DVS scheduling in a line or a star network of processors

Research output: Journal Publications and Reviews (RGC: 21, 22, 62)21_Publication in refereed journal

3 Scopus Citations
View graph of relations

Author(s)

Related Research Unit(s)

Detail(s)

Original languageEnglish
Pages (from-to)16-35
Journal / PublicationJournal of Combinatorial Optimization
Volume29
Issue number1
Online published13 Oct 2013
Publication statusPublished - Jan 2015

Abstract

Dynamic voltage scaling (DVS) is a technique which allows the processors to change speed when executing jobs. Most of the previous works study either single processor or multiple parallel processors. In this paper, we consider a network of DVS enabled processors. Every job needs to go along a certain path in the network and has a certain workload finished on any processor it goes through before it moves on to the next processor. Our objective is to minimize the total energy consumption while finishing every job before its deadline. Due to the intrinsic complexity of this problem, we only focus on line networks with two nodes and a simple one-level tree network (a star). We show that in some of these simple cases, the optimal schedule can be computed efficiently and interleaving is not needed to achieve optimality. However, in both types of networks, how to find the optimal sequence of execution remains a big challenge for jobs with general workloads.

Research Area(s)

  • Dynamic voltage scaling, Networks, Optimization, Sequencing