A methodology for predictable real-time system design

Timing predictability and analyzability are key requirements of real-time systems. As a result, a temporal model of the system must be used during the requirements, design and implementation phases. The temporal model is useful in representing latencies and delays associated with scheduling and data transmissions in distributed systems and for modeling various processor and network elements. Such a temporal model is directly applicable to all phases of system development: (a) the methodical capture of end-to-end predictability and performance requirements during the requirements phase, (b) the structuring of tasks, critical sections and shared data during the design phase, (c) system profiling and performance optimizations during the implementation phase. This paper discusses a state-of-the art methodology, a widely used temporal model called the rate-monotonic analysis (RMA) and TimeWiz/sup TM/, a commercially available integrated solution that allows uniform modeling, analysis and simulation of processing and networking components used in distributed real-time systems.