Invited Speaker: Philip Levis
Date:
November 21, 2008
Time:
1:00 PM - 2:00 PM
Venue: Boelter Hall 4760
I argue that an inability to observe the internal operation of a network -- a lack of visibility -- is a fundamental challenge in low-power, embedded networks such as wireless sensornets. While sensornets have seen proposals for a large number of powerful, complex protocol and application solutions, in practice network designs use only the most basic primitives and eschew efficiency for simplicity. Simple algorithms improve visibility by being easier to reason about, explain, and understand. The premise of this talk is that this need not be a tradeoff: with the right primitives, networks do not need to trade off efficiency for simplicity. I'll present three pieces of work that support this claim. The first is power locks, the mechanism TinyOS uses to manage both energy and concurrency. The second is the Collection Tree Protocol (CTP), the standard covergecast protocol in TinyOS. The third is Grant-to-Send, a novel collision avoidance mechanism for wireless networks. Each of these pieces of work involve a large number of collaborators; in particular, Omprakash Gnawali from USC is the lead student working on CTP.
Philip Levis is an Assistant Professor of Computer Science and Electrical Engineering at Stanford University. He received his Sc.B. from Brown University in 1999, his M.S. from the University of Colorado at Boulder in 2001, and his Ph.D. from UC Berkeley in 2005. He joined the Microsoft Research New Faculty Fellows in 2008. He researches low-power, embedded, and wireless networked systems. Jis current research focuses on designing simple protocol primitives and system abstractions to make these networks to build, deploy, troubleshoot, and maintain.