Tackling the Capacity Crunch

Fibre-optic technology is a clever way of harnessing light, but with data consumption increasing, how long can it keep delivering? Professor Andrew Ellis and his team at Aston University are tackling a growing crisis in optical communications.

Andrew Ellis

Light is one of our basic needs. Each human cell - every functioning organism on our planet, in fact - synchronises itself with the light from the sun. Experiments in sensory deprivation show just how important light is to the brain - remove it for long enough and hallucinations will occur. But light is so much more to us than a biological necessity. It is also a conduit for information in an increasingly networked world. Streaming through optical fibres, light offers a means of encoding and transmitting data across vast spaces. Powering everything from the internet to entertainment, it is quite literally the engine of economic and social progress. 

However, following decades of growth, the world of communications is now facing some significant challenges. Within the UK alone, there are 17.6 million UK mobile internet connections and 73 per cent of homes with broadband access - all of these are powered by high-capacity core networks. Some have likened these networks to the invention of lead pipes 2,000 years ago. Just as the pipes allowed the Romans to carry water from one place to another, fibre-optic cables transmit information between two places using light-based technology. When you consider that 99.9 per cent of all data passes over these fibres, there can be no doubt that this system - which started life in the early 1950s as a telephony network - has delivered considerable benefits. Yet the rise in smartphones and tablets, and our changing online habits (streaming high-definition videos for example), is putting the network under considerable strain. Experts warn that the demand for bandwidth will eventually outstrip the capabilities of fibre-optics - and it will be sooner than we thought.

Fibre Optic

How the current system works

So what are the current capabilities? A fibre-optic cable is made up of 100 or more thin strands of glass called optical fibres - each one is less than a tenth as thick as a human hair. In terms of capacity, current records in research are 101 Tbit/s of data for conventional fibres and ten times this amount for new fibres. Current products are more like 10 Tbit/s (to put that into context, each 10 Tbit/s is equivalent to 5 million ADSL connections or one quarter of a million BT Infinity/Sky Fiber connections or 250 DVDs per second). This might sound like a lot but as demand rises, the numbers quickly get smaller. 

“We don’t have one fibre per person, we put multiplex signals from different people together and aggregate them,” explains Professor Andrew Ellis of the Optical Communications Group at Aston University. “So if you take the population of the country, all looking at one or two Google server farms, that’s a lot of traffic going into and out of those data centres. We get large volumes of traffic and we squeeze it all into one fibre. Once we’ve aggregated that traffic, if you go from two megabits to 20 megabits, that’s ten times the capacity. We don’t want to install ten times the number of fibres in the network because we don’t want to have BT charge you ten times for the bandwidth provision. So we need to squeeze that in more tightly, and that’s starting to get difficult.”

This problem, popularly called the ‘capacity crunch’, is the focus of the PEACE (Petabit Energy Aware Capacity Enhancement) Project, headed up by Professor Ellis and part-funded by the Engineering and Physical Sciences Research Council. He and colleagues at Aston are working alongside industry partners not just to improve bandwidth on our optical fibre networks, but also to reduce energy consumption: another consequence of the demand for internet capacity. 

“Ten to 15 years ago, there was about a 50/50 balance between the power requirement on the transmission line and the power requirement in the terminals,” says Professor Ellis. “In order to increase the bandwidth, the terminals have got smarter using the power of electronics. Now it’s more like 90 per cent in the terminals and only ten per cent in the line system. So the dynamic of the network has changed and so have our habits. While we were doing web searches, looking at text pages and images - short packets - now we’re downloading movies. The switches are still carrying a legacy of that history of supporting telephony and supporting short bursts of data for web pages and trying to carry high definition video. So we need to look at how we develop the transmitters and receivers to get that energy consumption down.” 

