NEWS

CS Researcher receives €1.5m to tame the chaos of wireless networks

22 November 2011

Testing wireless coverage inside an anechoic chamber, designed to cut out interference from the outside world

Calibrating wireless equipment inside the anechoic chamber of the Electrical and Electronic Engineering Department, designed to cut out radio interference from the outside world

Dr Kyle Jamieson, a lecturer in the Department of Computer Science, has received a €1.5m ERC Starting Independent Researcher Grant. This award is one of only three nationally in the area of Systems and Communications Engineering, and rewards MIT graduate Jamieson’s ‘promising track record’ and ‘scientific excellence’.  He will use it to build a unique experimental facility in high-performance wireless networking, as well as recruit a postdoctoral associate and PhD students.  Jamieson and team will address a problem that is growing in our cities—the tangled messes that are our ‘chaotic’ wireless networks.

Mobile phone and TV networks are carefully planned in advance, to place masts efficiently and provide uniform coverage. But many modern wireless networks have no central planning—people just buy routers and set them up where they like, causing problems with interference, loss of connectivity, and security. These ‘chaotic’ networks are increasingly common in modern decentralized communications.

Coding out of the grey

Dr Kyle Jamieson and his student work on wireless problems

Dr Jamieson and student brainstorming ideas for new wireless protocols

Within wireless networks there will often be areas where you can receive a signal, but it’s barely strong enough to work with. These are known as ‘grey zones’, and crop up at the very edge of coverage, behind objects casting wireless ‘shadows’, or when the connecting device is moving too fast. Jamieson is looking at using techniques from coding theory to make the most of however much connection you have.

Currently, if part of a message sent wirelessly doesn’t quite make it, the receiver throws away the entirety of the message and waits for a whole new copy. If this copy isn’t perfect either, this process repeats again and again until a lucky message gets through entirely intact.  A more efficient possibility could be to hang on to the bits that did make it, and then patch together a complete message out of all the slightly mangled bits using error correction techniques.

Secure by nature

 

The eight antenna analysing the network coverage in all directions

Using multiple antennas, one can analyse the direction and strength of network signals

For a decade, the standard approach to security has been to put a security protocol on top of the wireless network. However, an arms race has developed between security protocols and exploits for them, with each repeatedly outstripping the other. In this situation, it’s impossible to be confident that your system is safe—sensitive data have been stolen before, and it’s feared they will be in future.

A promising alternative is to use the very nature of the wireless signal itself to ensure security properties.  Such an approach is possible because wireless signals vary in unforgeable ways depending on a user’s location, like a written signature on a document. Using many antennas, Jamieson believes that the infrastructure can construct these signatures based on wireless signal’s physical angles of arrival from a mobile device, such as a user’s laptop. The approach could mean an end to people parking in the street and breaking into the WiFi that leaks through from enterprise and domestic networks.

Don’t tread on me!

Today’s networks are dense, and wireless spectrum is scarce. As we demand more and more resources to transmit rich media, network capacity needs to keep increasing. So a future-proofed network needs to be totally scalable—able to provide the greatest number of people with the greatest number of bits, to transport ever-increasing content loads over the finite spectrum available.

Again, incorporating information that comes from physical wireless signals can assist here. If there’s no central planning for the location and spread of antennas, two may end up overlapping, both reducing transmission capacity and leading to deterioration of performance if they interfere with each other. Directional antennas can overcome this by recognizing each other and mutually ‘agreeing’ not to tread on each other.

Multi-layer cake

A close-up of a software-defined radio

Flexible radio platforms, such as the Rice University WARP system shown here, open the wireless airwaves up to innovation

Overcoming these problems requires work on all the levels that make up a wireless system—from the hardware of oscillating circuits, classically dealt with in an EEE context, to the classic Computer Science role of developing protocols.

Traditionally, the technology to produce radio signals is built into hardware and immutable, and then the other layers are ‘stacked’ up on that base. To find solutions that work in the real world, Dr Jamieson is taking a more flexible approach based on software-defined radio, experimenting with systems that can mimic many different radio devices—from long wave broadcasting to remote control for robots—simply by running different software.

Research into wireless networking is one of the most intellectually active and rapidly advancing areas in Computer Science today. What makes Jamieson remarkable in this area are his cross-disciplinary ability and interests—an unusual combination that makes him a real asset to the field. As the department that brought the Internet to Europe, UCL Computer Science has long been a leader in networking research. This highly competitive grant will support Jamieson, his postdoc, and PhD students as they shape the future of wireless networking, continuing that tradition of excellence.

Share on FacebookShare on Google+Tweet about this on TwitterShare on LinkedInPin on PinterestShare on Tumblr
Latest news from UCL Engineering
UCL recognised as Academic Centre of Excellence in Cyber Security Research
EPSRC funds Blockchain technology for Algorithmic Regulation And Compliance project
Collaborating AI learns to play StarCraft