Reducing energy consumption of network infrastructure using spectral approach

TL;DR

This paper proposes a spectral clustering approach to reorganize network topology and enable sleep modes in switches, significantly reducing energy consumption in LAN infrastructure.

Contribution

It introduces a novel spectral clustering method for reorganizing network topology to optimize energy savings by enabling switch hibernation during low traffic periods.

Findings

Spectral approach effectively reduces network energy consumption.

Switch hibernation based on clustering saves significant energy.

Method maintains network performance while decreasing energy use.

Abstract

The energy consumption by ICT (Information and Communication Technology) equipment is rapidly increasing which causes a significant economic and environmental problem. At present, the network infrastructure is becoming a large portion of the energy footprint in ICT. Thus, the concept of energy efficient or green networking has been introduced. Now one of the main concerns of network industry is to minimize energy consumption of network infrastructure because of the potential economic benefits, ethical responsibility, and its environmental impact. In this paper, the energy management strategies to reduce the energy consumed by network switches in LAN (Local Area Network) have been developed. According to the life-cycle assessment of network switches, the highest amount of energy is consumed during usage phase. The study considers bandwidth, link load and traffic matrices as input parameters which have the highest contribution to energy footprints of network switches during usage phase and energy consumption as output. Then with the objective of reducing energy usage of network infrastructure, the feasibility of putting Ethernet switches hibernate or sleep mode was investigated. After that, the network topology was reorganized using clustering method based on the spectral approach for putting network switches to hibernate or switched off mode considering the time and communications among them. Experimental results show the interest in this approach in terms of energy consumption. .