Let's get physical
While other researchers are busy developing new applications to squeeze into wireless devices, and add more services onto the network, Steve McLaughlin and his team of engineers are trying to extend the capabilities of the “physical” layer, but that doesn't mean they spend their time with screwdrivers, spanners and soldering irons – it’s all about maths... …
There are lots of major challenges in digital communications today, like trying to process more data – more quickly – and provide more services over the network, but when you also have to cope with “disruptive technologies,” the problems become even harder.
For Professor Steve McLaughlin (Chair of Electronic Communications at the Institute for Signal and Image Processing at the University of Edinburgh), his work in signal processing is all about “pushing the envelope” or extending the boundaries of digital communications systems.
The key is always compromise, McLaughlin says, and as well as reducing the cost per bit, researchers today also have to think about reducing the carbon footprint of networks, in the quest for what is now known as "green radio." When you add disruptive technologies into the mix, the problems multiply, because as soon as you design a new technology, people start doing strange things with it...
“What forces change in the network,” says McLaughlin, “is the way people use the services. And disruptive technologies don't only change the network architecture but also the business model.”
For example, he explains, when users discovered the joy of texting, the service providers were caught unawares. What started as a standby solution for technical problems became a new dimension of the industry which today accounts for about 40 per cent of total network revenues.
Similarly, Google, YouTube, the iPhone and the BBC’s iPlayer have all had unexpected effects on the system, changing the nature of traffic and vastly increasing the load.
So, faced with all these different pressures, how do we plan for the future? And how do we measure our progress?
The more things change...
Since the early 1980s, there have been dramatic changes in the industry. At that time, much of the focus was on military systems, while today it is mainly consumer devices. What used to take two years to bring to market can now be delivered in a couple of weeks – much smaller and much more intelligent, and also more dataintensive.
One job McLaughlin remembers in the mid 1980s was the challenge of developing a tactical radio system with a handset weighing less than one kilo, incorporating 18 layers of flexible circuits which today would be a few lines of code in some software, on a programmable semiconductor. “We did not even have any CAD tools,” says McLaughlin, “but even though we still deal with the physical layer, today we’re more concerned with mathematical problems like interference cancelling algorithms – and that is the smart part.“
Despite these huge advances, the “physical” challenges still seem the same – getting wireless devices to process more data (including compressed video) and use up less battery power, while sending and receiving signals faster, at any time in any location. The ultimate aim for McLaughlin is to optimise performance of the physical layer (the fundamental layer in communications systems which processes data for transmission and reception by different devices), and what is called “crosslayer optimisation,” balancing the needs of different layers – including applications, network, data link and physical layers.
As well as having to deal with the problems created by progress in other technological dimensions, and the problems created by disruptive technologies, researchers also have to “retrofit” communications systems, getting the existing network to do things it wasn’t designed for – e.g. using the telephone network to provide high-speed broadband.
Often, this means all you can deliver is a “best-effort” service, but it’s better than nothing...
Maths for engineers
Today, the research group working with McLaughlin focuses on “maths for engineers”, and members of his team can write new algorithms and program them onto a chip in a couple of weeks, taking advantage of the latest simulation tools. Among the work being done at the Institute is the development of audio signal processing components for the personalised audio market, and new ways of sampling video signals so they can be reconstructed more easily, improving image quality and maximising use of the available bandwidth. Another project involves reducing the power consumption of base stations by using algorithms to transfer the complexity to the receiver instead of the transmitter. The Institute is also doing groundbreaking work in relaying techniques and scheduling schemes (for more details, see sidebar).
For example, while other researchers come up with exciting new concepts like “Cognitive Radio,” which enables the service providers to dynamically sell their available bandwidth as if it’s a commodity, McLaughlin and his colleagues have to deal with major issues like latency and interference, or the system will simply not work. It is all about tradeoffs...
Personalisatiation of services delivered by “cooperative radio” also sounds very nice, says McLaughlin, but how do we incentivise users? If subscribers are asked to enable their mobiles to act as mini base stations for other users, bouncing signals from mobile to mobile, then how do we cope with the problems of battery power? Either the phone will shut down in a couple of minutes, or the user will have to pay for the privilege of being part of the cooperative network, helping other users. Ultimately the constraint is the physical layer, and that is where the maths comes in – and also where McLaughlin and his colleagues at the Institute will continue to focus their efforts, as they try to stretch the limits of the possible.
Everyone is asking what if questions, but to improve efficiency and reduce power consumption, they also have to ask how they will do it. If researchers at Strathclyde come up with something new that places extra demands on the network, Edinburgh have to write new algorithms so the physical layer can cope. If Edinburgh manage to crank extra performance from the physical layer, Strathclyde can think up new ways to exploit the extra capacity.
McLaughlin calls this process “dynamic interaction”, and for him and his colleagues, it equally applies to signal processing as human behaviour – and nothing is allowed to interfere.