Secure Estimation With Privacy Protection

• Published in: IEEE transactions on cybernetics Vol. 53; no. 8; pp. 4947 - 4961
• Main Authors: Liang, Shi, Lam, James, Lin, Hong
• Format: Journal Article
• Language: English
• Published: United States IEEE 01.08.2023; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Covariance matrices; Estimation; Optimal estimator (OE); Optimization; Privacy; privacy protection; Probability density function; Random variables; Security; security estimation; stability; State estimation; suboptimal estimator (SE)
• ISSN: 2168-2267; 2168-2275; 2168-2275
• DOI: 10.1109/TCYB.2022.3151234

In this article, we focus on the state estimation problems for a system with protecting user privacy. Regarding whether the user has conducted a sensitive action in the system as a kind of privacy, we propose a privacy-preserving mechanism (PPM) to prevent its action results from being disclosed or inferred. For such a system with the PPM, we first obtain the optimal estimator (OE). Subject to the inoperability of the OE in practice, we turn to designing a computationally efficient suboptimal estimator (SE) as an alternative. Then, we prove that this SE can remain stable while satisfying the user's requirements on both privacy protection and estimation performance. By solving a privacy-preserving optimization problem, a set of guidelines is established to customize a tradeoff between privacy and performance according to the user's demand. Finally, illustrated examples are used to illustrate the main theoretical results.