Measuring the Accuracy of Crowd Counting using Wi-Fi Probe-Request-Frame Counting Technique

Abstract

Wi-Fi in smartphones are designed to periodically transmit probe-request-frame to determine when a known access point is within range and by capitalizing this Wi-Fi behavior, crowd counting and analysis have been done by continuous monitoring and counting these Wi-Fi frames. The proliferation of Wi-Fi enabled mobile devices and the ever-increasing number of mobile devices in use, suggests opportunities for developing lowcost crowd counting and analysis solution. This work attempt to measure how well do monitoring and counting these Wi-Fi frames correlate with the actual number of people presence in a crowd. In this paper, we also compare the pros and cons of various crowd counting technologies, describe the system that we used for counting Wi-Fi frames and compare its accuracy against manual crowd counting technique in an event involving the public continuously for 8 hours. The results are promising, the correlation between manual counting and Wi-Fi frames counting is 0.89322. In addition to that, the Wi-Fi frames counting technique can even reveal the retention rate of the crowd.