Exploiting Dynamic Workload Variation in Low Energy Preemptive Task Scheduling (0710.4758v1)

Abstract: A novel energy reduction strategy to maximally exploit the dynamic workload variation is proposed for the offline voltage scheduling of preemptive systems. The idea is to construct a fully-preemptive schedule that leads to minimum energy consumption when the tasks take on approximately the average execution cycles yet still guarantees no deadline violation during the worst-case scenario. End-time for each sub-instance of the tasks obtained from the schedule is used for the on-line dynamic voltage scaling (DVS) of the tasks. For the tasks that normally require a small number of cycles but occasionally a large number of cycles to complete, such a schedule provides more opportunities for slack utilization and hence results in larger energy saving. The concept is realized by formulating the problem as a Non-Linear Programming (NLP) optimization problem. Experimental results show that, by using the proposed scheme, the total energy consumption at runtime is reduced by as high as 60% for randomly generated task sets when comparing with the static scheduling approach only using worst case workload.

• The paper introduces an offline scheduling algorithm that integrates ACEC and WCEC to optimize energy consumption in preemptive systems.
• It formulates the scheduling challenge as a nonlinear programming problem to effectively use slack time while ensuring deadlines.
• Experimental results demonstrate up to 60% energy reduction through tests on synthetic and real-life application tasks.

Low Energy Preemptive Task Scheduling through Dynamic Workload Variation

This paper articulates a strategy designed to minimize energy consumption in real-time embedded systems (RTES) by leveraging dynamic workload variations. It addresses an offline voltage scheduling technique for fully-preemptive systems, optimizing energy savings by combining the Average Case Execution Cycles (ACEC) with Worst Case Execution Cycles (WCEC) during the scheduling phase. This method diverges from traditional static scheduling approaches that rely solely on WCEC.

Methodology and Problem Formulation

The authors propose a novel offline scheduling algorithm formulated as a Non-Linear Programming (NLP) problem. The approach aims to capitalize on slack time by anticipating that tasks often require fewer cycles than the WCEC. The methodology constructs a schedule prioritizing energy efficiency under average execution cycles while assuring deadline adherence when WCEC is necessary.

The paper assumes a frame-based preemptive hard real-time system utilizing a rate monotonic scheduling policy — a prevalent model in RTES. By constructing a fully preemptive schedule, the method identifies optimal workload divisions for task instances and their sub-instances. The voltage scaling is then derived from the end-times of these sub-instances obtained through the scheduling algorithm.

Results

The experimental results demonstrate the effectiveness of this approach, achieving up to 60% energy consumption reduction compared to traditional static scheduling strategies. The experimental setup included randomly generated task sets and real-life applications, specifically CNC and GAP systems, confirming the robustness and applicability of the proposed method across different use cases.

Implications

This work contributes significantly to the field of low-power computing in embedded systems, providing a compelling alternative to static voltage scheduling. The integration of ACEC into the offline scheduling phase enhances slack utilization, leading to substantial runtime energy savings. The principal innovation lies in the simultaneous consideration of ACEC and WCEC for determining optimal end-times, paving the way for methodologies applicable to a broader class of systems, including those with non-preemptive architecture.

Speculations for Future Research

Further research could explore extending this approach to accommodate more complex task dependencies, multiple processors, and other voltage scaling constraints such as thermal design power limits. Additionally, the integration of online learning mechanisms to refine workload distribution predictions dynamically could refine accuracy, allowing further efficiency improvements.

In conclusion, this paper presents an NLP-based framework for preemptive task scheduling that effectively reduces energy requirements in RTES without compromising real-time performance. It opens new avenues for energy-efficient scheduling strategies by transcending beyond worst-case assumptions to exploit average workload scenarios.