A Stationary Random Process Based Privacy-Utility Tradeoff in Differential Privacy

Differential Privacy (DP) emerged as a promising solution to resist privacy attack and provided provable privacy guarantee. DP mechanisms typically introduce data perturbation to achieve almost equally statistical results regarding the presence or absence of an individual in a dataset. This perturbation in-volves adding noise to the dataset, however, it often compromises data utility despite its effectiveness in enhancing privacy. The critical parameter in determining the balance between privacy and utility is the privacy budget ε. But it lacks an effective approach to select an ε in current DP mechanisms. To address these challenges, we propose a novel privacy-preserving approach known as Differential Privacy with Stationary Random Process (DPSRP). We optimize the data privacy and utility of DP in the context of stationary random process. Specifically, we analyze the privacy and utility of DP within Markov chain. DPSRP facilitates the selection of an appropriate ε to achieve an optimal tradeoff between privacy and utility and yield high-quality differential private datasets. Extensive simulations conducted on real-world datasets validate the effectiveness of our theoretical results.