Indoor WLAN Anonymous Target Intrusion Detection Method Based On The Ray-aided Generative Model

With the wide deployment of Wireless Local Area Network(WLAN)and general support of IEEE 802.11 protocol by various mobile devices,the demand for intrusion detection services for anonymous target without any signal receiving or receiving devices is increasing dramatically.Anonymous target intrusion detection technology based on WLAN makes use of the correlation between the fluctuation characteristics of WLAN signal and the location of intrusive target to detect and locate the intrusive target.It can be widely used in many fields such as smart home management,security monitoring,counter-terrorism,and disaster relief.At present,the system usually relies on the construction and learning of WLAN intrusion fingerprint database,but there are three main problems: the huge cost of constructing WLAN intrusion fingerprint database in a large and complex monitoring environment,the accuracy of virtual WLAN intrusion fingerprint database is difficult to guarantee in a changing monitoring environment,andthe complex learning method of the constantly updating WLAN intrusion fingerprint database,it is difficult to promote.In order to solve the above problems,this thesis proposes an indoor WLAN anonymous target intrusion detection method based on the ray-aided generative model.The main content of this thesis includes the three following parts:Firstly,in order to achieve the low labor and time cost of constructing WLAN intrusion fingerprint database,this thesis uses genetic algorithm to improve the traditional ray-tracing algorithm.The adaptive-depth ray tree based quasi 3-dimensional(3D)raytracing model is proposed to to depict the difference of WLAN signal between the silence and intrusion environments with the purpose of constructing the virtual WLAN intrusion fingerprint database.Secondly,in order to improve the usability of virtual WLAN intrusion fingerprint database,this thesis ingeniously utilizes the adversarial learning idea of traditional Generative Adversarial Network(GAN)proposes the ray-aided generative model based joint virtual and actual learning framework to update the virtual WLAN intrusion fingerprint database through the actual unlabeled WLAN intrusion fingerprint data,and consequently obtain the refined WLAN intrusion fingerprint database.Finally,in order to reduce the complexity of WLAN intrusion fingerprint database matching,this thesis integrates Probabilistic Neural Network(PNN)intrusion detector into the ray-aided generative model based joint virtual and actual learning framework,and uses the multi-feature information of joint virtual and actual data to train it quickly.The correlation between WLAN intrusion fingerprint data and the location of the intrusion target is fully excavated to improve the robustness of target intrusion detection and localization performance.At the same time,extensive experiments are conducted to verify the proposed scheme can not only ensure high intrusion detection accuracy,but also significantly reduce the cost of WLAN intrusion fingerprint database construction...