Case study: From the universe to the Internet

Peter Cogan describes how the skills of a physicist are always in demand at Bell Labs, although the research focus is now on telecommunications rather than fundamental physics

My fascination with physics began at age 14 when I read an online article about the curvature of the universe. I found it intriguing to learn that a single parameter, Ω, can tell us whether the universe is open, flat or closed, and thus whether it is likely to end with a big crunch or a whimper. This experience got me hooked on physics, and it played a big part in my decision to study the subject at university. However, reading that article also taught me the importance and power of the Internet as an educational tool – so you might say that, in a roundabout way, it helped lead me to my current job at Bell Labs, the research division of telecommunications giant Alcatel-Lucent.

After getting a joint-honours degree in mathematical and experimental physics at University College Dublin (UCD), I stayed on to do a PhD in UCD's high-energy-astrophysics research group. When I joined the group in September 2003, its members were heavily involved in constructing a new instrument, the Very Energetic Radiation Imaging Telescope Array System (VERITAS), at the Whipple Observatory in Arizona, US, as part of an international collaboration. As a result, I spent most of my PhD building, testing and operating the array of four 12 m telescopes as the various pieces of the project came together.

Working on VERITAS often required a certain amount of inventiveness, which has served me in good stead in my subsequent career. I recall a particular night in spring 2005 when we brought the first sub-array of two telescopes online. Because the drive system in one of the telescopes was malfunctioning, it initially looked like we would not be able to test the sub-array, even though the hardware and software for the array were ready. Our solution was to steer the broken telescope up to the required elevation of 45° manually, using an electric drill and a very large gear ratio. This manual procedure was normally only used in emergencies to stow the telescope in the event of drive failure close to sunrise. Such an event could be catastrophic, because the telescope's mirrors would act like a giant solar concentrator, vaporizing the camera and other delicate instruments.

Once the telescopes were aligned, we switched the system on for the first time. At first, it did not work – it never does – but after a quick study of the software we were able to trace the problem to a stray minus sign in the coordinate transformation code, which calculates the angles between the telescopes on the ground and the direction in which they are pointing. Once this was corrected, we tried again and succeeded in taking a data run. I analysed the results immediately, and it was a real highlight to see effects we had predicted from simulations actually appear in live data.

Making the transition
After I finished my PhD, I became a postdoc in the high-energy astrophysics group at McGill University in Montreal, Canada, where I led the VERITAS data-analysis effort. One of the aspects of this work that I liked most was the combination of working on software, hardware and science. On a given day at the VERITAS site, I might spend the morning debugging hardware issues with the control system, lunchtime attending a seminar and the afternoon developing analysis code and writing papers.

By the end of 2008, I was ready to return to Ireland, and I also felt it was a good time to expand my career prospects beyond astrophysics. Bell Labs has an incredible reputation in the history of 20th-century physics, so its Dublin research centre seemed like an obvious place to move to next. Founded in 2003 and partly funded by the Irish government through the Industrial Development Agency, the centre, known as Bell Labs Ireland, is involved in a number of research projects associated with the telecoms industry. These include designing low-cost, high-power antennas and studying ways of optimizing telecoms networks. This kind of work appeals to me because it requires many of the skills that make a physicist a physicist, like data analysis, problem solving and project management. So although the goal of the research is very different compared with astrophysics, many aspects of the job remain the same.

Bell Labs' parent company, Alcatel-Lucent, is one of the world's largest telecoms firms, and the headquarters of Bell Labs in the US occupies a sprawling campus in Murray Hill, New Jersey. The Dublin centre is relatively small in comparison, with about 30 researchers in total. However, that is changing thanks to Bell Labs Ireland 2, a new five-year round of funding that will see the lab expand to more than 70 dedicated researchers. The Dublin centre is also heavily involved in the emerging GreenTouch consortium – a group of network operators, universities and government agencies with the lofty goal of making networks three orders of magnitude more efficient within five years (www.greentouch.org).

The research team at Bell Labs Ireland is diverse, with engineers and scientists from almost every country in the European Union, plus Turkey, Russia, China, India and Iran. This melting pot of cultures and languages is a fascinating place to work and interact. Bell Labs is also quite a strong academic environment, which I must admit is not what I had expected to find moving from academia to an industrial lab. Researchers are encouraged to participate in the wider scientific community through collaboration, conference participation and by publishing journal papers. There are also regular seminars where people can present their current work.

Ingenuity in action
One of the most interesting projects I have been involved in at Bell Labs is the investigation of network-diagnostic data. This project has been done in collaboration with scientists at Trinity College Dublin and the National University of Ireland Maynooth, and its ultimate goal is to make broadband Internet services more reliable. Although the backbone is ultra-reliable, problems with installation, faulty ports and poor wiring in the customer's home can have a detrimental effect on the reliability and quality of broadband services. By analysing data on parameters such as the bit rate and signal attenuation over a period of time, we hope to determine remotely and automatically whether a broadband connection has a problem, and, if it does, what that problem actually is. This would have a massive impact on the quality of service that customers receive from their Internet service provider. In order to achieve this, we have been investigating the use of tools such as artificial neural networks and other multivariate algorithms, which are able to find correlations and dependencies that are otherwise hard to discern in multivariate data.

As a physicist working in industry, I have found that many of the skills I learned during my PhD and postdoctoral work are essential. In particular, the ability to manage projects, communicate, learn on the fly and critique ideas at a fundamental level have all been very important. While fundamental-physics research is not undertaken at Bell Labs Ireland, I have found that there is almost no area of research at Bell Labs where the skills of a physicist cannot be put to use.

About the author
Peter Cogan is a researcher at Bell Labs Ireland