Publication (Conferences and Seminars)

Energy-based fair queuing scheduling implementation for battery-limited mobile systems

Wei, Jianguo; Ren, Rong; Juárez Martínez, Eduardo; Pescador del Oso, Fernando

Abstract:

Energy-based fair queuing (EFQ) is an implementation of the classical fair queuing scheduling algorithm in the energy domain. It is an energy-centric scheduling algorithm that manages energy as the first-class resource that is globally shared by different devices in the system. Recent high-level simulation and Linux-based implementation and experiment have proven that EFQ can achieve proportional power sharing by accounting the energy consumption on both CPU and I/O operations and scheduling tasks based on their normalized energy consumptions. It has also been demonstrated that EFQ can achieve effective and flexible time-constraint compliance. In this paper, the core Linux implementation of EFQ is introduced in detail, and the experimental platform as well as the benchmarks are presented for the experiments. Experimental results on EFQ scheduling show that it is more effective and flexible than the default Linux scheduler in optimizing the user experience of battery-limited systems.

Research areas:

• Image and Video Technologies

Year:

2014

Type of Publication:

Conferences and Seminars

Keywords:

CPU operation; EFQ; I-O operation; Linux; Linux-based implementation; battery-limited mobile systems; Energy consumption; energy domain

Editor:

IEEE

Organization:

XXIX Conference on Design of Circuits and Integrated Systems (DCIS 2014)

ISBN:

978-1-4799-5743-9

DOI:

10.1109/DCIS.2014.7035594