Privacy Risk Assessment of Travel Trajectories in Metro AFC Data

The rapid expansion of metro systems in major cities worldwide has resulted in the accumulation of vast amounts of travel data through Automatic Fare Collection (AFC) systems. While this data is crucial for enhancing and optimizing transportation networks, it also raises significant concerns regarding passenger privacy due to the potential exposure of individual travel patterns. In this paper, we propose a novel privacy risk assessment model aimed at quantifying the uniqueness of travel trajectories and evaluating the associated privacy threats. Utilizing AFC data from Chengdu collected in March 2021, we first employ an information entropy approach to assess the uniqueness of travel trajectories across different time granularities. We then apply the K-Means clustering algorithm to classify these trajectories into categories based on their uniqueness levels, enabling us to investigate how factors like travel time and routes influence trajectory uniqueness. To further understand the privacy implications, we simulate attacker scenarios by replicating the process of identifying users based on known travel trajectories, thereby assessing the risk of privacy exposure under various time scales. Our experimental results reveal that metro travel trajectories exhibit high uniqueness at finer time resolutions, and that travel routes significantly affect this uniqueness. Notably, even at coarser time granularities, nearly half of the users remain susceptible to identification risks. These findings highlight the critical need for effective privacy protection strategies in the management of AFC data. The insights provided by this study are essential for policymakers and transit authorities seeking to safeguard passenger privacy while leveraging AFC data for transportation improvements.