IM-LDP: Incentive Mechanism for Mobile Crowd-Sensing Based on Local Differential Privacy

In recent years, the rapid development of embedded technology has given rise to mobile crowd sensing (MCS) systems to outsource sensing tasks to the public crowd equipped with various mobile devices. Sensing data often involves the workers' privacy, but overprotection of workers' data leads to the decrease of the data accuracy. Therefore, a crucial issue in such systems is how to balance workers' data privacy and data aggregation accuracy. The local differential privacy guarantees the data privacy by returning the privacy budget to workers. However, existing works only considered the workers' reputation as the weight of the aggregation result, but did not correlate with the rewards that workers deserve, which restrained workers' incentive of participation. Different from these works, by quantifying workers' reputation, we propose IM-LDP, an incentive mechanism for MCS based on local differential privacy, which includes four mechanisms of incentive, reputation, data perturbation and data aggregation. Specifically, incentive mechanisms are able to select workers who can provide more accurate data and compensate themselves for their privacy costs. The reputation mechanism quantifies the workers' reputation to improve their payments. The data perturbation mechanism ensures the tradeoff between the data privacy and aggregation accuracy, and the data aggregation mechanism generates highly accurate aggregation results. We evaluate the proposed IM-LDP through theoretical analysis and extensive experiments.