Phase type approximation for optimal auto-sleep scheduling

  • Authors:
  • Hiroyuki Okamura;Tadashi Dohi;Shunji Osaki

  • Affiliations:
  • Department of Information Engineering Hiroshima University Higashi-Hiroshima, Japan;Department of Information Engineering Hiroshima University Higashi-Hiroshima, Japan;Faculty of Mathematical Science and Information Engineering Nanzan University Seto, Japan

  • Venue:
  • Mathematical and Computer Modelling: An International Journal
  • Year:
  • 2003

Quantified Score

Hi-index 0.98

Visualization

Abstract

This paper addresses the problem of how to determine optimal sleep timing when the computer user should turn the hard disk or the display off in order to save electric power after the computer has not been used. We propose a stochastic model to obtain an optimal sleep timing strategy which minimizes the expected power consumed per unit time in the steady state, where access requirements of the system occur in a renewal process and processing is described by a general service time. Phase type approximation is proposed to represent the expected power consumed per unit time. We investigate the performance of this approximation through a simulation study, and design an optimal auto-sleep schedule for a computer workstation system using real data.