Real-time computer systems perform computations and I/O operations in a time frame defined by processes in the external environment. This research addresses the class of real-time systems characterized by the existence of strict timing constraints that must be satisfied if the system is to function correctly. The goal is to develop methods for assessing and ensuring the temporal correctness of such real-time systems. There are two main thrusts to this research. The first is the creation and analysis of more pragmatic models of real-time systems. The goal is to extend the theory of deterministic scheduling and resource allocation to encompass more realistic assumptions about program behaviors and to accommodate new forms of timing constraints motivated by the development of technologies such as digital video and high speed local area networks. The second focus is on the design and construction of operating system kernels that efficiently support the process model and scheduling policies indicated by the analysis of the resource allocation problems. The goal is to provide a facility for the predictable execution of real-time applications.
