Research On Intrusion Detection Technology Of Wireless Sensor Networks Based On Behavior Trust

With being widely applied in military surveillance, medical treatment and environment monitoring, the security issues of Wireless sensor networks (WSNs) have become increasingly prominent. As a reasonable complement for encryption, authentication and other security mechanisms, intrusion detection technology has become an important domain of WSNs security research. So it has very academic and practical value to design an intrusion detection approach adapting to wireless sensor networks characteristics.In recent years, network trust research has become a hot topic. So from trust point, after analyzing on security need and current technology of intrusion detection, we design a reputation model and an intrusion detection model for wireless sensor network on the basis of nodes' behavior trust. The simulation experimental results show these models are effective.The main research works are as follows:1. Briefly introduce the status of domestic and overseas research on information security guarantee technology of wireless sensor network, and put forward the research background, meaning, and main tasks of our research.2. Analyze the security requirements in wireless sensor network and the researches of existing intrusion detection technology, prepared for the future work on intrusion detection technology fitting for wireless sensor network.3. According to analyze current trust mechanism, and for the lack of node energy in BRSN model, a reputation computing model of wireless sensor network is designed on the basis of the node's behavior. In order to evaluate a node's reputation, the model firstly calculates direct trust parameters from the node's behavior, then uses indirect trust parameters as reference, and finally integrates these parameters with energy reputation to get the synthetic reputation evaluation value. If the evaluation value is larger than or equal to the specified threshold, we can confirm the node is credible.4. At the basis of the reputation-computing model we designed, and after analyzing features of the sensor network's intrusion detection mechanism, we proposed an intrusion detection model based on trusty nodes behavior. The model structure is composed of clusters. The nodes inside clusters make use of the reputation-computing model to calculate the coordinate nodes'synthetic reputation values to check whether some invasion has taken place. If some node is being intruded, the messages will be reported to cluster head. When the number of the reporting nodes is more than the specified threshold, the cluster head can confirm this node being intruded, then the head will report the information to sink node and start intrusion response at the same time.5. The reputation-computing model and intrusion detection model all have been simulated by experiments. The experimental results show that the designed models can identify vicious nodes and low competitive nodes, helping to promote the security and reliability of the network, and embodying the equity of the network. But there are still some shortages in these models; we will focus on them in the following study.