A critical path approach for testing distributed real-time systems

The distinguishing feature of real-time computing is its insistence on predictable timing behavior. The authors introduce a novel approach to testing timing criteria of distributed real-time systems. They describe major problems related to estimating the worst case response times of distributed real-time tasks, concentrating on the blocking time due to limited resource availability. A static analysis approach is used to generate time-critical program paths based on a distributed system control flow graph. The authors present a three-step algorithm which, when applied to a task system flow graph, creates the corresponding execution graph. The properties of the task oriented execution graph allow for easy generation of time-critical distributed program paths. These paths, for which input data could be generated automatically, are then used for testing the timing behavior of the system. The authors also define and solve a maximum simultaneous request subset problem for a periodic task set. The solution of this problem allows for significant reduction of time-infeasible program paths. The method described is part of an integrated testbed project, designed for testing and fine tuning all relevant parameters of a distributed real-time system