printer friendly version

Of fuzzy logic and fusion engines

The twenty-first century has thrown up entirely new paradigms of crime and terror in scale as well as in remoteness and anonymity from which the new criminal can strike and unleash untold violence on unsuspecting citizens – at schools, malls, offices, crowded public places and even homes. To compound the situation, the police and security agencies often do not have the luxury of well-defined space in which to search and indefinite time in which to arrive at deductions.

Sanjay Dharwadker

Often, the masterminds of unspeakable horror could just lurk in cyberspace, or appear and disappear at places so ephemerally and in such deceiving ways that conventional techniques, such as biometrics when used in isolation, could be effective only by remote chance.

In such a situation, we seem to have arrived into the future where a new generation of devices and techniques are being deployed in an integrated and comprehensive manner, that are yielding spectacular results already.

One of the key elements of such strategies is the effective use of multi-modal biometric search engines. It is well known by now that no single biometric when used in isolation is effective any more. Fingerprint repositories, the legacy of colonial times, assume repeat crimes by the same individuals. This is well understood today, even by common thieves and thus routinely circumvented. DNA evidence can be planted and is often found to be so. Faces and voices can be camouflaged, but taken together with other evidence can be effective, and similarly so with samples of handwriting and signatures. Match this with a vehicle or house and street and the picture seems to gather focus, both in space and time.

Thus, such systems need to be able to process together, not only various kinds of biometrics, but also text, numbers and images from structured databases, geographical information systems or even disparate, unstructured or incompatible masses of data. In such situations the search and match processes are neither precise nor binary, but fuzzy – best described as being imprinted upon layers of information in shades of grey – arriving at degrees of recognition just like a human mind does; putting together shape, sound, smell, image and taste as well as their synthesia in different contexts.

In recent years, this has been made possible as four distinct elements of information technology have been fused together – large data storage, algorithms that can rapidly traverse and find, analytical tools that can join the dots, and finally field devices that can reveal the possibilities, instantaneously at the scene of action.

Such state-of-the-art systems for the search and match of identities are already in place, not only within criminal justice departments, but also border control for both passengers and cargo – where often quick reflexes have become the order of the day, in order to prevent deliberate planned acts of crime and violence.

Bombings in crowded places, like the London Subway and the Boston Marathon were successfully investigated by rapidly analysing terabytes of CCTV footage, zeroing down to a few single faces, and then tying this into other elements of bio-data. Mysteries such as the MH370 would be one day explained when the dots are finally joined, perhaps in this way.

Individually, many such algorithms and tools are known to be effective, but what is it that makes them successfully work together? While there can be many attributes to this, the single most important cornerstone seems to be the standards – that seem to have pervaded the systems and processes that are now expected to seamlessly work together for identity search and match, the most critical of these being BioAPI, CBEFF, PIV, ANSI INCITS 378 and 381 as well as ISO 19794 parts 4 and 5 – to name a few.

Another critical aspect is the global arena within which such systems are expected to successfully operate and thus must not only conform to prevailing multi-lateral frameworks, but also national and international law pertaining to customs and immigration. For this, such systems must be versatile and flexible and thus be able to fit into existing situations.

Also importantly, as Moore’s law goes, such systems are becoming increasingly cost-effective and can be afforded by a large number of governments and agencies, and could even be oriented into regions such as the SADC and can be jointly owned and operated by groups of nations having common interests. An important factor would be the training of personnel to manage, operate, maintain and thus ensure the overall sustainability of such systems in the long run, and of course become an important focus area of future multilateral cooperation.

So it seems that the post-modern Sherlock Holmes would do well to have identity search and match capabilities in his Watson – that can do fuzzy logic and of course, fusion engines – all in one.

Share this article:

Further reading: