CSI-based passive intrusion detection bound estimation in indoor NLoS scenario

Passive intrusion detection is important for many indoor safety applications. By utilizing widespread WiFi infrastructures, channel state information (CSI) based intrusion detection methods have attractive advantages such as low cost, non-line-of-sight (NLoS) support, and privacy protection. However, existing CSI-based methods lack in-depth and intensive analysis of intrusion detection bound. To the best of our knowledge, this is the first work that precisely characterizes and models CSI-based intrusion detection bound in typical indoor NLoS scenario. A bound is defined as an intruder's farthest position from the transmitter during intrusion detection. To derive a model for the bound, we first derived an intrusion-disturbed NLoS channel model by analyzing the influence of human intrusion on the NLoS wireless channel. Subsequently, based on the channel model, we further derived an intrusion detection bound model. Based on the derived bound model, we proposed a practical system to estimate the real intrusion detection bound. Extensive experiments were conducted based on practical systems. The experimental and simulation results verified and demonstrated the effectiveness of the derived bound model. Our work not only reveals the fundamental performance limit of the basic intrusion detection method in an indoor NLoS scenario, but also provides a valuable reference for bound estimation for other fine-grained wireless sensing applications.

Automated summary

Passive intrusion detection using WiFi infrastructures has advantages such as low cost and privacy protection. However, existing methods lack in-depth analysis of intrusion detection bound. This study precisely characterizes and models CSI-based intrusion detection bound in an indoor NLoS scenario and proposes a practical system for estimating the real intrusion detection bound.