The Friday Interview: Dr Debasis Mitra, Bell Laboratories


14 Oct 2005

If anyone sums up the global nature of modern industrial research it is Dr Debasis Mitra (pictured). Indian-born, British-educated, he has spent his working life with Bell Labs — the research and development (R&D) arm of US telecoms equipment giant Lucent Technologies. He even went to the trouble of learning Russian in the Sixties in order to tap into the “great eruption” of new theories emanating from what was then the Soviet Union.

Mitra is vice-president of the Mathematical and Algorithmic Sciences Centre at Bell Labs’ headquarters in Murray Hill, New Jersey. The overall facility employs a total of 500 researchers; Mitra’s unit accounts for about 40 of these, drawn from a range of disciplines, from mathematics and statistics to electrical engineering and computer science.

Mitra has come to know Ireland quite well since Bell Labs established its first international research lab in Blanchardstown a year ago. After its first director Dr Lou Manzione resigned in August to take up a position back in the US, Mitra assumed one of his responsibilities — that of chairman of the Science Foundation Ireland (SFI) Mathematics Advisory Committee.

The main reason for this particular visit, however, is to deliver a talk to the Royal Irish Academy as part of the Hamilton Bicentennial celebrations. The noted 19th-Century Irish mathematician, who did pioneering work into optics and light waves not to mention algebra would doubtless have seen Mitra — a mathematician whose enquiring mind had led to breakthroughs in global communications systems — as something of a fellow traveller.

Over the past four decades, Mitra has played no little part in transforming our communications networks from the essentially voice-based systems of yesteryear to the information-rich data conduits or so-called packet-switched networks that span the globe today. The mechanics of how this has been achieved is not something that would occur to the ordinary citizen but the truth is that it could never have happened without mathematics. For example, the work of which Mitra is most proud is in the area known as “statistical multiplexing” — how data of different types can be shoved down the same pipe in a way that optimises the use of the communications channel while ensuring users get the performance levels they need. The work was underpinned by mathematics, specifically theories of applied probability.

The application of mathematics is not limited to products; it can also be applied to processes. Mitra and his team were, for example, deeply involved in the supply chain optimisation programme at Lucent Technologies, which aimed to strip out cost and increase efficiency. A key part of this was applying complex forecasting models to predict both customer demand and the flow of components from suppliers.

“It’s been a revolution,” remarks Mitra. “The whole process of manufacturing bears no relation to what existed as recently as five years ago. Inventory and turnover rates are orders of magnitude better than what they were.”

Mathematics or more specifically game theory has even had a direct input to boardroom strategy. Mitra’s team has used the game theory notion of stability to model the evolution of the telecoms industry and so giving, for example, insights into how alliances between different telcos might fare.

In the future, Mitra predicts that the research being done at Bell Labs will help transform networks from mere delivery vehicles into intelligent infrastructure that telecoms firms can use to gain competitive advantage. “One of the challenges is developing networks that learn, that have the ability to gather a lot of information on users and find out their preferences — and then how to mine that information in a manner that is of value to the user. The network is not just a bandwidth pipe and information processor; it’s as much a database,” he observes.

For researchers of network systems, one of the great motivating factors is that their work is never done. As soon as one problem is resolved, another emerges to take its place. While most of the remaining technical challenges lie in the area of wireless networks, Mitra points out that the fixed-line arena can still generate a few real conundrums – such as how to accommodate multiple users and rich content on the so-called ‘last mile’ of the network into people’s homes.

“How to allocate bandwidth to the many users sharing that channel is a big issue right now and a lot is riding on it because if you want to send TV and video over the last mile such issues will need to be resolved,” he says.

Bell Labs has clearly had a huge business impact over the years. The main reason, Mitra believes, is that it has managed to balance the roles of basic and applied research.

“We work in two planes: the project or real-world plane and the abstract, discipline-based plane. The success of ventures such as Bell Labs so essentially depends on the interworking of these two planes – the ability of researchers to spend time in both of these planes … Hiring the right people, allowing them to grow in this capability is a very important part of our function in the past and today,” he observes.

He believes the same principle can be used to help increase the quality of mathematicians within the Irish research system. In fact, an important mission of the SFI’s Mathematics Advisory Committee he chairs is “to couple the mathematicians with the real world through partnerships,” he says.

The lack of mathematicians is not unique to Ireland; it also applies to the US and to the entire developed world. Mitra feels a number of factors are to blame, primarily globalisation, which has the effect of paring costs in all areas including R&D, and the deregulation of telecoms markets and the consequent break-up of companies such as the old AT&T. While accepting that liberalisation was “probably exactly the right thing to do” Mitra feels that the loss of big players that had an end-to-end view of the supply chain has been detrimental to R&D.

“Ideal research would require you to take a holistic view; since that’s not possible you need to think about environments that would allow you to recreate it. The old Bell Labs achieved that but that was a long time ago. So the question is what can we do to recreate that end-to-end view in what is now a decentralised and fragmented industry.”

By Brian Skelly