Research On The Trust Management System Of Wireless Sensor Network Based On Poisson Distribution

In recent years,the breakthrough of sensor related technology has promoted the rapid development of wireless sensor networks,which has led to a series of security problems.Although traditional security mechanisms such as authentication and network protocol can effectively resist external attacks,the defense against network internal attacks caused by node selfishness or node compromise is particularly poor.In addition,the traditional security mechanism is not suitable for wireless sensor networks with limited resources due to its high computational overhead and high algorithm complexity.Trust management system has emerged as an effective means to resist internal attacks in wireless sensor networks.The paper has surveyed the significant technologies of trust management in wireless sensor network.On the basis of fully considering the unique nature of the wireless network,we put attention on the trust between nodes to design a trust management system,which aims to solve the problem of mathematical distribution and node energy utilization in the construction of trust management system.In order to improve the speed of trust evaluation of nodes and enhance the ability to resist internal attacks,which largely deal with the on-off attack.In addition,we have proposed a multi-fusion trust decision mechanism based on Bayesian theory and Markov chain to solve the problem of subjectivity and uncertainty in evaluating the reputation of unevaluated nodes according to observation behavior,which aims to improve the speed of trust evaluation of nodes and enhance the ability to resist collusion attacks.More specifically,the main contributions of this paper can be summarized as follows:(1)We have reviewed the trust management in wireless sensor networks in recent years,and deeply analyzed the relationship of the related mathematical distribution.The expression of trust and reputation based on Poisson distribution is deduced and incorporates into the trust management system,which weakens the potential hypothesis.The entropy theory is introduced to evaluate the certainty of the direct trust value of nodes.Only when the entropy of direct trust value is high,the indirect trust is integrated for enhancing the trust relationship between nodes.The method of calculating indirect trust on demand saves the energy of nodes and reduces the complexity of the algorithm to a certain extent.The simulation results show that the trust management system can evaluate the trust of nodes more quickly and efficiently resist the on-off attack than the traditional one.(2)By reviewing the trust management in wireless sensor networks in recent years,there exists a certain extent of subjectivity and uncertainty in the observation process,which reputation is obtained from the observation of the current behavior of nodes by Bayesian theory.Therefore,a multi-fusion decision mechanism based on Bayesian theory and Markov chain is proposed.We verified the credibility of the behavior of the observed nodes by constructing the transfer matrix and judged the current interactive behavior of the nodes according to the evaluation results,so as to evaluate the trust value of the nodes objectively and accurately.The simulation results show that the proposed method can evaluate the trust of nodes more quickly and efficiently resist the collusion attack than the traditional one.