The global communications network has grown so quickly that its energy use is becoming a problem. Now the industry is responding.
In January, 13 organisations came together at a hastily arranged press conference in London to announce the GreenTouch alliance, an effort to cut the energy used by communications networks by 1000-fold. They were driven by the realisation that, with the mass uptake of mobile data services, and the continuing expansion of the mobile subscriber base beyond the 4.4 billion users connected today, the issue could quickly become an economic, political and reputational problem for the industry.
How big is the problem? According to current estimates, today's global communications network is responsible for 300 million tonnes of carbon dioxide emissions a year - the same output as 20 per cent of the cars on America's roads. Reducing this figure and the impact of new usages and new users is not a problem that will be easily solved.
'If we take everything we know today and apply it, it cannot compensate for this rise. At the very best we can hold emissions constant,' said Gee Rittenhouse, vice president and head of research at Alcatel-Lucent's Bell Labs.
The response from Dr Jeong Kim, president of Bell Labs, last summer was to challenge two of his researchers to take a fresh look at the fundamentals of communications technology, much as Nobel Prize-winning physicist Richard Feynman did for the electronics industry in his 1959 talk 'There's Plenty of Room at the Bottom', which laid out the fundamental limits of device scaling.
A team of two looking at the thermodynamics of information storage and transmission had grown to a multidisciplinary team of two dozen by autumn. It concluded that the energy consumption of the global communications network could be cut 10,000 fold in theory, or by 10 million-fold for wired networks alone. The team also realised - before it had even presented its conclusions to management - that the job was bigger than any one company could tackle. Hence the GreenTouch alliance, formulated in about six to eight weeks, with the goal of demonstrating a set of technologies to cut network energy consumption by 1000 fold (leaving an order of magnitude headroom to the theoretical limit) within five years.
'This is a broad and daring initiative to tackle one of the biggest problems we are facing,' said Kim.
Politics and opportunities
It's also going to be one of the most difficult initiatives to drive forward. The GreenTouch alliance will be run along open innovation lines, a strategy increasingly used by industrialists to collaborate on research with academia and each other. Open innovation can broaden the expertise that is applied to solving a problem, but it can be difficult to manage, especially if some participants have not worked this way before, or the terms of reference are not clearly stated at the outset. Intellectual property (IP) can also be difficult - it's fine to assert that all the IP developed during joint research will be shared, as the alliance has, but one piece of shared IP is often built on another, proprietary piece of IP that its owners may still want to control.
There are also bound to be conflicting interests. The alliance is open for anyone to join, with its initial members forming a steering committee to get things off the ground. But at time of writing no other equipment vendors had announced their involvement, perhaps concerned about working so closely with competitors on future technology. There are more subtle issues, too, at least according to Ulf Ewaldsson, vice president and head of the radio product area at equipment maker Ericcson, which is not yet an alliance member. He points out that one of the best ways to reduce basestation power requirements is to get rid of all the standard interfaces between system functions that have become part of their design over the past few years. For operators, though, these interfaces have enabled them to drive down costs by specifying commoditised subsystems rather than proprietary yet highly optimised complete systems.
The alliance is looking for new technologies, but there are optimisations available today that could save substantial amounts of energy if they were consistently applied. Ewaldsson estimates that a macro basestation used to consume 3 to 4kW, but technology improvements have cut that to 1kW. Moving the signal processing electronics close to the antenna, rather than driving the antenna through a long piece of feeder cable, could reduce this to less than 500W. He estimates that a further fourfold reduction in power consumption is possible, just by applying technology options that are available today such as sleep modes for unused circuitry and antennas during low-load situations. There are also opportunities to use more complex coding algorithms to squeeze more data through a channel without increasing the power used to drive it.
Beyond the harvesting of such low-hanging fruit, the alliance will have to work on both better point technologies and better systemic optimisations.
'To have a true impact we need to consider the whole system in a realistic way versus the technology and the operators' use models,' said Muriel M'dard, who runs the network coding and reliable communications group at MIT, at the GreenTouch alliance launch. 'The history of theory going into products is good, for example MIMO, which began on the theory side and has migrated quite quickly. into Wi-Fi. If advances are made in a way that is judicious they have a history of making very considerable differences.'
Daniel Kilper, a member of the technical staff of the optical networks department at Bell Labs agreed: 'It's not about making a better transistor, it's about the combination of things; for example, small cell sites demand less energy.
Many communications systems are optimised for costs and performance, but M'dard argues that optimising for energy use as well need not be seen as another constraint - it could instead be viewed as another dimension in which to find an optimal solution: 'In a single channel you can trade energy, bandwidth, and time for a given noise in the channel.'
She added: 'Consumers and operators will decide on the optimal point and then you need to be working with optimisation in mind.'
It's not going to be easy. Kilper said: 'Optimisation problems are especially hard when you want to solve them exactly. But Bell Labs has lots of experts in heuristics who are good at getting close to optimal solutions.'
But it is going to be worth it. Kilper said: 'The Internet has been rushing along like a train on the tracks. Now we're going to define a new set of tracks so the Internet train can keep going for many decades to come.'