Abstract: Cyber-physical systems (CPS) -- such as cars, factory control networks, medical devices, and power plants -- interact directly with the physical world. Because of this, CPS need to be highly reliable: when they fail, the consequences can be serious, including physical damage and loss of life. However, the CPS landscape has been changing rapidly: complexity has been increasing, and new features (such as cloud computing, multicore, or virtualization) are being added in rapid succession. As a result, existing reliability techniques are slowly reaching their limit, e.g., in terms of scalability and efficiency.
In this talk, I will present three examples of my recent work in this space. I will describe 1) a way to add real-time capabilities to cloud platforms, to make them compatible with CPS; 2) a technique for diagnosing subtle timing bugs in complex CPS, using a new variant of network provenance; and 3) a radically different security technique for CPS that leverages CPS-specific properties to provide resilience against attacks.
Bio: Linh Thi Xuan Phan is an Associate Professor of Computer and Information Science at the University of Pennsylvania. She works in cyber-physical systems and distributed systems. Her research focuses on theoretical foundations and systems techniques for safety, performance, and security guarantees. Her research has received several awards, including the NSF CAREER Award, the Dean's Graduate Excellence Award (from NUS), best paper and outstanding paper awards at RTAS, and best paper award nominations at RTSS and EMSOFT conferences. Phan has been serving on the Executive Committee of the IEEE Technical Committee on Real-Time Systems (since 2016) and on the ACM Future of Computing Academy, and she is currently the Secretary-Treasurer of ACM SIGBED.