Towards Optimal Incentive-driven Verification in Social Sensing Based Smart City Applications

Social sensing has emerged as a new smart city application paradigm that achieves low deployment costs by leveraging privately owned edge devices for data collection and processing. Unfortunately, such an application paradigm is vulnerable to result spoofing, where attackers fabricate computation results and convince the server that they are valid. This problem has not been adequately addressed due to three important challenges. First, the raw data originates at the edge, and therefore the application server cannot directly verify the computed results sent by the edge devices without receiving the raw data as well. Second, personally owned edge devices can easily collude to convince the server that fabricated results are real, even under sophisticated scrutiny. Third, the high level of churn common in such systems nullifies established reputation-based schemes. To address the above challenges we develop a novel incentive-driven verification system in which risk is minimized under a budget constraint. We compare our solution to state-of-the-art computation verification schemes through an extensive evaluation. The results show that our solution achieves the least risk of computation spoofing under a constrained budget when compared to all of the baselines.