Interview with Parul Gupta

Featured Engineer

Parul Gupta

Parul Gupta - Wireless Research Engineer

What are your favorite hardware tools that you use?

A lot of my work on algorithms for wireless systems involved writing equations, hence my favorite hardware tools have mostly been a pen and paper :-) I also use stationery heavily – highlighters, colored pens, post it note tags – I can almost never read a paper without highlighting, underlining or taking copious notes in the margins!

But these are slowly getting replaced by “software tools” – PDF annotators. Amongst the more conventional type of tools, I like spectrum analyzers for the wealth of information they provide while testing actual wireless hardware.

What are your favorite software tools that you use?

MATLAB. Because it is so versatile, so user-friendly and has such good data visualization features. I have been using it for the last decade for many different purposes – simulating complex wireless systems, analyzing large data-sets, for course projects from control theory to matrix algebra, or simply as an all powerful scientific calculator.

What is the hardest/trickiest bug you have ever fixed?

The bugs I have found the hardest to trace are memory errors in huge multithreaded programs in C. To the best of my knowledge, there aren’t any great tools which give exact information on the cause of the crash, though valgrind and gdb often help clean up some. For one project, we built a Wimax simulator in C which had many threads for different data plane and control plane functions. Depending on the data or exact sequence of events, the program would take a different execution path and errors weren’t always easily reproducible. The trickiest ones usually were conquered only after many many printfs and tedious scrolling through pages of error traces. Actually this is a very rife area for good contributions – a lot can be done to make multithreaded programming easier and more efficient.

What is on your bookshelf?

Most of my reading these days happens on the computer, but I like collecting books and often have a much longer pile and a more ambitious reading list than I find time for.

Amongst technical references, I like Vishwanath & Tse’s “Fundamentals of Wireless Communications”, Krovella and Krishnamurthy’s “Internet measurement” and Cormen, Leiserson aIand Rivest’s “Introduction to Algorithms”.

In non-technical stuff, I have enjoyed reading from many different genres, “O Jerusalem” by Lapierre and Collins, “Persepolis” by Marjane Satrapi, “Fountainhead” by Ayn Rand and “Gone with the wind” to name a few.

What has been your favorite project?

My favorite project so far is the current one that I am working on, called the “Wireless network cloud”. It brings together recent advances on Software Defined Radio and Cloud Computing technologies. Our aim is to build a system where instead of having dedicated proprietary hardware at Base Stations, telecom service providers can offload their processing to a cloud and get all the advantages of elastic capacity and pay-per-use models. I like it since it has the potential to radically change the way wireless connectivity is delivered, also because it challenges me to learn many new things because of its wide scope. For those interested in knowing more please see the article by Lin et al titled “Wireless Network Cloud: Architecture and System Requirements”, in IBM Journal on Research & Development, Vol 54, No. 1, Jan/Feb 2010

Do you have any note-worthy engineering experiences? (blowing up things, getting shocked, etc.)

Nothing as ‘physically’ dramatic except crashing a few servers and losing a lot of data! The experiences that remain in memory when something that challenges a widely-held belief is discovered.

There have been many such – for example, once we were trying to build a better equalizer for an OFDM based wireless system and after trying all the tricks like smoothing filters, decision feedback etc, we were still far from ideal and were getting only insignificant incremental gains with any more tweaks. I had a big discussion with my supervisor on how a particular type of weighted soft-decision viterbi decoding might be a good idea but he disagreed since we had already incorporated erasure based decoding (a special case of that). I went ahead to implement and test its performance and the SNR-PER curves shifted by 2 dB, the largest ever performance gain we got with any other technique! We later filed a patent on this.

In summary, never hesitate to challenge the ‘more experienced’ views, sometimes naivety has its benefits :-)

What direction do you see your business heading in the next few years? What challenges do you foresee in our industry?

I currently work at IBM Research India and this is not a marketing gimmick, I sincerelybelieve in IBM’s “Smarter Planet” dictums of the a moreinstrumented, interconnected and intelligent world. There will be alot more information coming not just from people, but also machines and sensors of various types. Some people also call it the Internet of Things. I think that’s where the world is headed. It will be big challenge to transport such huge amounts of data and a bigger challenge to make sense of it for better solutions. Analytics will be big – there is a lot to do!

Previous Next