Configuring Hard Real-Time Distributed Systems

Abstract

My work is concerned with configuring automatically distributed hard real-time systems. A large facet of this work is the development of analysis which can take a given configuration of a distributed hard real-time system and determining whether the timing constraints are met. To do this I have developed extensive scheduling theory, assuming a static priority pre-emptive scheduling algorithm. I have been largely successful in developing this theory, and can now determine worst-case response times (i.e. the time between a task being released and the task terminating after executing for a bounded of computation time) for a number of different task types. One of the most powerful and expressive pieces of theory allows a set of task transactions to be described: each transaction consists of a set of tasks which are released at fixed offsets in time relative to each other. The transactions can be used to implement distributed hard real time systems, and allow complex timing patterns to be described.