Intrusion Detection Research In Wireless Sensor Networks Based On The Barrier Coveage

Wireless sensor networks always detect the monitoring area by using a large number of wireless and self-organization sensor nodes. Since the physical and inner characteristics of the sensor nodes, they are easily to be disturbed by external interference factors or interior factors when transmitting data. As sensors are always deployed in harsh and sparsely populated environment, it is not easy to replace the dead sensor nodes. So how to efficiently use a limited number of sensor nodes to achieve the object of monitoring the target area is one of our research purpose.Firstly, this thesis introduces some characteristics and key technologies of wireless sensor networks, and pays much attention to the barrier coverage. Then the k-barrier coverage model has been studied, and its mathematical conditions has been analyzed. Since the barrier coverage of wireless sensor networks is the most widely used in intrusion detection with high efficiency, which is the basement of our thesis.Secondly, this thesis presents the intrusion detection system of wireless sensor networks briefly and summarizes the overall framework of the intrusion detection system. Two commonly algorithms of the intrusion detection system based on anomaly,(Traffic Prediction based on DoS attack Detection) TPDD algorithm and(the CUMULATIVE SUM) CUMUM algorithm, has been studied. The intrusion detection system has been studied in the aspects of network communication. The formal is based on traffic forecasting, and the latter is based on statistical behavior. Then the relationship between the detection probability and data rata of the two algorithms has been presented by simulation results.Thirdly, a k-barrier constructing algorithm has been put forward in this thesis, and it has been promote extened from a special circular monitoring area to more general area, such as a rectangular area. According to the characteristics with limited energy of sensor nodes in wireless sensor networks, a barrier mended model of node failure has been considered. The neighbor sensors of the failure one can find a replaceable node, which can prolong the lifetime of the whole k-barrier coverage model. We prove that the k-barrier coverage put forward in this paper has more advantage in consuming sensors when building barriers.At the last part of this thesis, we analyse the application in one-way intruder detection basing on the k-barrier coverage in wireless sensor networks, which is different with two neighbor barrier coverage algorithm. Comparing with the simulation performance of our k-barrier constructing algorithm and the two neighbor barrier algorithm, we find that there is litter difference between both of them in terms of building the same number of barriers. But the complexity of implementation is different. Our algorithm reduces the difficulty of building two neighbor barrier, on the country, it makes the requirements of constructing barrier coverage more easily.