An Energy-Efficient Fault-Tolerant Scheduling Scheme for Aperiodic Tasks in Embedded Real-Time Systems

This paper investigates how to combine fault tolerance with power management in embedded real-time systems at the same time. Fault tolerance is achieved by check pointing, and power management is carried out via dynamic voltage and frequency scaling (DVFS). Also, we take tasks' average switched capacitances into consideration. We present a fault-tolerant schedulability analysis for a periodic tasks and derive the optimal number of checkpoints. The optimal number of checkpoints can help the task to guarantee the timing constraints and minimize the worst case execution time in the presence of faults. We then propose a scheduling scheme which carries out DVFS on the basis of the schedulability analysis for the problem of static task scheduling and voltage allocation. The problem is addressed and formulated as a linear programming (LP) problem. The simulation results show that an increase of the number of variable voltages can reduce energy consumption. Also, selecting suitable voltages for tasks can lead to drastic energy reduction even if the number of variable voltages is very small.