Anonymous People Flow Monitoring System Leveraging Bloom Filters

In the last decade, our cities have undergone significant transformation due to the widespread adoption of IoT (Internet of Things) sensors and smart devices owned by citizens, resulting in significant improvements in urban life. One of the most common and relevant research fields concerns the monitoring of people's movements, with the aim of enhancing services for citizens. This type of monitoring also pertains to optimizing transportation infrastructure, bike lanes, and public thoroughfares. However, it is also of crucial importance in identifying and quantifying individuals in sensitive areas, such as ensuring safety during large-scale events or at transit points. Additionally, it is essential to introduce new technologies for tracking individuals in emergency situations, to reduce response times and increase the likelihood of success. In the context of this thesis, we focus on the WiFi fingerprinting technique, which uses the MAC address of mobile devices as a proxy for people counting. Due to European GDPR regulations and the stringent measures taken by major smart device providers to enhance user privacy (e.g., through MAC address randomization), most of the techniques examined in the past must be rethought and redefined. For this reason, we have adopted efficient probabilistic data structures called Bloom filters for storing sensitive information. Thanks to their formal "deniability" property, we are able to offer a solution that safeguards user privacy. Our solution is also compatible with trajectory-based crowd monitoring, as these elements enable set operations, such as intersection, for determining flows.