Seven years ago, Peter Rabinovitch attended a MITACS conference in Ottawa as a Master's student at Carleton University. It was his first exposure to this nation-wide network of Ph.D.-level mathematicians.

Today, as a Senior Research Scientist at Alcatel-Lucent Canada in Ottawa, he is working side-by-side the next generation of mathematicians to make networks more secure and smarter.

Students participating in the MITACS Internship Program are augmenting the scientific capability of the telecommunications giant. Alcatel-Lucent's first intern, Helen Tang, worked on an Internet traffic generator tool used by researchers to simulate and analyze the performance of routers. She was also involved in analyzing IP flow statistics, resulting in a model that can be used to better design and engineer some of Alcatel-Lucent's new products. Dr. Tang is now working as a defence scientist with the research division of National Defence.

"We had an idea for a project that required some pretty advanced mathematics – expertise that is in short supply in house," says Mr. Rabinovitch, who is doing his Ph.D. in probability at Carleton. "So we inquired with MITACS as to whether they had any student programs. That's how the MITACS internship program was born."

Today, three other MITACS interns – Paul Boone, Pin Yuan and Yihui Tang – are applying their mathematical skills to address a new set of networking challenges at Alcatel-Lucent.

As part of a MITACS project led by Dr. Michel Barbeau at Carleton, Paul is developing a new algorithm for XML routers that could be incorporated into sensor networks to alert authorities to a potential forest fire, a break in a security fence or even a leaky water main. As Mr. Rabinovitch explains, sensor networks generate so much data that it's difficult to isolate the useful nuggets. Like a race car fanatic that only wants to receive magazines on racing, he says, Paul's work will help to create a system where sensor networks publish all information they receive, and individuals can subscribe to the data that's most relevant to them.

"If your job was to locate forest fires, you would request any piece of information that relates to a sudden temperature increase in this geographic location. The sensor in the forest would still publish all the data it receives, but the end work station wouldn't receive it because the XML router would try to match the publications with the subscriptions. Deciding what to forward, and where, can be a very difficult job and that's the algorithm Paul is working on."

Two other Carleton graduate students, Pin and Yihui, are developing algorithms to boost network security. Their goal is to come up with new tools that would help network administrators identify malicious content, priority data that deserves high-level service or something new that's never been seen before. Yihui's job is to identify changes in the traffic stream. Pin then compares how the traffic has changed to determine what action should be taken.

"This is about being able to look at network traffic without having to look way into the packet and seeing what every bit and byte means and what applications generated it," says Mr. Rabinovitch.

On the skills side, MITACS offers companies expertise in several mathematical fields, including statistics, probability, algorithm development and simulation abilities. And, because Alcatel-Lucent is very picky about the students it chooses, "the ones we end up with are probably the best," says Mr. Rabinovitch. There are the other advantages as well. "Graduate students have the right kind of aptitude for doing research. If you're a Master's or Ph.D. student, you're comfortable reading huge amounts of literature trying to find the interesting little nugget and how it fits our problem," he explains. Through the students we also get better contacts with the professors and the research going on at the university. The end result is that we do even more collaborative research with universities."