|제목||[세미나] Towards Predictable Real-Time Performance in a Cyber-Physical World|
제목: Towards Predictable Real-Time Performance in a Cyber-Physical World
연사: Hyoseung Kim (김효승 교수)
소속: University of California, Riverside
일시: 7/25 화요일 11:00 - 12:00
장소: 제2공학관 26312
Abstract: Cyber-physical systems integrate sensing, computing, communication and actuation capabilities to monitor and control operations in the physical environment. A key feature of such systems is the need to have predictable real-time performance: the system’s performance should be analyzable and guaranteed to meet timing constraints imposed by the physical world. This requirement for timely execution is particularly important for safety-critical applications, such as automotive, aerospace, defense, and manufacturing.
This talk focuses on the challenges arising from the use of advanced computing technologies in cyber-physical systems. Even as of today, modern multi-core platforms are rarely used in safety-critical applications primarily due to the performance interference caused by contention on various hardware resources shared among processor cores, such as caches, memory buses, I/O devices, and GPUs. Such interference is hard to predict and can significantly increase task execution time, e.g., up to 12x on a commodity quad-core platform. Our research adopts a unifying approach that coordinates analytical techniques with abstractions in systems software to predict and bound performance interference on multi-core platforms. We also apply this notion of isolation with our software primitives and algorithms for operating systems and hypervisors. We finally provide an illustration of our technologies in the context of a self-driving car.