Detecting Mobile Crowdsensing Sybil Attackers via Presence Verification

Mobile crowdsensing (MCS) relies on smart, portable devices to conveniently collect sensory data from our surroundings. MCS-based apps, e.g., Google Maps, are already well-integrated into our everyday lives. However, Sybil-based attacks, with an attacker creating many fake identities and the illusion of numerous contributors to influence MCS-based functionality, pose a significant threat. MCS systems need security, including mechanisms to vet incoming users and prevent the introduction of Sybil nodes. Intuitively, each incoming contributor can be verified to be an actual device near other devices by other newcomers and contributors already part of the MCS system. We propose a novel cooperative MCS user presence verification protocol based on this idea, also ensuring users are physically present in locations relevant to the MCS tasks. The protocol leverages a commodity component, Bluetooth, with each user broadcasting to prove their presence to users listening and recording Received Signal Strength Indicator (RSSI) values in multiple randomized rounds. The presence verification is done by a simple server tasked with registering users and orchestrating the protocol based on the collected information. The protocol identifies a broadcast signal on behalf of multiple users, indicating a potential Sybil behavior. We conduct extensive simulations to evaluate the performance of the proposed method, demonstrating its ability to find Sybils with high accuracy even when Sybils are nearly the majority in the protocol session.