A Geometric Approach for Generalized Encrypted Control of Networked Dynamical Systems

This grant focuses on establishing methods to enhance cybersecurity for networked motion-control systems utilizing somewhat homomorphic encryption. Industry 4.0 will transform the conventional automation systems to efficient cyber-physical systems by taking advantage of today’s information technology. Rapidly transforming automation system architecture introduces cybersecurity risks that did not exist in the past.

While protection of cyber-physical systems at the communication level has been extensively studied, there is a void in the study of protection at the motion control level. Allowing malicious system identification and data breach attacks to a motion controller would result in a) leaking of controller architecture, gains, and models, b) interception of motor commands and monitoring signals, and c) system disruption due to falsification of controller gains.

The projected approach will encrypt control algorithms, sensor signals, model parameters, and feedback gains, and perform necessary computation of motion commands to servo systems in the ciphertext space without a security hole. The project in particular studies encrypted motion control systems including unilateral remote assembly systems and bilateral teleoperation systems.

Success of this research project will foster innovation by changing current motion control practices for configuring networked automation systems, which would have a positive impact on the advancement of the future infrastructure and factory automation. The project will transform the current education in system dynamics and control to that with a clear and strong emphasis on cybersecurity.

The project will conduct unique interdisciplinary research covering system dynamics, control, communication, and cybersecurity engineering. The encrypted control framework configures motion systems with improved robustness against data breach and system identification attacks. Factors that limit performance in terms of computational load and signal quantization will be characterized.

In particular, security parameters such as cryptographic key lengths associated with somewhat homomorphic encryption will be determined by analyzing the trade-off between security and computational stability. Expression tree methods will efficiently characterize how different algebraic expressions of the control scheme impact the integrity of somewhat homomorphic encryption.

Dynamic key management and attack detection based on disturbance observation will be developed as additional security measures. Efficacy of the technique will be demonstrated using the networked control platform. In the future, the encrypted control framework may be combined with multilayered security measures with monitoring of process behaviors.

This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.