Privacy Preservation by Public Design in Cloud‐Based Cooperative LQG Control Systems

ABSTRACT In this paper, we study a cooperative linear quadratic Gaussian (LQG) control system realized by a cloud‐based structure, consisting of a user and a server. In this system, the user runs a plant to control and employ the server, sharing necessary information with it to obtain the optimal LQG control inputs. However, the user considers its state trajectories as private. Therefore, we propose a privacy scheme utilizing a public privacy processor, whose certain parameters are also open to the server. This scheme manages to make the server unable to directly access the states of the user, so that the user's privacy is preserved; meanwhile, the server's public knowledge of the privacy processor allows the user to have satisfactory performance of LQG control. We then analyze the performances of privacy preservation and LQG control. Lastly, we propose an optimization problem to analyze the trade‐off between privacy and LQG control, which we solve using an efficient numerical method.