Abstract:

The negative selection algorithm in intrusion detections (IDs) for wireless sensor networks (WSNs) generally adopts the r-continuous binary string matching mechanism, which leads to low detection rate and cannot reflect the dynamic features of WSNs in a period of time. Aiming at the problem, we present a real-value negative selection algorithm called RNS-WSNs, which uses the change rate of the attribute value during an interval of time as the antigen and antibody, and the Manhattan distance between the two vectors as the affinity. Simulation results on network simulator 3 show that the real-value negative selection algorithm has higher detection rate under the same energy consumption and false positive rate.

Key words: wireless sensor networks (WSNs), artificial immune, negative selection, real-value