intrusion Detection can resist attacks under cryptography technologies are unavailable, and provide strategies for network recovery. This paper proposed a novel attack detection based on RSSI technology. It collects multi-observed RSSI values to form vectors, and uses mean vector confidential interval examination of multivariate normal population to detect malicious packets. Real experiments showed that vector-based detection approach have a better detection sensitivity and fault tolerance than single value detection approach.