Souvik Sen views his research as a bridge between computer science and electrical computer engineering.
The Ph.D. student in his final year at Duke is studying wireless networking and mobile computing. He notes that wireless research has both CS and ECE perspectives — with electrical engineers studying the communication perspective and computer scientists looking at networking. Yet little communication has occurred between the two fields. In pulling these two communities together, Sen is finding results in his research. This past fall, he won the Outstanding Ph.D. Preliminary Exam Award for work on his thesis, which examines how to improve wireless performance.
“Souvik has made some important contributions to the area of cross-layer protocol design, where he leveraged ideas in wireless communication to redesign networking systems,” said his adviser, Romit Roy Choudhury, Associate Professor or Electrical and Computer Engineering and Computer Science.
One of Sen’s projects has demonstrated that communication protocols from wired networking can be applied to wireless networks to detect collisions in transmissions — something networking textbooks say is not possible. Another work has shown that contention resolution in wireless networking can be moved from the time to the frequency domain to improve performance. This less conservative design breaks away from 38 years of research, Choudhury noted.
“These demonstrate Souvik’s ability to think differently,” the professor said, “as well as his passion to work through various hurdles that come along the way.”
Through his investigations, Sen also has demonstrated that sensing information from the communication layer — or physical layer — of computers can be used to benefit wireless and mobile applications.
“This multipath information, which you can obtain from the physical layer, has rich information about the ambience in which your mobile device is in,” Sen noted. “You can do very precise indoor localization.”
Such technology, he said, could be used by museums to allow visitors to automatically retrieve information on artifacts they view.
“We have done experiments inside Nasher Museum,” he said. “We could say you are in front of a particular painting 90 percent of the time.”
Other applications of spot localization include advertising in a real-time manner.
“If you’re looking at a Samsung TV, then Samsung might give you a coupon to get you to buy,” Sen said. “Maybe Starbucks wants to give you a coupon whenever you are in a Starbucks. Localization can tell Starbucks whenever you are in one.”
Sen — who has been involved in other ways at Duke, including helping to found the Joint Youth Organization of Indians at Duke and finding ways to improve safety on and around campus — currently is looking at faculty positions and a research lab. He graduates in May 2012.