AdDQ: Low-Energy Hardware Replication for Real-Time Systems through Adaptive Dual-Queue Scheduling

AdDQ: Low-Energy Hardware Replication for Real-Time Systems through Adaptive Dual-Queue Scheduling

Abstract

Low energy consumption and high reliability are two major design objectives for real-time embedded systems. Beside other techniques, hardware replication (e.g. standby-sparing) can provide high reliability while keeping the energy consumption under control. In this paper, we consider two replicated processors as a standby-sparing system where main copy tasks on primary are scheduled by Earliest-DeadlineFirst (EDF) while backup copy tasks on spare are scheduled by our proposed Adaptive Dual-Queue (AdDQ) scheduling. AdDQ provides the best opportunity to postpone the spare executions as much as possible to minimize execution overlaps between main and backup copy tasks. Therefore, when a copy task finishes successfully a larger portion of its corresponding copy task can be cancelled, resulting in a significant amount of energy saving. To achieve further energy saving, we use Dynamic Voltage Scaling (DVS) and, Dynamic Power Management (DPM). The main reason of using DPM is that, once a copy of task has finished successfully, its other copy task is terminated, and if there is no more task for execution the processors go to a low-power mode. We evaluated our AdDQ technique under various system configurations. Experiments show that AdDQ provides up to 36% (on average by 14%) energy savings compared to four state-of-the-art techniques.

References