WiFi Sensing Model for Intrusion Detection in Smart Home Environment

Recently, WiFi signals are not only used for communication purposes but also for various sensing applications. The Channel State Information (CSI) of the received WiFi signal captures the environmental dynamics. This CSI data can be used for motion detection, gesture recognition, localization, human presence detection, environmental monitoring, and other device-free sensing applications. The amplitude and phase of CSI data give fine-grained information about the changes in the signal transmission path due to disturbances caused by physical and environmental changes. This paper focuses on the monitoring of intrusion-related activity by using low-cost WiFi-enabled ESP32 micro-controller devices. We performed four different intrusion-related human activities in the indoor environment. Performed feature engineering on received CSI data for all the activities such as entering into the room, leaving the room, sneaking into the room without entering, and doing suspicious activities in front of the room. We used a Low-pass filter to filter out the noise from the received signal. To classify the activities accurately, we used different lightweight machine learning algorithms such as Support Vector Machine (SVM), Random Forest (RF), and K-Nearest Neighbor (KNN). The result shows that the KNN performs better than other models with an accuracy of 95 %.