Cyber-physical systems have received a great deal of attention in recent years, but decades after we began to use computers to control physical systems. We will start with a few remarks on the definition of cyber-physical system and challenges in the field. We will then go on to describe two research projects on CPS at very different levels of physical granularity.
At the small end of the scale, we explored the relationship between schedulability and stability in control systems in collaboration with Prof. Fumin Zhang’s group. Both control theory and real-time systems theory traditionally view the sample period as involate. However, most real-time scheduling algorithms do introduce jitter in the execution interval of tasks. Until recently, the relationship between the schedulability of tasks and the stability of the control system had not been explored. We used Lyapunov analysis to show that, for a class of simple control systems running under rate-monotonic analysis, the goals of stability and schedulability are, in fact, consistent.
At the large end of the scale, my student Umer Tariq and I are collaborating with Prof. Santiago Grijalva’s group on distributed system architectures for smart energy grids based on the prosumer model. We have developed a multi-layer architecture for distributed control of smart energy systems based on a Web services model. At the RTOS layer, we use an API translator to map generic software onto target RTOSs. At the Web level, we have developed a model for hard real-time Web services.