Towards Personalized Privacy Preference Aware Data Trading: A Contract Theory Based Approach

A large number of IoT devices geographically distributed in urban areas makes it attractive to collect massive data in a crowdsourced manner. To fully exploit the value of crowdsourced data, there is a significant growth in demand for data trading recently. However, data trading puts data owners at risk of privacy breaches. Existing studies based on centralized differential privacy are impractical due to the assumption of a trustworthy data collector, high risk of privacy leakage and extensive communication cost, let alone data owners’ personalized requirements on privacy protection. To this end, we propose a contract theory based personalized privacy-aware data trading approach which provides a set of optimal contracts specifying different privacy-preserving levels and data trading prices to selfish data owners who upload perturbed data according to negotiated privacy-preserving level, and finally aggregates data using a group-weighted maximum likelihood estimation method. The proposed private data trading approach not only achieves desirable data utility in terms of accuracy but also satisfies budget feasibility, individual rationality, and incentive compatibility through theoretical analysis and extensive experiments.