Bandwidth allocation for fixed-priority-scheduled compositional real-time systems

Abstract

Recent research in compositional real-time systems has focused on determination of a component's real-time interface parameters. An important objective in interface-parameter determination is minimizing the bandwidth allocated to each component of the system while simultaneously guaranteeing component schedulability. With this goal in mind, in this article, we explore fixed-priority schedulability in compositional setting. First we derive an efficient exact test based on iterative convergence for sporadic task systems scheduled by fixed-priority (e.g., deadline monotonic, rate monotonic) upon an explicit-deadline periodic (EDP) resource. Then we address the time complexity of the exact test by developing a fully-polynomial-time approximation scheme (FPTAS) for allocating bandwidth to components. Our parametric algorithm takes the task system and an accuracy parameter ε > 0 as input and returns a bandwidth which is guaranteed to be at most a factor (1 + ε) times the optimal minimum bandwidth required to successfully schedule the task system. We perform thorough simulation over synthetically generated task systems to compare the performance of our proposed efficient-exact and the approximate algorithm and observe a significant decrease in runtime and a very small relative error when comparing the approximate algorithm with the exact algorithm and the sufficient algorithm.