WiId: A WiFi-BASED INDOOR INTRUSION DETECTION SYSTEM

This paper proposes an indoor intrusion detection system WiId based on WiFi Channel State Information (CSI) for effective detection and discrimination of personnel intrusion activities. By examining the relevance between human motion and amplitude information in channel state information, the system classifies different activities during intrusion. For better overall performance, this paper proposes dynamic adaptive link selection based on the maximum range. While fusing the subcarrier information in multiple links, it deletes signal data insensitive to human behavior features for higher system recognition accuracy. The experimental results indicate that this method has an average recognition accuracy of 96.2%, demonstrating good robustness and stability. © 2022, Politechnica University of Bucharest. All rights reserved.