Industrial process control systems increasingly rely on cloud computing and third-party platforms for computational efficiency, yet this shift introduces significant security and privacy risks. Sensitive operational data, control algorithms, and system parameters could be exposed to unauthorized access or misuse. A research team has addressed this challenge by developing an encrypted model predictive control system that maintains full protection of both process information and control logic throughout operation.
The approach utilizes polynomial approximation techniques to express the optimal control law in a form compatible with fully homomorphic encryption, a cryptographic method that allows computation on encrypted data without requiring decryption. This enables the control algorithm to execute on cloud or third-party platforms while keeping all sensitive information encrypted at all times.
A critical innovation in this work is preserving the mathematical properties essential for safe operation. Traditional encrypted control methods struggle to maintain closed-loop stability and recursive feasibility—the guarantee that the controller can always find valid solutions within physical constraints. The proposed method overcomes these limitations by carefully structuring the polynomial approximation to retain these stability guarantees even under encryption.
Laboratory-scale experimental validation confirms that the privacy-aware control method successfully operates on simulated cloud infrastructure without performance degradation. The system maintains accurate control of process variables while keeping both the control law and operational data completely encrypted. Process operators never expose unencrypted sensitive information to the computing platform.
This development has important implications for industrial automation, particularly for facilities managing proprietary processes or operating in competitive industries where process data represents competitive advantage. It enables organizations to leverage cloud computing's cost and scalability benefits while maintaining stringent cybersecurity and privacy standards. As industrial systems become increasingly connected and distributed, encrypted control methods will likely become essential infrastructure for secure modern manufacturing and process operations.
