A fault-tolerant strategy for real-time task scheduling on multiprocessor system

With more and more multiprocessor computers coming to the market and applications, system reliability is becoming a key issue of multiprocessor systems. As there is an inherent characteristic of redundancy existing in multiprocessor systems, the way to increase the reliability is aimed at software strategy rather than hardware redundancy. Aiming at realizing fault tolerance and making better utilization of system resources, this paper presents a fault tolerant strategy to increase reliability for multiprocessor systems, which utilizes a common class of operating system algorithm referred to as Self-Scheduling (SS). Simulation results have been analyzed to show the feasibility of the proposed fault tolerant policy.