One day, the work Marquis Scholar John Kolba ’06 (Chelmsford, Mass.) is conducting for his senior honors thesis could be used to perfect hearing aid technology or help the United States government improve radar transmissions.
An electrical and computer engineering major, Kolba is breaking new ground by working with a set of signal processing techniques called blind source separation (BSS) to determine which method is most effective at finding sources of mixed signals. Ismail Jouny, Dana Professor and head of electrical and computer engineering, is Kolba’s thesis adviser.
Jouny has published numerous articles in academic publications, many co-authored with Lafayette students. He has served as a summer faculty fellow at the U.S. Naval Air Base in Patuxent River, Md., and Naval Underwater Warfare Center in Newport, R.I. Jouny also has a U.S. patent pending, has served as a consultant in industry, and has been listed in Who’s Who in Science & Technology. The international publication for members of the Institute of Electrical and Electronics Engineers has cited Jouny’s research on landmine detection in an article about developments in the field.
“BSS is a way of taking data from several sources with a series of sensors and processing it to find out what information came from what source,” Kolba explains. “In the case of audio, different speakers comprise the different sources, and the BSS processing determines which sounds came from each speaker.”
His research includes comparing full band BSS, sub band BSS, and wavelet-based sub band BSS. In full band BSS, Kolba processes the signal all at once. For standard sub banding, he splits the signal into equal pieces of the frequency spectrum, performs BSS on each piece, and adds the results. With wavelet-based sub banding, he splits the signal into smaller and smaller pieces and adds those results together.
“With this process, I am hoping to find which one of the methods is more effective so we’ll be able to process the different signals more efficiently,” he says. “That could have applications in the science of hearing aids and also in eavesdropping techniques along the lines of espionage.”
The project’s radar applications include identifying different sources of mixed signals. Scientists around the world are researching the same topic.
“The immediate application is something called the ‘cocktail party problem,’” Jouny says. “At the party, you are trying to listen to one person, but there are just too many people talking.”
Separating signals and strengthening the most important one would allow the person at the party to understand the speaker more clearly.
“John’s making a small contribution to the field, but it’s a ‘divide and conquer’ technique,” Jouny explains. “His specific niche is doing the division using uniform sub banding and wavelet sub banding.”
Later this year, Kolba will use BSS on images.
“An image is thought of as a two-dimensional object, but in the process of rearranging the signals, you turn the image into a long, one-dimensional object, reprocess it through the separation techniques, and then recombine the image into two dimensions,” he explains.
Understanding signal sources in a video image is a helpful defense tool.
“You could take one image and superimpose another image onto it, something that’s faint enough so that it cannot be seen with the eye,” Kolba says. “If you want to hide a message or data or watermark an object, this technology would be useful.”
Almost no research on BSS has been conducted. While doing something that has never been done before is exciting, the work with audio BSS technology is equally interesting for Kolba.
“It’s fun right now, working with the signals, playing back the mixed signals, running the processing, and separating them,” he says. “I had a course where we learned to separate signals and how the computer could pull the pieces apart, so I knew it would work. I had never seen it in action until this semester; it’s been really neat seeing it do what it’s supposed to do.”
Kolba chose this topic because of its wide range of applications and because it is a good foundation for post-graduate studies.
“I know I like working in signal processing,” he says. “This is a specialty area of signal processing that I haven’t worked in before, so I am exposing myself to a new portion of the field and getting a broader picture of what signal processing is all about. The project has applications to radar, which is something I could see myself doing in the future. Ultimately, I want to work in research and development at a company that does Department of Defense contracting, similar to the summer work I have had the past several years.”
Kolba adds that Lafayette provides outstanding resources for student research.
“Lafayette is solely an undergraduate institution, so the professors can give us more attention than they might be able to at a university with a graduate program,” he explains. “A thesis is much more open-ended than research in a regular class. When you’re working on a thesis, you have the whole semester to be doing as much as you can. I have to make sure I take the time to work on it now and not sit back and let the time slip by. I have to really keep my goals in focus until the end. The project is definitely educating me in a manner different than any class I’ve ever taken.”
Kolba is a member of the student chapter of Institute of Electrical and Electronics Engineers, College Theater, the drama society Five Actors in Search of a Director, and the Marquis Players, a student-led acting troupe that produces an annual play for charity. He is vice president of History Club and teaches Sunday school at St. Peter’s Lutheran Church.
Chosen from among Lafayette’s most promising applicants, Marquis Scholars like Kolba receive a special academic scholarship and distinctive educational experiences and benefits, including a three-week, Lafayette-funded course abroad or in the United States during January’s interim session between semesters or the summer break. Marquis Scholars also participate in mentoring programs with Lafayette faculty and cultural activities in major cities and on campus.
Honors theses are among several major programs that have made Lafayette a national leader in undergraduate research. The College sends one of the largest contingents to the National Conference on Undergraduate Research each year; 39 students were accepted to present their research at this year’s conference.