Fixed-Priority Scheduling Policies and Their Non-utilization Bounds

Abstract

Utilization bounds for schedulability are one of the simplest approaches for admission control in real-time systems. Even though they are not resource optimal, they offer the advantages of simplicity and computational efficiency. Liu and Abdelzaher proposed a universal bound that was customized to any fixed-priority scheduling policy by choosing the corresponding load metric. In this paper, we extend Liu and Abdelzaher’s work by combining different fixed-priority scheduling policies, and give their non-utilization bounds. We have considered the combination of deadline monotonic scheduling (DMS), shortest-job-first scheduling (SJF), and velocity monotonic scheduling (VMS). The reason why we take this combination is because we can take advantage of the merits of each scheduling policy and at the same time overcome their shortcomings.