An Effective Dynamic Task Scheduling Algorithm for Real-time Heterogeneous Multiprocessor Systems

Abstract

Real-time systems require both functionally correct executions and the results that are produced in time. Fault-tolerance is an important requirement of such systems, due to the catastrophic consequences of not tolerating faults. In this paper, we propose an effective Load-driven Adaptive Scheduling Algorithm (LASA) to dynamically schedule real-time tasks with fault-tolerance used in heterogeneous multiprocessor systems. In LASA, it has an adaptive mechanism to monitor the processor utilization and determine the number of backup copies been scheduled. Besides, we also design a task deferment mechanism to improve the utilization of reclaimed computing resources from redundant copies. According to our simulations, LASA makes a trade-off between the guarantee ratio and the reliability of fault-tolerance, and obviously outperforms other related methods.