WiSH: WiFi-based real-time human detection

Abstract

Sensorless sensing using wireless signals has been rapidly conceptualized and developed recently. Among numerous applications of WiFi-based sensing, human presence detection acts as a primary and fundamental function to boost applications in practice. Many complicated approaches have been proposed to achieve high detection accuracy, but they frequently omit various practical constraints such as real-time capability, computation efficiency, sampling rates, deployment efforts, etc. A practical detection system that works in realworld applications is lacking. In this paper, we design and implement WiSH, a real-time system for contactless human detection that is applicable for whole-day usage. WiSH employs lightweight yet effective methods and thus enables detection under practical conditions even on resource-limited devices with low signal sampling rates. We deploy WiSH on commodity desktops and customized tiny nodes in different everyday scenarios. The experimental results demonstrate the superior performance of WiSH, which has a detection accuracy of >98% using a sampling rate of 20 Hz with an average detection delay of merely 1.5 s. Thus, we believe WiSH is a promising system for real-world deployment.