Indoor WLAN Intelligent Target Intrusion Sensing Using Ray-Aided Generative Adversarial Network

An indoor target intrusion sensing technique has been used in many fields, such as smart home management, security monitoring, counter-terrorism, and disaster relief. At the same time, with the wide deployment of wireless local area network (WLAN) and general support of the IEEE 802.11 protocol by various mobile devices, the target intrusion sensing can be realized based on the existing WLAN infrastructure without requiring the target to carry any special device. However, the existing indoor WLAN target intrusion sensing approaches usually depend on the radio map construction with huge labor and time cost, which is the major barrier of current systems. In response to this compelling problem, we propose the new ray-aided generative adversarial model (RaGAM) to automatically construct the radio map, which is used for the indoor WLAN intelligent target intrusion sensing and localization. To achieve the low labor and time cost, the RaGAM uses the adaptive-depth ray tree based the quasi three-dimensional ray-tracing model to depict the difference of WLAN signals between the silence and intrusion environments with the purpose of constructing the synthetic radio map. Considering the gap between the synthetic and actual radio maps, we modify the conventional generative adversarial network by the joint synthetic and unsupervised learning (or called S+U learning) from the actual unlabeled received signal strength (RSS) data to improve the precision of the proposed ray-tracing model, and consequently obtain the refined radio map. After that, the statistical characteristics of the refined radio map are utilized to construct the training set for the probabilistic neural network (PNN), and then by using the well-trained PNN to classify the newly collected RSS data, the target intrusion sensing, and localization are achieved. The experimental results show that the proposed approach cannot only perform well in terms of computation cost and the ray-tracing accuracy, but also sense the target intrusion states accurately.