Privacy for Switched Systems Under MPC: A Privacy-Preserved Rolling Optimization Strategy

Area of Science:

• Control Systems Engineering
• Cybersecurity
• Applied Mathematics

Background:

• Differential privacy methods often add noise to transmitted data, potentially compromising control system accuracy.
• Data accuracy is critical for the performance of control systems.

Purpose of the Study:

• To propose a novel privacy-preserved rolling optimization strategy (PP-ROS) for switched systems.
• To address the challenge of maintaining data accuracy while ensuring privacy in control systems.

Main Methods:

• Developed PP-ROS by adding Laplace noise to prediction and control horizons, not transmitted data.
• Presented privacy definitions for horizons and designed a private model predictive control (P-MPC) controller.
• Provided methods to prove and calculate privacy levels for control input and system output.

Main Results:

• The PP-ROS and P-MPC ensure privacy of controller parameters, control input, and system output.
• Adding noise to horizons effectively preserves data accuracy compared to traditional methods.
• Simulation examples demonstrate the availability and benefits of the proposed strategy.

Conclusions:

• The PP-ROS and P-MPC offer a viable solution for privacy-preserving control in switched systems.
• The proposed method effectively balances data privacy with system performance and accuracy.