Crowd Size Estimation Using WiFi Eavesdropping Despite MAC Address Randomization
By: Darren Newell and Taylor Artunian
Department: Embedded Electrical and Computer Engineering
Faculty Advisors: Dr. Yiyi Wang and Dr. Xiaorong Zhang
The consequences of poor urban planning affects more and more lives as the world becomes increasingly urbanized. Engineers need adequate space utilization data to assess how to optimally design urban infrastructure. Using embedded systems and the Internet of Things, our research looks at using Wi-Fi probe request frames as a potential source of crowd size estimation. Crowd size estimation paired with timestamps can provide critical insight to how public space is used, and is relatively difficult to capture without high cost or privacy intrusion. Wi-Fi Probe request frames are emitted from all Wi-Fi enabled devices, and we seek to utilize the information elements emerging from these probe request frames to count unique devices through advanced data analysis techniques. No personally identifiable information is obtained, only the characteristics of a phone's make, model, and relative signal strength.
Our research work involves devising lab and field experiments to determine the best methods of obtaining Wi-Fi radio data, designing embedded devices to capture the signals, and creating software data analysis tools using MATLAB, Python, and Wireshark. With these tools we aim to derive a meaningful and accurate count of unique devices passing through a given area over a defined period of time. Our end goal is to make a scalable, low-cost, portable, and privacy preserving counting device that can provide civil engineers reliable traffic data for their infrastructure improvement and planning work.