MC-Fluid: rate assignment strategies

In this paper, we consider fluid scheduling of mixedcriticality implicit-deadline sporadic task systems. Fluid scheduling allows tasks to be allocated fractional processing capacity, which, although hard to implement, significantly improves the schedulability performance. For dual-criticality systems, dualrate fluid scheduling in which each task is assigned two execution rates depending on the system criticality level has been proposed in the past. An optimal rate assignment algorithm for such systems called MC-Fluid that assigns rates with polynomial complexity has also been proposed. Another rate assignment strategy called MCF has recently been proposed with linear runtime complexity. Although MCF results in a lower schedulability ratio when compared to MC-Fluid, it is shown that both the strategies result in a speed-up optimal scheduling algorithm for dual-criticality systems. We propose two new algorithms to assign execution rates called MC-Sort and MC-Slope, both with linearithmic (i.e., nlogn) complexity in the number of tasks n. The proposed algorithms have a schedulability ratio that is significantly better than MCF and almost as good as MC-Fluid, but with a reduced run-time complexity when compared to MCFluid.