Wi-Fi and Radar Fusion for Head Movement Sensing Through Walls Leveraging Deep Learning

The dataset pertains to head movements, preserved in a manner that respects privacy. Specifically, radio frequency (RF) sensing technology was employed to capture the unique channel variations caused by head movements. The USRP x300, equipped with a VERT2450 omnidirectional antenna for transmission and a HyperLOG 7040 X for reception, was utilized. Additionally, the experiment was replicated using a Xethru UWB radar to capture Doppler frequency shifts resulting from lip movements. Six classes of lip movements were identified: Head Up, Head Down, Right 90, Left 90, Right 45, and Left 45. These classes were recorded in scenarios both with and without a wall, using Wi-Fi and radar signals. We succeeded in detecting head movements even when obstructed by a wall. Four subjects—two males and two females—participated in the experiments.