Development of Doppler CSI model for Wi-Fi based hand gesture detection application
Date:
Abstract Various studies have proposed the use of RF signals to realize a non-wearable hand gesture sensing applications in the environments where commercial vision-based motion sensing cannot properly operate such as non-line-of-sight (NLoS) environment and private places. One way to sense the hand motion is through the Doppler frequency extracted from channel state information (CSI) of Wi-Fi chips. Although many works have proposed the solutions for CSI-based hand gesture sensing, there is no model that could properly explain the behavior of hand movement in terms of Doppler profiles. In this paper, the deterministic CSI Doppler model has been developed which describes the Doppler frequency profile caused by particular gestures. Through simulation and discussion, it was found that Doppler frequency profile depends on the speed of the motion, position, and trajectory relative to transmitter and receiver. The density of the Doppler power spectrum is the result of the constructive accumulation of a various number of propagation paths experiencing similar Doppler frequency shift. In addition, the proposed model could be beneficial for the gesture recognition as it could generate the Doppler profile for pattern training.