Generalized reliability-oriented energy management for real-time embedded applications

Abstract

DVFS remains an important energy management technique for embedded systems. However, its negative impact on transient fault rates has been recently shown. In this paper, we propose the Generalized Shared Recovery (GSHR) technique to optimally use the DVFS technique in order to achieve a given reliability goal for real-time embedded applications. Our technique determines the optimal number of recoveries to deploy as well as task-level processing frequencies to minimize the energy consumption while achieving the reliability goal and meeting the timing constraints. The recoveries may be shared among tasks, improving the prospects of DVFS compared to existing reliability-aware power management frameworks. The experimental evaluation points to the close-to-optimal energy savings of our proposed technique.

Shared recovery for energy efficiency and reliability enhancements in real-time applications with precedence constraints

“…To address the conservativeness of such an individual-recovery based approach, considering independent frame-based task model, a preliminary version of SHR technique for independent frame-based tasks was first proposed in Zhao et al [2009]. Recently, SHR has been further extended to generalized shared recovery (GSHR) technique [Zhao et al 2011] through which a small number of recovery tasks are shared by all the tasks while satisfying an arbitrary system-level target reliability.…”

Real-Time Scheduling of Embedded Applications on Multi-Core Platforms

“…Same as [128,131], the P ind is varied between [0.05, 0.35] for each node v i and the maximum slack-execution-ratio is fixed at S/C = 1.4. According to 7.2, the larger the P ind , the higher the energy consumption.…”

Energy-aware Fault-tolerant Scheduling for Hard Real-time Systems

“…Based on this model, they proposed a recovery scheme to schedule real-time tasks that can reduce energy consumption without degrading the reliability. They further proposed to reserve computing resources that can be shared by different tasks to improve the energy-saving performance [133]. These algorithms work only for frame-based realtime systems, i.e.…”

“…the probability that q faults are detected. If the failure distribution is modeled as a Possion process with a failure rate λ as in [136,134,133], then the reliability function is shown in equation (3.9),…”

“…We adopt the power model in [56,133] by considering the frequency-independent and frequency-dependent power components. Specifically, the overall power consumption P can be formulated as…”