A new dynamic scheduling algorithm for real-time multiprocessor systems

Abstract

Real-time systems require both functionally correct executions and results that are produced in time. Thus, the scheduling algorithm is an important component of these systems. Several dynamic scheduling algorithms for real-time multiprocessor systems using heuristic approaches such as famous myopic algorithm and its variations have been proposed. However, the task assignment policies used by all these scheduling algorithms hinder the improvement of the scheduling success ratio. In this paper, we propose a new dynamic scheduling algorithm, called "thrift algorithm", for real-time multiprocessor systems. By using a new task assignment policy in this algorithm, thrift algorithm improves the scheduling success ratio successfully. To study the effectiveness of thrift algorithm, we have conducted extensille simulation studies and compared its scheduling success ratio with that of myopic algorithm when several task parameters are changed. Simulation results demonstrate that the scheduling success ratio of thrift algorithm is superior to that of myopic algorithm. © 2001 by Springer Science+Business Media New York.