News

• [Sept '11] I am starting as a postdoc at Stanford. And my journal paper on a theory for green communications appears in JSAC!


• [July '11] How do we know when cognitive radios are in near field, and when they are in far field? How would it change our system design? Discussed in our paper that got accepted in Globecom '11.


• [June '11] Talk in the 2011 Georgia Tech summer school on cyber-physical systems (June 28) on the need to rethink models in controls, comm, and computing.


• [May '11] Received the Eliahu Jury award for “for outstanding achievement in the area of Systems, Communications, Control, or Signal Processing.”


• [May '11] Visiting Muriel Medard and Anantha Chandrakasan at MIT (May 8-13).


• [Apr '11] My first student's first paper got accepted at DAC '11!


• [Mar '11] Talk at CISS on wiring complexity of decoding, and implications on green communication. Slides can be found here.


• [Feb-Mar'11] Visiting Cedric Langbort and Tamer Basar at UIUC (Feb 21 to Mar 4).


• ITA Workshop talk on Green Codes on Feb. 10.


• [Jan '11] Green codes will have decoders with short wires. And this is a new code-design problem! My recent paper accepted at CISS talks about this issue.


• [Jan '11] Talk at U Michigan (Ann Arbor) on Jan. 7, 2010.


• [Dec '10] Received the best student paper award at IEEE CDC 2010! Talk slides of my paper: “Is Witsenhausen's counterexample a relevant toy?” are here.


• [Dec '10] Finally (!), my dissertation “Actions can speak more clearly than words” is done! It can be found here.


• [Nov '10] My article on why Witsenhausen's counterexample is important (a more traditional outlook): “Demystifying the Witsenhausen counterexample” appeared in “Ask the Experts” section of the IEEE Control Systems Magazine, Dec '10.


• [Oct '10] Seminars at MIT, UCLA, USC on speaking through actions.


• [Sep '10] “Is Witsenhausen's counterexample a relevant toy?” nominated for student best paper award at CDC'10 (a total of four nominees). The conference is in December!


• [July '10] Three talks at ISIT'10. Distributed signal cancelation inspired by Witsenhausen's counterexample [Slides], Shannon meets Tesla: Wireless information and power transfer [Slides], and Info-theoretic tradeoffs of throughput and power for decoding [Slides].


• [June 2010] Talk at Berkeley Wireless Research Center (BWRC) retreat, June 7, 2010: “Low-power wireless systems: A theory-practice ping-pong.” Email me for slides.


• [June'10] Nominated for ISIT best paper award for “Distributed signal cancelation inspired by Witsenhausen's counterexample ”


• [April '10] Three papers accepted at ISIT'10. Two are on green radios. And I defended (yay!). My thesis defense slides can be found here.


• [March '10][Slides] and [Handout] for my talk at the ISS Seminar at Princeton University, March 26, 2010. “Distributed control: a new frontier of information theory”

Click here for more news



[Slides] and [Handout] for my Graduation Day talk at ITA Workshop, San Diego, March 3, 2010. “Distributed control: a new frontier of information theory”


[Slides] and [Handout] for my talk at UCSD Control Center Symposium, Nov. 13, 2009.“Understanding implicit communication in distributed control”


Witsenhausen's counterexample: Why now?



Sept '09: A generalized Witsenhausen counterexample, that has costs on all the inputs and states. We solve this problem to within a constant factor. Presented at Allerton'09. Suggests that implicit communication is what makes decentralized control hard.


May'09: Tightened lower bounds on Witsenhausen's counterexample using a new technique. The new bound is much tighter, yielding a constant factor of 1.3 for the infinite length problem. Is deeply connected to Steinberg's information embedding problem. Accepted at ITW'09


Feb '09: An approximate solution to Witsenhausen's counterexample: we show that quantization based strategies can obtain within a factor of 8 of the optimal costs for the original Witsenhausen counterexample for all values of the problem parameters. Further, lattice based strategies obtain within a constant factor of optimal for vector Witsenhausen counterexample of any length. These are first results that show approximate optimality of any scheme for this longstanding problem. Accepted at ConCom'09.


Feb '09: How to do green broadcasting? We show that time-division multiplexing performs better than superposition/dirty-paper coding based schemes at short distances if decoding power is also taken into account. Optimal schemes, therefore, must have some component of time-division multiplexing. Accepted at ISIT'09. The paper also improves on the best known outer bounds on the error exponents for Gaussian broadcast channels.


Nov '08: Final version of Journal paper on vector version of Witsenhausen's counterexample is now up. We show that for an infinite length vector Witsenhausen problem, we can achieve costs within a factor of 2 using dirty-paper coding based strategies. Accepted for publication in IJSCC, 2008.


September '08: Work on the vector version of Witsenhausen's counterexample got accepted at IEEE Conference on Decision and Control (CDC) 2008.


May '08: Camera ready version of the work on Green Codes is now up. Appeared in the Proceedings of ISIT'08.

Invited talks and seminars