Non-preemptive scheduling of periodic tasks with specified release times

Abstract

Hard real-time systems are characterized by the fact that the execution of the tasks must be completed in time and must be done correctly. Thus, the task scheduling which allocates tasks to resources in such a way that their executions are being completed before the deadlines, is one of the most important activities in real-time systems, such as multimedia applications. Many of multimedia applications in distributed systems should transmit continuous audio/video non-preemptively to preserve the continuity across networks when clients request. Typically, messages for continuous media are heterogeneous and are split into periodic, different sized packets with deadlines. In this thesis, we present a static scheduling strategy and a dynamic scheduling algorithm to improve the schedulability of non-preemptive scheduling for periodic tasks with specified release times, known as a {\it NP-hard} problem. Moreover, we apply them to multimedia applications in distributed systems. First, we present the Time-Division Scheduling Strategy (TDSS) for the static non-preemptive scheduling. The TDSS transforms the given problem of non-preemptive scheduling for periodic tasks with specified release times into two subproblems of that with same release times within specific time intervals. Two time intervals for scheduling are computed with release times and periods. The TDSS finds two subsolutions for two respective subproblems and combines them into a complete solution of the given problem by using the repetitiveness and the predictability of periodic tasks. Therefore, the TDSS makes the previous optimal algorithm to find optimal solutions for the restricted periodic tasks with specified release times. For the general periodic tasks with specified release times, the TDSS gives higher schedulabilities than the previous scheduling strategy by the forward or the backward scheduling. Second, we present the Precalculated Deadline-Miss Avoidance (PDMA) algorithm which is a heuristic al...