WiFi Fingerprinting For Controlling Social Distancing

Internet of Things (IoT) has become a cutting edge research topic in the recent years to extend the massive internet connectivity to every physical device used in our daily-life routine. As the COVID-19 pandemic hits the smart cities as Turin, it is essential to manage the public transport boarding limitation to maintain the social distancing as much as possible. In this thesis work, the analysis of people's boarding on bus is implemented by electromagnetic fingerprinting using a Raspberry PI 3 with a sniffer mounted on the bus, that is connected to the internet by a 4G SIM modem and connected to network of the bus that provide essential information (e.g. door status, location, line ID, etc ..). The sniffer captures WiFi probe request messages sent frequently by active devices searching for known access points. Each message contains a randomized MAC address that needs to be matched with other random MAC addresses corresponding to same device for it to be counted as a unique passenger. Taking to account that not all people keep the WiFi interface enabled, or not all of them have a smart device, a correction factor and a smoothing process is applied. At each bus stop, our system returns an estimated number of passengers to be stored in a database and then visualize it on a real-time dashboard. Test results prove that our system is efficiently and accurately classifying the bus utilization based on a confusion matrix. Moreover, it is possible to identify frequent patterns of a specific line at different time slots of any given day. Also it is relevant to see the effect of the COVID-19 lockdown regulations on such patterns.