Indoor device-free passive localization for intrusion detection using multi-feature PNN

Indoor device-free passive localization is an emerging technique that can be used in a variety of fields, like the intrusion detection and smart homes, which does not require the target to carry any devices or participate actively during the localization. In this paper, we rely on the Probabilistic Neural Network (PNN) algorithm which has been widely used in pattern recognition in combination with the device-free passive localization technique to realize the intrusion detection. We utilize the variance of RSS to classify the different intrusion states. Due to the limitation of single-feature input in providing information for classifier, we propose the multi-feature PNN to improve the accuracy of intrusion detection, as well as area localization. Our experiments conducted in an actual indoor Wi-Fi environment shows that the multi-feature PNN can reach better performance than the PNN with a single-feature input. Finally, the proposed approach achieves higher accuracy compared to some exited device-free passive detection approaches, and our approach can locate the area which the intruder is really located at accurately.