Towards Secure and Resilient Real-Time Cyber-Physical Systems

Date: Jan 13, 11:45am-12:45pm

Abstract: Recent advances in sensing, communication, and computing have revolutionized the accessibility and integration of safety-critical Cyber-Physical Systems (CPS). However, these systems face stringent timing constraints and escalating security challenges. In this talk, I will present my research on enhancing the resilienc e and security of CPS. To address temporal resilience, I will introduce a genera lized mixed-criticality systems (MCS) model and innovative techniques that impro ve MCS performance. Modeling the system as an MCS can enable secure recovery fro m memory-based cyber-attacks, ensuring safe operations in safety-critical applic ations. On the security front, I will discuss backdoor-resilient algorithms that secure the deployment of pretrained machine learning models in resource-constra ined CPS, addressing vulnerabilities in both in-domain and cross-domain inferenc e scenarios. Finally, I will showcase my research on timing analysis for the Rob ot Operating System (ROS 2) and a security analysis framework for virtual realit y applications using sensing technologies, emphasizing their impact on advancing CPS resilience and security.

Biographical Sketch: Abdullah Al Arafat is a Ph.D. candidate in Computer Science at North Carolina St ate University, advised by Dr. Zhishan Guo. His research focuses on real-time sy stems, robust learning, and cyber-physical systems. He has authored several publ ications in top-tier conferences, including RTSS, RTAS, EMSOFT, ICCV, ECCV, and CCS, with eight papers featured in CSRankings-listed venues. Notably, his DAC 20 22 paper was selected as a publicity paper. Abdullah’s work bridges the domains of real-time systems and machine learning, addressing challenges in secure and r esilient system design for safety-critical applications.

Location and Zoom link: Zoom only at https://fsu.zoom.us/j/94396085712