Multiuser Wireless Hand Gesture Recognition by Spatial Beamforming

Abstract

Hand gesture recognition through wireless sensing is a new method of human-machine interaction, and is an important research direction for next generation integrated sensing and communication systems. Recently, although there are plenty of works on wireless sensing and hand gesture recognition, most of them are applied in single user scenario. Since the recognition algorithm is mainly realized by analyzing the variation of wireless propagation channel, the simultaneous movement of multiple people will cause superposition of the dynamic channel responses and cause interference to each other. This article proposes to use spatial beamforming to alleviate the mutual influence of multiuser hand gestures. Since the hands of multiple users reflect the same signal, the channel responses are usually not orthogonal even if the users are well separated in different locations, and there is no reference signals to help distinguishing the channel response reflected by different users. We propose a preamble gesture scheme to estimate the spatial channel of dynamic reflections under the impact of strong phase noise, and use the Doppler variation feature to verify the channel response belongs to hand movement. We have a thorough analysis of the inter-user interference impacted by the transmitter-user-receiver geometry relations, user movement speeds and directions. The interference suppression effect is demonstrated by prototype experiments under real LOS and NLOS scenarios exploiting LTE signals.