WiN: non-invasive abnormal activity detection leveraging fine-grained WiFi signals

Abnormal activity detection has recently drawn significant research attention, because of its potential applications in providing critical and severe emergency information. Existing non-invasive activity detecting approaches rely on radio signals, wearable sensors or specialized hardware. Motivated by the observation that the amplitude and the phase information of channel state information CSI are highly sensitive to activity variation, we propose WiN, a non-invasive abnormal activity detection system, based on fine-grained physical layer channel state information, which is available from commercial off-the-shelf WiFi devices. We implement WiN and evaluate its performance in IEEE 802.11n devices. Extensive experiments in typical real-world environments demonstrate that WiN can achieve impressive performance in abnormal activity detection.