"In doing science, you can't just use a black box; you have to know what is inside, understand it, and be able to explain how it works."
Associate professor of computer science
Felipe Perrone compares his latest project to "putting training wheels on a really dangerous, fast bike." The bike in this case is a tool used to simulate and study computer networks; the training wheels help researchers produce more credible data from their simulation studies.
Simulations are invaluable for studying networks. Imagine you want to see whether a wireless network with 100,000 nodes can survive the random destruction of some fraction of those nodes. Conducting this experiment in the real world would be exorbitantly expensive and a logistical nightmare. With simulation, you can conduct the entire study in a digital world, with minimal expense and equipment.
Unfortunately, simulation studies have experienced a "crisis of credibility," Perrone says. Researchers cannot always replicate each other's results because the underlying assumptions and methods are often not explicit. "It is as if you are just using a black box and getting a magic result," he says. "In doing science, you can't just use a black box; you have to know what is inside, understand it, and be able to explain how it works."
Another problem has been that researchers don't always have all the expertise they need. "There were a lot of people doing simulations of networks who were experts in simulation but not in networks, and there were a lot of people who were conversely experts in networks but not in simulation; each of the two camps would commit some capital sins here and there," he says.
Perrone and his colleagues are creating user-friendly frameworks for ns-3, a popular network simulator. Their tools will automate the staging of experiments and recording of data, guide novice users through the proper use of the simulator, and provide experienced users with the power and flexibility they need. Perrone, who received a grant from the National Science Foundation to fund the work, will spend his sabbatical this fall working on the frameworks project with students. "It's great to be able to create learning opportunities for undergraduates, but then again, I'm working with two outstanding undergraduates so it's a pleasure all around."
Perrone is a firm believer in the value of studying abroad. "It is life changing and mind expanding," he says. "In the global economy we have today, it's a requirement of all engineers to be able to work with multiple cultures and ethnicities." Toward that end, Perrone worked with Tim Raymond and Margot Vigeant, associate professors of chemical engineering, to take the Engineering 290 course to his native Brazil this past summer. Titled "Engineering in a Global Societal Context," the three-week course provides engineering students with an opportunity to travel abroad in an academic context.
As for what's next after the frameworks project, Perrone sees endless possibilities, which include letting his students propose new directions. "My own interest will continue to have some focus on this area of networking," he says. "For the sake of creating more opportunities for students, I would like to open up to their input and see what problems they would like to study with simulation."
Posted Sept. 20, 2010