Sybil-Based Attack Impact on Mobile Crowdsensing-Based Navigation: Privacy Versus Utility

Abstract

Mobile Crowdsensing (MCS)-based data collection can facilitate a wide range of services. Navigation MCS (N-MCS)-based applications provide the Expected Time of Arrival (ETA), helping millions in their daily lives. N-MCS users share their locations and speeds to build a current state of traffic. However, malicious users, often termed Sybils, submit falsified data from remote locations using multiple accounts to mostly fake congestion on road segments and redirect benign N-MCS users. We investigate the impact of Sybil-based attacks on N-MCS, given the sensitive nature of location data, with and without privacy-enhancing technologies (PETs). Our simulation-based N-MCS system models both PETs and non-PETs variants, as well as Sybil-based attacks targeting critical links connecting communities in the road network, identified by the Infomap algorithm. With our extensive set of experiments, we have shown that, comprising on average only 3.8% of the N-MCS population, with PETs it is significantly easier to reroute the users. The impact on the travel time and trip length heavily relies on the traffic state, and in some cases PETs can significantly increase the targeted users’ travel time (trip length) median percentage change more than non-PETs by at least 12.81 pp (9.89 pp).