Passive Intrusion Detection Algorithm Based On Multiple Base Stations In Cellular Network

The passive intrusion detection is an emerging technology enabling the detection of entities that do not carry any devices nor participate actively in the detection process using the already installed wireless infrastructure. It mainly relies on the fact that RF signals are affected by changes due to the motion of entities in the environment. It has the advantage over traditional device-free detecting systems such as video and image detection systems which are limited to the light and line-of-light vision. And it has the advantages of simple structure, easy deployment, lower cost, etc. It has potential applications in the smart home, police security, enterprise management and defense military fields.Existing detection methods include the following features: Firstly, the sliding mean or variance of the received signal strength(RSS) was directly as signal features of the system. But time-varying characteristics of the signal were not considered. Secondly, the system achieved the detection based on a single signal stream that did not guarantee the stability of the detection system. Thirdly, the applications of detection technology was restricted because detection algorithms were studied under wireless LAN(WLAN) and wireless sensor networks(WSN). Based upon the conditions described above, we will research the passive intrusion detection algorithm based on multiple base stations using cellular network environment. This paper mainly includes the following three studies:Firstly, we analyzed the feasibility of passive intrusion detection which includes theoretical analysis about the impact that the body may influence the electromagnetic wave propagation combining with variation of the real RSS values in the real cellular network. Then disadvantages of existing algorithms were studied. And the passive intrusion detection program and research ideas in cellular network were analyzed by closely observing changes of the real signal.Secondly, we started to research the passive intrusion detection algorithm based on a single base station. The RSS signal feature extraction was considered primarily. In order to overcome the problem that the detection performance would degrade due to time-varying characteristics of the signal, the system employed the difference of RSS as the basic input signal features. Further, the nonparametric kernel density estimation function was chosen to estimate the distribution of signal features under the static environment. This distribution also will provide the basis for the online detection phase. This algorithm was finally evaluated in the open and multipath environment. And the experimental results showed that it could obtain a better detection performance.Thirdly, the passive intrusion detection algorithm jointing multiple base stations was further researched and discussed based on the study about the detection algorithm using a single base station. To gain higher detection performance, how to calculate the optimal detection threshold was focused. In addition, not all base stations are beneficial to the improvement of the detection performance, so we need to measure detection capability of every base station. The algorithm calculated the score of each signal feature through evaluating the contribution degree of every data stream combining with the sigmoid function. And then the score matrix was building using all scores of each stream to realize the joint detection of multiple base stations. The system accomplished the model establishment and decision of signal features by employing the support vector machine(SVM) to train the score matrix. This algorithm was finally evaluated in the open and multipath environment. Also we compared the performance of the proposed system with the single base station detection method and traditional methods. And the results showed that the performance of proposed algorithm in this paper was better than other several algorithms.