Privacy-aware control of cyber-physical systems: analysis and synthesis

In his thesis, Teimour Hosseinalizadeh explores privacy-preserving control mechanisms for Cyber-Physical Systems (CPSs), which connect physical systems to the internet. Firstly, Hosseinalizadeh investigated cloud-based Model Predictive Control (MPC). The analysis reveals vulnerabilities in existing methods, allowing the cloud to potentially infer sensitive system parameters. To address this, Hosseinalizadeh proposes a novel cloud-based data-driven control design that combines transformation-based methods and robust control theory. This approach preserves privacy for crucial system matrices while maintaining real-time performance.

Secondly, Hosseinalizadeh focused on protecting the initial state of a system in data-releasing scenarios. By introducing correlated Gaussian noise, he demonstrates the creation of "confusion sets" that hinder accurate state estimation by a monitoring center.

Finally, Hosseinalizadeh tackled privacy challenges in networks of agents that need to compute polynomials over their neighbors' private data. He presents a novel algorithm that utilizes Paillier encryption, secret sharing, and a new polynomial handling technique to enable accurate and privacy-preserving computations across the network.

In summary, this research contributes to the development of secure and private control strategies for CPSs, addressing critical concerns in areas like cloud computing, data sharing, and distributed control.

Dissertation: https://hdl.handle.net/11370/66f3fec0-f80f-4d9f-b7a2-9e5422a3b1ec