A Dynamic Scheduling Algorithm for Improving Performance Index in Multiprocessor Real-time Systems

In real-time systems, there exists a trade-off between schedulability and reliability. Most of the existing faulttolerant scheduling algorithms for real-time systems tend to favor schedulability over reliability. This may not be desirable when tasks have wide-varying utility values. This motivated the need for performance-oriented reliability measures, such as performance index, which inherently captures both schedulability and reliability in a single integrated metric. In this paper, we propose a dynamic scheduling algorithm for multiprocessor real-time systems with an objective of maximizing the performance index of the system.