Strategic Information Transmission Against Malicious Eavesdropper and Semi-Honest Estimator: A Tripartite Game Analysis

This paper is concerned with privacy-aware remote state estimation in the presence of a malicious eavesdropper and a semi-honest (or honest-but-curious) estimators via strategic information transmission. We first establish a tripartite game framework to analyze the relationship between the sensor, malicious eavesdropper, and semi-honest estimator, where the sensor is the leader in the game and designs the optimal encoder, and then the eavesdropper and estimator follow to design the decoder. Then we formulate the problem as the minimization of a global objective by the sensor subject to the trace of the malicious eavesdropper error covariance above a given threshold by designing the optimal encoding parameters. For the malicious eavesdropper possessing different encoding knowledge, we consider two wiretap scenarios of the limited and complete encoding information. In both scenarios, we transform the original problem into a constrained difference-of-convex optimization problem and obtain locally optimal solutions by applying the penalty convexconcave programming. Finally, numerical examples are given to illustrate theoretical results.