Abstract
Fundamentally, real-time computation is about interaction between software and physical dynamics in what many people today call `cyber physical systems.' CPS applications include much of today's industrial economy, including for example robotic manufacturing systems, self-driving and refuse-to-crash cars, electric power generation and distribution, process control in chemical factories, fly-by-wire aircraft, and military systems. The `real' in `real time' refers to the time of the physical dynamics. But how does that translate into a notion of time in the world of computation? In that world, the notion of real time encompasses a number of disparate concepts, including simply fast computation, or more interestingly, bounded execution time, computation on streaming data, temporal semantics in programs, and temporal semantics in networks. These concepts translate into significantly different approaches to design and implementation of cyber-physical applications. This talk will analyze each of these concepts and illustrate the potential consequences. I will focus particularly on safety-critical systems, such as self-driving cars and fly-by-wire aircraft, because these present the most daunting challenges.