Pedestrian Flow Estimation Through Passive WiFi Sensing

In public places, even if pedestrians do not have their mobile devices connected with any WiFi access point (AP), WiFi probe requests will be broadcast, so that WiFi sniffers can be employed to crowdsource these WiFi probe packets for use. This paper tackles the problem of exploiting the passive WiFi sensing approach for pedestrian flow analysis. To be specific, a passive WiFi sensing model is first established based on a probabilistic analysis of interactions between WiFi sniffers and the moving pedestrian flow, capturing the main factors affecting pedestrian flow characteristics. On that basis, a sequential filtering algorithm is proposed based on the Rao-Blackwellized particle filter (RBPF) to produce simultaneous and efficient estimates of the pedestrian flow speed and pedestrian number utilizing the real-time sniffing results. In order to validate this study, an experimental pedestrian surveillance system using WiFi sniffers is deployed at the transfer channel of a metro station in Guangzhou, China. Extensive experiments are conducted to verify the passive sensing model, and confirm the effectiveness and advantages of the proposed algorithm. The pedestrian flow estimation not only helps to improve the safety and facility management and customer services, but also paves the way for introducing other novel applications.