The problem of light intensity

The intensity of the light in the fibres is another problem. Once a few thousand customers are aggregated, and taking into consideration the different colours of light wavelengths, optical fibres are probably transmitting around 100 milliwatts of power. Less than the power of domestic lights - you might think - but when you push 100 milliwatts through a super-thin strand of glass, the concentration is like standing close to the sun.

“Despite the fact that it’s the world’s purest glass, there is some interaction between the light and the glass as the refractive index changes, which distorts the signal. So we’re looking in the PEACE project at how we can do clever things in the optical domain to allow us to increase the signal power - so we get a better signal-to-noise ratio, so we can increase the capacity.”

Swiss Physicist Jean-Daniel Colladon supposedly invented fibre-optics in the 1840s when he shone a light down a water pipe and discovered that the water carried the light by internal reflection. Yet, arguably, it was Newton’s experiments with prisms that really paved the way for optical communications. First published in his book Optiks (1704), Newton used pairs of refracting prisms to split white light up into a rainbow, and then recombine it into a single beam of white light using an inverted prism. His experiment relied on a phenomenon called dispersion. Up until relatively recently, a similar principle was still in use by employing two different types of optical fibres - one to spread the beam and one to narrow it.

“This works well at low intensities,” says Professor Ellis, “since two different fibres can have opposite dispersions, just like a prism and an inverted prism.” Unfortunately, the nonlinear dispersion which results from the high intensities of today’s networks is always the same sign (in other words, it always makes the blue part of the spectrum travel faster), and so can’t be compensated for using a fibre with opposite properties. “Newton’s experiment can be replicated using two prisms with the same orientation,” Professor Ellis explains, “and a mirror is used to reflect the beam between the two prisms. To counter dispersion what we do is essentially place a special mirror in the middle of the fibre. If you think about a prism, you know a prism spreads light around. You go through two identically orientated prisms, it spreads and then it spreads some more. Gathering all of that light is quite a complicated procedure because the light’s over a large spatial area. And in an optical fibre the light doesn’t spread around - it’s guided - but the same effect which causes wavelength-spread through a prism translates into time-shift in an optical fibre, so without using the ‘mirror’ a receiver has to have a large memory to catch up with all of the different delays. That’s what’s causing the power consumption.

“What we’re doing in the PEACE Project is we’re using a special equivalent of a mirror, known as a phase conjugate mirror which basically replicates the effect of carefully placing a mirror between two Newton’s prisms in an optical fibre. This reduces the pulse spreading at the receiver, the equivalent of the compact pencil beam which emerges from Newton’s prisms, and this allows the receiver to be simpler and more efficient. This special trick is particularly useful in the case of the nonlinear dispersion. Using a phase conjugate mirror to combat linear dispersion also compensates for nonlinear dispersion.”

The future

The implications of the PEACE Project are widespread, both at home and abroad. If these problems can’t be solved, networks can’t be grown, and that impacts on our ability to connect rural areas of the UK (a big issue when you consider that EU farming subsidies increasingly rely on access to the internet). Meanwhile, although developing countries aren’t facing the capacity crunch, they are increasingly networked and need affordable technologies that allow them to participate in the global economy.

The truth is that consumers want data provision to be cheap, reliable and limitless, but the reality - as we are increasingly realising with energy provision - is that we can’t have it all. All things considered, fibre optics have been an immensely useful technology, but the only approach now, argues Professor Ellis, is to rethink the whole communications infrastructure. “This increasing reliance on the existing network - it’s like having an oil shortage and basing your industrial strategy on more and more cars. You have to do things differently.”

There is only one thing - market forces - that will determine the demand for bandwidth and, consequently, the shelf-life of fibre-optic technologies. That is a notoriously difficult thing to predict. But solving the capacity-crunch would bring huge benefits to everyone’s lives and one thing is certain - there’s no time to waste.

This article first appeared in the Sep 2015 edition of Aston in Touch >>

Find out more about Professor Andrew Ellis

Find out more about Professor Andrew Ellis
It’s like having an oil shortage and basing your industrial strategy on more and more cars.