Research On Intrusion Detection Algorithms In Wireless Sensor Networks Based On Background Updating

Recently intrusion detection technology in wireless environment is attracting more and more attention, and has been widely applied in military protection, energy saving and fire rescue. Intrusion detection is refers to detecting the breaking or the attempt to break into system by manipulating behaviors, security logs, audit data or other information available on the Internet. There are two kinds of intrusion detection system based on wireless sensor network:active intrusion detection and passive intrusion detection. Active intrusion detection needs to detect target carry wireless transceiver equipment, while the passive detection can implement the intrusion detection without carrying any wireless transceiver equipment, which changes the special hardware support conditions of active intrusion detection and makes the universality of detection technology stronger. This paper mainly studies a passive intrusion detection based on wireless sensor network based on the signal changes.The work is mainly about research and application of intrusion detection algorithms based on wireless sensor networks. The present passive intrusion detection algorithms, such as traditional detection algorithm, differential detection algorithm and interval differential detection algorithm are based on the change of signals to implement intrusion detection. Specifically, traditional detection algorithm directly uses the signal difference to determine presence of invaders, and only a threshold is needed which is associated with experimental environment. Differential detection algorithm is based on the characteristic value of two adjacent signals, determined by the strict statistical results, to judge the existence of invaders. The implementation of the interval difference detection algorithm is similar to the differential detection algorithm, but it has the advantage of expressing intruder information graphically. Though easily to implement for the above three detection algorithms, signal changes as the change of environment, thus reducing the detection rate and raising the error detection rate. With the goal of improving the detection rate and reducing false detection rate, this paper proposes an intrusion detection algorithm based on the update of background.In the proposed algorithm, this thesis implements a wireless sensor network, it consists of one wireless controlling node and multiple wireless scanning nodes, collected data will be send to the PC to process. The intrusion detection algorithm based on background update consists of three stages, that is, setting up background model, background difference and background model update. It can automatically update the background model as with the change of environment, so that can adapt to environment changes in real time. However, its simplex background model cannot express environmental factors accurately, and it is hard to adjust real-timely with the in complex environment changes. Gaussian mixture background model updating method is employed, which can adapt to more complex environmental changes, to obtain higher detection rate and lower false detection rate even in a complex environment.Finally, five kinds of algorithm are evaluated on their detection rate and false detection rate. Statistical results show that the intrusion detection algorithm based on the update of background and the Gaussian mixture model updating method have higher detection rate and lower false detection rate, and optimistic detection results can be obtained even in the bad environment.