Impact of Synthetic Task Set Generation Methods on Schedulability Performance
Saehwa Kim · Journal of Computing Science and Engineering · 2021
This paper addresses the various alternative methods of synthesizing task sets even when the continuous uniform distribution of their task utilizations is guaranteed. There are four methods that have been widely used in literature; LinearC, LinearT, LogT, and HarmonicT: C and T represent the worst-case execution times and periods, while linear, log, harmonic represent the spaces for the random generation of C or T. We have demonstrated that the schedulability performances of the task sets generated by those methods are very different. Specifically, the schedulability performance for the fixed priority scheduling is in the decreasing order of LogT, HarmonicT, LinearC, and LinearT. We have introduced notions of C-difference and T-difference, which have been used to demonstrate that the larger the value induced the better schedulability performance.