A Fluid-Flow Interpretation of SCED Scheduling

TL;DR

This paper introduces a fluid-flow perspective on SCED scheduling that simplifies deadline calculations by leveraging an isomorphism between min-plus and max-plus calculus, avoiding complex pseudo-inverse computations.

Contribution

It presents a novel fluid-flow interpretation of SCED scheduling that enables straightforward deadline derivations for general convex or concave service curves.

Findings

Deadlines can be computed without pseudo-inverses.

Applicable to convex and concave piecewise linear service curves.

Simplifies the analysis of SCED scheduling.

Abstract

We show that a fluid-flow interpretation of Service Curve Earliest Deadline First (SCED) scheduling simplifies deadline derivations for this scheduler. By exploiting the recently reported isomorphism between min-plus and max-plus network calculus, and expressing deadlines in a max-plus algebra, deadline computations no longer require pseudo-inverse computations. SCED deadlines are provided for general convex or concave piecewise linear service curves.