Encrypted control for privacy-preserving and secure cyber-physical systems

点击翻译按钮获取中文摘要

Driven by advances in cloud computing and wireless communications, many control systems are increasingly integrated with sensing, communication, and computing devices, leading to the so-called cyber-physical systems (CPS). Typical examples of CPS include critical water, gas, and power infrastructures, swarm robotics, and vehicle platoons. Those control systems usually have their feedback control loops closed via communication networks, leading to critical privacy and security challenges due to the vulnerability of (shared) communication networks. In fact, concerns on security and particularly privacy have spurred various legislative polices and recommendations for many privacy-and-security critical CPS like smart grids and intelligent transportation systems. For example, the National Highway Traffic Safety Administration has required that privacy issues must be addressed when implementing vehicle-to-vehicle communications, and privacy measures must be undertaken to prevent individuals from being tracked in location. Although researchers in the control community and the computer science community have proposed several technological solutions to enhance the privacy and security of CPS, existing approaches fail to combine and synergize control-theoretical approaches with conventional information technology solutions like encryption, which significantly restricts the overall achievable strength and scope in protecting the privacy and security of CPS.This proposal seeks to provide an encrypted control approach for CPS privacy-and-security protection by seamlessly integrating control and encryption. The integration creates tremendous synergy and enables new capabilities: not only does it enable decentralized privacy-and-security protection without any data aggregator or third-party, it also avoids compromising accuracy, a typical problem with differential-privacy based privacy-protection approaches. Furthermore, the seamless integration guarantees low computational complexity as well. The project will: i) develop an encrypted control framework and corresponding theories for cloud-based control to protect privacy with guaranteed stability despite encryption-induced representation errors; ii) under the encrypted control framework, develop new verification mechanisms to combat integrity attacks on both communication links and shared computing platforms such as the cloud; iii) extend the encrypted control to completely decentralized multi-agent settings without any data aggregator or third-party; and iv) systematically validate the approach using a multi-robot platform. The proposed research will provide privacy-and-security protection for general CPS such as smart grids and connected vehicles, in which privacy and security are crucial. It is also expected to impact many other critical infrastructures such as water and wastewater systems where control over communication networks is becoming the norm. The project will significantly enrich existing graduate/undergraduate courses in cryptography and control, which the PIs regularly teach. It will also directly address the underrepresentation of ethnic groups and women in science and engineering through leveraging scheduled training activities under an existing NSF RTG grant as well as through ongoing outreach activities to minority middle-school girls under the Clemson WISE (Women in Science and Engineering) program.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.