Crowd Size Estimation with Passive Electromagnetic Sensing

Abstract

Crowd size estimation has become of even greater significance than before COVID-19 restrictions. At the same time, wireless technologies are ubiquitous and change rapidly; we hypothesise that we can estimate crowd sizes from electromagnetic (EM) activity, agnostic from the used communication technology. In this experimental study, we investigate crowd size estimation based on the activity that ambient WLAN devices generate inside specific WLAN channels. By recording and analysing this EM activity, we tested if this corresponds to the crowd size at a large music festival with thousands of attendees. We identify and compare trends within a data set of 186 samples over three consecutive days. We demonstrate a significant correlation between our approach and the data of three other crowd estimation systems: entrance and exit counts, MAC address counts, and device-free sensing using sub-GHz attenuation. Our study is the first step to passively and technology agnostic estimate large crowd sizes by capturing the electromagnetic activity inside the 2.4 GHz band with Software Defined Radios.