Towards strategies for Power-Aware Networking

Lead Research Organisation: University of Cambridge
Department Name: Computer Laboratory

Abstract

The Internet and the World Wide Web are using between 2% and 5% of the energy in the developed world and this is fraction increasing rapidly as demand for new services such as online video (Youtube) and Social Networking, Internet voice and video communication, and online shopping expand. We need to characterise where and when this power demand comes from, and then see if we can apply techniques to trade of power and performance - if demand falls, then can we use energy proportional techniques to slow down computing and communications technology, so that (e.g. at night time) we get an equivalent reduction in consumption.This project is to fund a sabbatical visit by a leading practitioner in Internet Measurement and Green Networking, as part of a larger initiative to tackle these problems. Professor Barford has proposed a particular approach to one aspect of the problem, which is to devise a microrouter , which is a kind of small building block, that provides the essential communication facility of the Internet, but can be stacked up (metaphorically speaking, a bit like Lego bricks) to build faster (but more energy hungry) communication when needed, or pub back in their box (turned off, logically) when not needed. Finding the best dynamic strategies (when to add more microrouter blocks, when to take them away) is the second main outcome of this project.

Planned Impact

The principle impact of this proposal is on people, through binding together disparate groups in the UK working on energy aware networking, via visits, seminars, and a workshop. Secondary impact will be via the existing EPSRC Program Grant, INTERNET at Cambridge and Leeds, which has economic/industry impact plans that include direct exploitation of inventions, and standardization. Underlying this, the societal value of reducing the blatantly unnecessarily large electricity consumption by large network and service providers clearly has an impact on the GDP since it reduces the non-productive expense, and improves the environment, although the latter point is a relatively minor contribution. However, low-hanging fruit should not be sneezed at! And the lessons from measurement and adaptation from the work are applicable in other utilities, and therefore of general value.

Publications

10 25 50

publication icon
Vallina-Rodriguez N (2013) Energy Management Techniques in Modern Mobile Handsets in IEEE Communications Surveys & Tutorials

 
Description This grant funded a 1 year visit by Professor Paul Barford, who worked in connection with the EPSRC funded Internet project on Power Aware Networking. As well as giving several seminars, he also assisted with PhD advice and produced 4 papers on relevant topics, and provided fantastic assistence getting data and access to expertise on power grid and router hardware architecture, which will be reported elsewhere (under the EPSRC INTERNET project grant Research Outcomes, which is due later).
Exploitation Route The outcomes are several - one is we have used the advice to help BT, CIsco and the BBC to look at ways to reduce power consumption in their infrastructures.
Sectors Digital/Communication/Information Technologies (including Software)

 
Description This work led to better models of Internet Energy use, and helped disseminate work from the EPSC INTERNET project to US researchers as well.
First Year Of Impact 2013
Sector Digital/Communication/Information Technologies (including Software)
Impact Types Economic