An Efficient Deadline and Period Assignment Scheme for Maintaining Temporal Consistency Under EDF

Abstract

In order to maintain real-time data freshness (or temporal consistency) while minimizing the imposed processor workload, much work has been done to address the period and deadline assignment problem for real-time update transactions, such as the More-Less scheme [18] for fixed priority scheduling, and the ${\mathcal{M}}{\mathcal{L}}_{\rm EDF}$ and ${\mathcal{H}}{\mathcal{S}}_{\rm EDF}$ algorithms [19] for dynamic priority scheduling. This paper studies the period and deadline assignment problem for update transactions scheduled by the earliest deadline first (EDF) scheme. Based on a sufficient feasibility test for EDF [3], we formulate the assignment problem as a Linear Programming problem and propose the Improved More-Less (${\mathcal{I}}{\mathcal{M}}{\mathcal{L}}_{\rm EDF}$) algorithm. ${\mathcal{I}}{\mathcal{M}}{\mathcal {L}}_{\rm EDF}$ can achieve a considerably low workload in a time-efficient manner. Extensive experiments are conducted and the results verify the efficiency of $\mathcal{I}\mathcal{M}{\mathcal{L}}_{\rm EDF},$ when compared with existing methods with comparable quality.

This work was partially supported by the National Science Foundation of China [Award No. 60873030] and the Research Fund for the Doctoral Program of the Ministry of Education of China [Award No. 20090142110023].