To solve real world problems, A. Kevin Tang reduces them to their essence. In the process, he sometimes solves not only the problem at hand, but others in seemingly unrelated fields.
“I’m not only looking at specific networking applications, I’m looking more at fundamental issues. It’s more generic, less protocol specific,” he says. “I look at the fundamental but my research still results in specific solutions, so in that sense, I’m still engineering.”
He offers Internet congestion control schemes as an example. “You have to have a feedback control system to monitor the packet loss rate or delay observed and then design algorithms based on that to adjust your transmission speed,” he says. “It’s actually a fantastic thing because the Internet is arguably the largest manmade feedback control system and it’s purely distributed—there’s no central control at all.”
The distributed nature of the system presents a problem for any program designed to estimate network conditions. “The challenge is because it’s decentralized, heterogeneous, and there’s some delay when something happens before you know it,” explains Tang. “You really have to have robust things that work for all these sorts of possibilities to produce that constantly changing download speed in a distributed manner.”
Stripped bare, this problem is really no different from that of how to best dynamically distribute a desirable commodity. “This is really a resource allocation problem with the bandwidth as the goods, because there’s fixed bandwidth in the network for each router,” Tang says. “That’s not new at all; it’s something economists have been researching for decades.”
Tang’s work in this area while a Ph.D. candidate at Caltech won him the George B. Dantzig Award for Best Dissertation from the Institute for Operations Research and Management Sciences. “I made that link with operations research and microeconomics, borrowed from them, and contributed back,” he says.
The prospect of collaborating with other Cornell faculty to make similar contributions is what drew Tang here. “This department I really like because if you divide the different research areas into different clusters—communications, signal processing, and control theory—they are all strong,” he says. “And there’s also a strong computer science department and a strong operations research department, so I can reach out to these strong potential collaborators easily, but at the same time, for my areas, it’s relatively open. It is quite ideal for a starting faculty member.”
Tang is already working on a proposal with Eric Friedman in the School of Operations Research and Information Engineering and Steve Strogatz in the Department of Theoretical and Applied Mechanics. “They told me the opportunity that there’s a new proposal initiative from NSF in complex systems with an emphasis on complex dynamics, and we teamed up as our expertise complement very well for that” he says. “This kind of team, you can’t do this everywhere. I really learn and benefit a lot.”
Prof. Tang's Web site
