An optimal resource sharing protocol for generalized multiframe tasks

We consider sharing of non-preemptable resources in real-time task models with flexible job release patterns. Resource sharing is an inherent property of many real-time systems. At the same time, flexible task models are needed to precisely express their workloads. Exact analysis and optimal scheduling of systems with shared resources have been available only for relatively simple task models, such as the sporadic task model. We propose a new algorithm for scheduling systems with shared resources. The key idea behind the algorithm is to take the tasks’ structures into account when predicting possible resource contention. We show that the algorithm is optimal for scheduling generalized multiframe task sets with shared resources. We also present an efficient feasibility test for such systems, and show that the test is both sufficient and necessary.