Fair lateness scheduling: reducing maximum lateness in G-EDF-like scheduling

Abstract

Existing research in soft real-time scheduling has focused on determining tardiness bounds given a scheduling algorithm. In this paper, we study lateness bounds, which are related to tardiness bounds, and propose a scheduling algorithm to minimize lateness bounds, namely the global fair lateness

Minimizing response times of automotive dataflows on multicore

“…The most well-known window-constrained algorithm is G-EDF, which uses the deadline of each job as its PP. However, Erickson et al [9] demonstrated that allowing PPs set earlier than deadlines can result in better response-time bounds. The G-FL scheduler, mentioned in Sec.…”

“…Recently, Erickson et al showed that "G-EDF-like" (GEL) schedulers, such as the global fair-lateness (G-FL) scheduler, have provably smaller response-time bounds than G-EDF [9]. Under GEL schedulers, priority points (PPs) distinct from deadlines are defined to minimize worst-case response times.…”

Reducing Tardiness under Global Scheduling by Splitting Jobs

“…An improved, but more complex, bound was introduced in [9]. While these works focused on G-EDF itself, [8] proposed G-FL as a new scheduler with analysis similar to G-EDF in [9]. G-FL usually provides a smaller maximum lateness bound for the task system.…”

“…One of the advantages of G-FL promoted in [8] is that it is a job-level static priority (JLSP) algorithm, which is important for synchronization mechanisms such as those discussed in [3]. However, when splitting is introduced, the algorithm is no longer truly JLSP.…”

“…In prior work on tardiness bounds, the global earliest deadline first (G-EDF) scheduler and the improved global fair lateness (G-FL) scheduler have received considerable attention [6,8,9,10]. G-FL is considered a G-EDF-like (GEL) scheduler because, like G-EDF, each job's priority is defined by a fixed point in time after its release.…”