Malicious Node Detection Strategy Based On Trust Evaluation

Due to the nature of energy-limited and broadcast makes wireless sensor network more vulnerable to be attacked. Therefore, in order to ensure reliable data transmission and to resist various attacks, it is an important problem to specially design security mechanisms for WSNs.As a kind of security mechanism, trust model is used in designing WSNs security mechanisms. However, trust model also can become the object of attack due to itself own defects. And the exist of malicious nodes in sensor network will affect the user’s determination. Therefore, this paper studies the contents as follows:Firstly, this paper proposes a trust model-based malicious nodes detection strategy for WSNs. the proposed trust model brings in an exponential attenuation factor to depict the variation of dynamic behavior on nodes and gradually increases the proportion of negative experience in trust evaluation. For malicious recommendation problem, it calculates the mean difference of trust evaluations between trustor and recommender for the public interaction nodes to evaluate the recommendation credibility, and the recommendation credibility is acted as the recommended weight to reduce the affect of malicious recommendation. When calculating the overall trust value, the adequacy of the interactive experience between nodes is used for the dynamic weighting factor to balance the direct and indirect trust value. At last, the simulation results verify that our scheme not only has better ability to defense malicious nodes but also has higher accuracy.Secondly, this paper also designs a malicious node detection strategy based on weighted trust. The strategy employs a dual threshold to cope with the trade-off between event detection accuracy and event misdetection rate. In order to reduce misjudgment for event boundary nodes and their neighbors, the proposed strategy designs an event boundary detection algorithm. Thus, it makes the proposed strategy improve malicious node detection and reduce the misdetection rate. Simulation results verify that the effectiveness of the strategy.