An adaptive wireless passive human detection via fine-grained physical layer information

Wireless device-free passive human detection is a key enabler for a range of indoor location-based services such as asset security, emergency responses, privacy-preserving children and elderly monitoring, etc. Since the feature of received signal varies with different multipath propagation conditions, an labor-intensive on-site calibration procedure is almost indispensable to decide the optimal scenario-specific threshold for human detection. Such overhead, however, impedes readily and fast deployment of wireless device-free human detection systems in practical indoor environments. In this work, we explore PHY layer multipath profiling information to extract a novel quantitative metric K_s as an indicator for link sensitivity, and further exploit a linear detection threshold prediction model. We design an adaptive device-free human detection scheme that automatically predicts the detection threshold according to the richness of multipath propagation within monitored areas. We implement our scheme with commodity WiFi infrastructure and evaluate it in typical office environments. Extensive experimental results show that our scheme yields comparative performance with the state-of-the-art, yet requires no on-site threshold calibration.