Sabina Zejnilovic, Dual Degree Ph.D. Student: “I would like to make a positive impact in our everyday life”
Networks fascinate her, “whether they are connecting communication equipment, as in a telecom network, or computers on Internet, or even people in a social network.” As part of her Ph.D., she recently presented a poster titled “Communication Complexity and Energy Efficiency on Consensus Algorithms,” on the Inaugural Symposium of the CMU Portugal Program. Sabina explained that, in conjunction with her two advisors, they “discovered that some slower algorithms are in fact more efficient than the faster ones, as they require overall less communication to achieve the same goal,” adding that “we have also shown that in some network topologies, the algorithms are as efficient communication wise, as they are in complete networks.”
As a Ph.D. student she does not have a clear idea of what she wants to do after completing her graduation: either pursuing academic research or going back to industry. Of one thing she is sure: “I enjoy solving interesting problems, applying engineering practicality and creativity, and seeing the solutions making positive impact in our everyday life,” Sabina Zejnilovic said.
CMU Portugal: You are a dual degree Ph.D. student in ECE, at IST/UTL and CMU, since 2011/2012. Before joining this doctoral program, you left your work as a software Engineer at BH Telecom d.d. Sarajevo. Why did you decide to go in this direction?
Sabina Zejnilovic [SZ]: After completing my Bachelor’s degree in Telecommunications at University of Sarajevo, I was not sure in what directions to pursue my career, if academia or industry. Hence, I did both. I worked as a System Engineer for Bosnian Telecom operator, while keeping ties to my University as a part time teaching assistant and completing a Master’s degree. At that point it was pretty clear that if I wanted to do research at a Ph.D. level, it would require a full time commitment and quite a change in my life. And after years spent working in industry, I felt that I wanted that change and a chance to dedicate myself to a creative and challenging work of a doctoral student. Not to mention that a Ph.D. program abroad, especially a dual degree program, set both in Europe and the USA seemed like an exciting and enriching environment for this line of work.
CMU Portugal: How do you comment on your experience on the dual degree Ph.D. Program so far?
SZ: It turned out to be challenging and exciting in so many aspects. Academic-wise, trying to fulfill the requirements of a doctoral program, not only intellectually challenging research, but all the accompanying aspects of coursework, qualifying exam and teaching internship is certainly no small matter by itself. On top of that, all this is to be done in two different institutions, with two advisors and in two different continents. But in this process what I find the most valuable is that I have an incredible network of top-level researchers, with whom I can exchange ideas and discuss work, a truly inspiring environment.
CMU Portugal: How was the adjustment to these two different cities, two different environments?
SZ: Logistics is not easy. Settling down in a foreign country, dealing with all the cumbersome administration and paperwork, navigating through a new culture and adjusting my lifestyle, only to do that all over again, and then again, well that is quite an experience. However, being exposed to new cultures and meeting new people positively broadens not only work, but also generally, a person’s life perspective. The two cities, Lisbon and Pittsburgh, are so different, both beautiful in their own ways, and offer experiences which I always look forward to come back to.
CMU Portugal: In the Symposium held in January by the CMU Portugal Program, you presented a poster titled “Communication Complexity and Energy Efficiency on Consensus Algorithms.” What have you discovered so far?
SZ: Consensus algorithms have been widely researched, not only by social scientists, but by engineers as well. Studying the process of how different agents in a network exchange information and agree on average of their initial beliefs has wide implications. Since so many different consensus algorithms exist, an interesting problem becomes how to compare them, in terms of their performance: how close they are to agreeing; and in terms of accompanying cost: how many messages had to be exchanged to accomplish that. Communication complexity is a metric that captures both aspects and characterizes energy efficiency of these algorithms. My work involved evaluation of efficiency of some popular consensus algorithms and studying the effect of different network topologies on this efficiency. We have discovered that some slower algorithms are in fact more efficient than the faster ones, as they require overall less communication to achieve the same goal. We have also shown that in some network topologies, the algorithms are as efficient communication wise, as they are in complete networks.
CMU Portugal: What are your main research interests?
SZ: Networks have always been in the focus of my interest, whether they are connecting communication equipment, as in a telecom network, or computers on Internet, or even people in a social network. Modeling and studying the diffusion of phenomena in a network has so many different flavors to it, depending on: is it the passing of information or infection in a communication network, or spreading of a trend or a rumor in a social network that we are interested in. Even though the applications are so numerous and diverse, yet there are a lot of unifying elements. Currently, I am interested in the problem of localizing the origin of such diffusion, which can either be the first infected computer in a network or a trendsetter in a social network. Solving this becomes even more interesting considering the fact that we do not always know the full structure of a network, as some devices and connections may be hidden, and not all friendships are featured on Facebook.
CMU Portugal: How do you envision yourself in the future?
SZ: It is still a fuzzy problem without a clear-cut solution. Although plenty of tasks have been accomplished, still plenty remain until I can defend my thesis. Therefore, I opt to further defer to the future the choice of going into academia or working an industry job. I see the lines separating these two nowadays are blurred, as being in academia may require a bit of entrepreneurship, and working for a company may involve theoretical based research. Wherever it may be, I enjoy solving interesting problems, applying engineering practicality and creativity, and seeing the solutions making positive impact in our everyday life.
May 2013
