Towards Data Poisoning Attacks in Crowd Sensing Systems

Abstract

With the proliferation of sensor-rich mobile devices, crowd sensing has emerged as a new paradigm of collecting information from the physical world. However, the sensory data provided by the participating workers are usually not reliable. In order to identify truthful values from the crowd sensing data, the topic of truth discovery, whose goal is to estimate each worker's reliability and infer the underlying truths through weighted data aggregation, is widely studied. Since truth discovery incorporates workers' reliability into the aggregation procedure, it shows robustness to the data poisoning attacks, which are usually conducted by the malicious workers who aim to degrade the effectiveness of the crowd sensing systems through providing malicious sensory data. However, truth discovery is not perfect in all cases. In this paper, we study how to effectively conduct two types of data poisoning attacks, i.e., the availability attack and the target attack, against a crowd sensing system empowered with the truth discovery mechanism. We develop an optimal attack framework in which the attacker can not only maximize his attack utility but also disguise the introduced malicious workers as normal ones such that they cannot be detected easily. The desirable performance of the proposed framework is verified through extensive experiments conducted on a real-world crowd sensing system.