The Center for Education and Research in Information Assurance and Security (CERIAS)
The Center for Education and Research inInformation Assurance and Security (CERIAS)
Principal Investigator: Inseok Hwang
Intelligent Transportation Systems traditionally consider ground-based vehicles operating on roads, streets, and highways. However, substantial technological advances in electrically powered aerial vehicles have paved the way to incorporate low-altitude, on-demand air transportation services to reduce the economic costs and commuting burdens of an ever-increasing ground traffic. This newly emerging transportation system is called the Urban Air Mobility (UAM) system. Envisioned as a highly automated, decentralized, cooperative passenger and cargo-carrying air transportation system, the UAM is expected to accommodate on-demand services, such as cargo delivery, medical service, and passenger transportation, in large cities. With such a large dependence on the interconnectivity of its physical and cyber components, the UAM is a large-scale safety-critical cyber-physical system (CPS) whose operation in highly populated cities means that its failure consequences can be disastrous. Therefore, UAM integration and operations within the National Airspace System (NAS) require developing new safe and secure decision-making and control frameworks that are resilient to cyberattacks. However, any such developed framework must allow for UAM scalability, and give guarantees on safety and cybersecurity while accommodating the inherent heterogeneity (of dynamics, physical constraints, communication network topologies, resource constraints, etc.) of the subsystems within the UAM.
To address the technical challenges toward safe and secure integration and operation, we first develop an explicit mathematical model of the UAM CPS under different cyberattacks. An explicit multi-agent system model of the UAM CPS is developed to describe the physical and logical behavior of the heterogeneous decentralized UAM system, which facilitates the identification and evaluation of different cyberattack pathways and the inherent structural vulnerabilities of the UAM CPS. Unlike random disturbances or faults that can disturb the UAM CPS, the configuration of cyberattacks (e.g., Denial-of-Service (DoS), Man-in-the-Middle (MITM), False-Data-Injection (FDI), etc.) is dependent on the attacker’s knowledge and intent and can be sophisticatedly designed to degrade the safe and efficient performance of UAM operations.
With this model in hand, we use an array of mathematical tools and techniques to i) detect and identify the different types of cyberattacks that can disrupt UAM operations, ii) analyze the security and safety risks and potential consequences of these cyberattacks on UAM operations, and iii) develop resilient attack detection and mitigation algorithms for risk and contingency management to allow safe, secure, and efficient large-scale UAM operations. Such algorithms are designed to allow for theoretical and experimental analyses on the performance of the UAM CPS with respect to resiliency to cyberattacks of varying severities. To detect and identify cyberattacks, we leverage state-of-the-art machine learning technologies to develop artificial intelligence (AI)-driven watchdog procedures that can be trusted to monitor UAM behavior in real-time for cyberattack intrusions. To mitigate the effects of these cyberattacks, we use a combination of techniques on adaptive estimation and reachable set analysis to avoid unsafe paths of malicious agents in the UAM, and self-organized topology reconfiguration for resiliency to cyberattacks. Furthermore, we benchmark, through rigorous mathematical analyses and in-house high-fidelity UAM simulations developed by our industry partners, the resilient performance of these algorithms for the operation of the UAM CPS in the presence of varying severities of cyberattacks.
All things considered, we design and develop a suite of resilient and risk-mitigating decision-making and novel control algorithms with benchmarked resiliency to varying severities of cyberattacks for theoretical and experimental validation before real-time UAM deployment.
Other PIs: Dengfeng Sun, Shaoshuai Mou, Mahmoud Mahmoud (NCAT)
Students: Sounghwan (Eric) Hwang, Shanelle Clarke, Omanshu Thapliyal, Chan-Yuan (David) Kuo, S M Nahid Mahmud, Mohammed MynUddin (NCAT)
O. Thapliyal and I. Hwang, “Learning based Cyberattack Design and Defense for Supervisory Control Systems”, 2021 European Control Conference, Rotterdam, Netherland, June 29- July 2, 2021
Richmond Asiedu Agyapong, Mahmoud Nabil, Abdul-Rauf Nuhu, Mushahid I. Rasul and Abdollah Homaifar, “Efficient Detection of GPS Spoofing Attacks on Unmanned Aerial Vehicles Using Deep Learning”, 2021 IEEE Symposium Series on Computational Intelligence (SSCI)
G. Clarke, O. Thapliyal, S. Hwang, and I. Hwang, “Attack-Resilient Distributed Optimization-based Control of Multi-Agent Systems with Dual Interaction Networks”, 2022 AIAA SciTech Forum: Intelligent Systems, San Diego, CA, January 2022
Jiazhen Zhou, Dawei Sun, Inseok Hwang, Dengfeng Sun, “Control Protocol Design and Analysis for Unmanned Aircraft System Traffic Management”, IEEE Transactions on Intelligent Transportation Systems, 2021.
Jiazhen Zhou, Dengfeng Sun, “Safe Link Transition for Unmanned Aircraft System Traffic Management”, AIAA SCITECH 2022 Forum.
Y. Xie, S. Mou and S. Sundaram, "Towards Resilience for Multi-Agent QD-Learning," 2021 60th IEEE Conference on Decision and Control (CDC), Austin, Texas, 2021.
Keywords: Control Theory, Cyber-physical system, Cybersecurity, Urban Air Mobility