Research On Passive Intrusion Detection Method Based On Channel State Information

With the development of wireless network technology,researchers apply Wi-Fi signals to passive intrusion detection methods,aiming to achieve passive intrusion detection of personnel only with the existing commercial wireless network equipment.Compared with the initial research which using the received signal strength of the wireless signal as the detection feature,the channel state information of the wireless signal can describe the environmental feature information of the wireless signal propagation process with a fine-grained manner.Using the channel state information as a detection feature can effectively improve the accuracy of passive intrusion detection method.However,the current passive intrusion detection methods based on channel state information have some problems: Firstly,different wireless links have different sensitivity to personnel activities.The terrible wireless links may reduce the accuracy of the detection method.Secondly,it is difficult to balance the time consumption of the detection method and the detection performance.The detection methods with better detection performance are often accompanied by higher algorithm complexity.Thirdly,the frequency domain characteristics of wireless signal data are not fully utilized.For solve the above problems,this paper designed a passive intrusion detection method based on channel state information.Our main work and innovations are as follows:(1)We propose a wireless link selection method based on the VIKOR method.The channel state information in a static environment is collected as the evaluation attribute of the wireless link,using the VIKOR method which belong to the field of multi-attribute decision-making to sort the performance of the four wireless radio links,and select the best wireless radio link for achieving passive intrusion detection.The feasibility of the method is verified through the collected data.(2)We propose a passive intrusion detection method based on dynamic threshold.The initial threshold is set based on the channel state information in a static environment,and the threshold is dynamically maintained by the exponential moving average algorithm.The method can achieve quick response to personnel intrusion behavior and it is verified in an indoor environment.The experimental results show that the comprehensive detection accuracy of the detection method can reach more than 97%,and it can achieve a quick response to the intrusion behavior of personnel within 1 second.(3)We propose to utilize the power spectral density distribution of the channel state information as the detection feature.Based on the correlation between each subcarrier,the first principal component of the data is selected by principal component analysis,and the power spectral density distribution of the channel state information is extracted from it as the detection feature.We compare the differences between the power spectral density distribution of the static environment and the dynamic environment to determine whether there is an intrusion by a person.