Researchers keep an eye on the future of security

Researchers at the University of Kent are using bespoke software to centrally manage multiple types of biometric data to increase...

The idea of checking physical characteristics to authenticate a person's identity has a long and distinguished history.

Identification has been used to catch criminals for more than 100 years, thanks to breakthroughs in areas such as fingerprint and DNA analysis.

Now identification using biometric data is creeping into IT systems, with the creation of applications that use biometric technologies such as iris and fingerprint recognition.

In the UK the Home Office has evaluated biometric systems, which authenticate a user's identity by taking a reading, such as a fingerprint, and comparing it to an original template, for use in areas like passport control and entitlement cards. An iris recognition system could be in place at UK airports by the middle of next year.

However, there are still barriers to the widespread adoption of the technology - barriers that a research team at Kent University aims to overcome by developing more robust and reliable detection systems.

"One of the key problems is that there is no single biometric device that is reliable and accurate enough for all applications, and not everyone recognises that," said Mike Fairhurst, a professor in the university's department of electronics. "People need to be more flexible in their approach."

Fairhurst's team has developed an "intelligent processing framework" that uses bespoke software to centrally manage multiple forms of biometrics, choosing the most appropriate for the job or combining different forms to increase accuracy and reliability.

One such project is Iambic (Intelligent Agent for Multimodal Biometric Identification and Control), which is being run in collaboration with technology developer Neusciences.

The team has created a demonstration Iambic system that uses a laptop fitted with biometric sensors and a microphone as the client, with a networked desktop PC acting as the server.

Users enrol using as many of the three biometrics - voice, fingerprint and facial recognition - as they can and create a template which is held on the client. On subsequent visits they have to verify for each reading and are given a confidence rating for each. The idea is that different thresholds would be set according to the user and the application.

A key driver behind the project is that even the most accurate and reliable biometrics suffer from problems of acceptability. Iris recognition, for example, is fairly obtrusive as users have to place their eye close to the reader. Users can find it tricky to align themselves correctly. Worn fingers and the general likelihood of smudged fingerprints can restrict the efficacy of fingerprint recognition systems.

In its Iambic trials, the team found that users had difficulties enrolling with the voice recognition component. People may also have a disability or injury that prohibits them from using a particular mode of biometrics.

"The performance figures when working with real people can surprise you. Sometimes the devices are very good but people find it hard to interact with them. A lot of work needs to be done to improve the interface and make it more intuitive," said Fairhurst.

Another problem is impostors faking an identity, using techniques such as high-tech contact lenses to mimic someone's iris reading, showing a biometric reader a photograph or even chopping someone's finger off to foil a fingerprint recognition system.

"As techniques become more sophisticated so do the countermeasures," said Fairhurst.

Farzin Deravi, a member of the research team at the University of Kent, said a more sophisticated system of enrolment is needed as existing systems tend to assume users are who they claim to be. The Kent team is looking at adding a smartcard reader to help validate identity at the enrolment phase.

Despite all the barriers to adoption, Deravi predicts there will be a creeping introduction of biometrics alongside other technologies.

"The process has begun. We are quite confident that biometrics in some shape or form will play a large part in people's lives in the future," said Deravi. "It is inevitable - the future is biometric. But you have to be realistic about biometrics. You have to steer a way between the hype and the paranoia."

The Kent University research team is backed by British Trade International, a government body set up to support the UK's trade and investment strategy.

CV: Mike Fairhurst

Mike Fairhurst is a professor at Kent University's department of electronics, where he has worked since 1972.

Fairhurst heads the computer vision and image processing research team, which develops techniques for image recognition and the analysis of image structure. His interests include image analysis, handwriting analysis, biometric processing and neural architectures.

Kent University has been conducting research into biometrics for about five years and aims to become Europe's first centre of excellence for biometrics and security.

Getting wired: tell us the future

Research work being undertaken at universities today will change the way we use IT, and Computer Weekly is on a mission to showcase their cutting-edge IT research. We would like to hear from researchers who think they might have made a breakthrough. Each week we will feature innovation in the field of IT, giving a glimpse of how technology will evolve in the coming years.

Email:  [email protected]

What is BTI?

British Trade International is the government organisation set up to support Britain's trade and investment strategy. It brings together the work of the Foreign and Commonwealth Office and the Department of Trade & Industry on trade development and promotion of inward investment. BTI's two operating units are Trade Partners UK, which aims to help UK companies trading overseas, and Invest UK, which aims to promote the UK as an inward investment location.


Read more on IT strategy