Nov 29, 2016: Mohammadsadegh Mohagheghi: Improving Utility-Aware Scheduling Heuristics for Soft Real-time Systems with Stochastic Running Times

November 29, 2016Improving Utility-Aware Scheduling Heuristics for Soft Real-time Systems with Stochastic Running Times
Room: HB 2AMohammadsadegh Mohagheghi
12:30-13:30

Time Utility Functions are used to precisely specify the semantics of soft time constraints in many real-time cyber-physical systems. These functions characterize the utility of completing jobs as a function of time and can be considered as a generalization for the concept of classical deadlines.

The scheduling criteria of TUF-based soft real-time systems are based on optimizing accrued utility from those jobs, i.e. maximizing the expected sum of the jobs' attained utilities. Such criteria are called Utility Accrual (or UA) criteria and UA scheduling algorithms are those ones that consider UA criteria and should compute the optimal expected values and policies. Because of time and memory constraints that are usual in real-time embedded systems, a good heuristic should be used to compute near-optimal scheduling.

In this presentation we review some previous heuristics and improve them to give better approximations of the optimal scheduling. We also present an extension of these heuristics for battery based UA scheduling and compare our improvement with previous heuristics.