Trust Model In Wireless Sensor Network Intrusion Detection

Wireless sensor network, as a new emerging technology, is still in its infancy, and the relevant technique researches are also in discuss. Wireless sensor network is a major research priority and hot spot. Network security, which is prerequisite of network to normally work, is one of the most popular researches.As an important branch of WSN (Wireless Sensor Network) Intrusion Detection, malicious or compromised node identification is discussed in this paper. In light of this, we are introducing a trust-based mechanism for identifying malicious nodes. And a detection model is established, which consists of distinguishing mechanism, trust index calculation, trust index update and trust index distribution. Distinguishing mechanism is employed to identify whether the declarations from common nodes are true, according to the WSN high-density and large-scale distribution features. It can improve efficiency and restrain jamming attacks. Update weights is introduced to meet the diversity of the WSN application; to adjust the ratio of historical and present trust indices; to prevent malicious nodes accumulate high trust indices. Reward and Punishment mechanism is designed to encourage common nodes work more actively, and punish those malicious tendencies. Trust index distribution is specially designed in order to serve the actual situation better. For wireless sensor network's event-driven character, we proposed a distributing method for monitoring nodes which is event-driven mechanism mainly and time-window mechanism as supplement.Evaluations are performed by simulation; the results indicate obvious advantages of the approach in identifying malicious or compromised nodes, and it can ensure wireless sensor network runs normally under a safe circumstance after be invaded. Also, compared to another model RFSN, both accuracy and speed in identifying malicious nodes, this approach is better than RFSN. Furthermore, it plays a good effect in the robustness of network.