Keynote Speakers

SpaSWiN'06 will be opened by two keynote speakers from different research communities,

• P. R. Kumar (University of Illinois at Urbana Champaign)
  Random graph models of wireless networks: Connectivity, clocks, computation and capacity
  At one granularity of analysis, wireless networks can be modeled as random Euclidean graphs. Nodes can be modeled as randomly located in two or three-dimensional space, and connected by edges to neighbors at a certain distance away, or to a certain number of nearest neighbors. One question of interest is whether such graphs are connected. A second question is how to accurately synchronize clocks over multi-hop networks, which is motivated by envisaged time driven computing applications of wireless networks to sense and interact with their physical environment. A third question is how much traffic wireless networks can carry. A fourth question is how to compute over wireless sensor networks. We address all four questions, with more of an emphasis on the issue of clock synchronization. [Joint work with A. Agarwal, V. Borkar, A. Giridhar, P. Gupta, R. Solis, F. Xue].
• R. Meester (Free University of Amsterdam)
  From statistical physics to information science; how to use math for communication networks
  In this lecture, I will review, emphasise and discuss various ideas from the mathematical community that have turned out - or are expected by me - to be useful for answering questions arising in the engineering community. In particular, I will discuss results (1) from percolation theory, (2) from the theory of Poisson approximation, and (3) some results from general probability theory concerning so called sharp phase transitions. Elaborating a bit: (1) Typically, results in percolation theory deal with the infinite plane, while applications often require statements about what happens in a bounded domain. To this end, I will emphasise that in percolation theory, it is often possible to draw useful conclusions about finite-domain behaviour when it comes to (almost) connectivity questions. We also discuss problems arising from dependencies that are NOT local. (2) I will discuss some applications of the powerfull Chen-Stein method that sometimes enables us to deal with dependencies, leading to asymptotic results about Poisson convergence. (3) This part is mainly speculative, and deals with the interesting fact that most phase transitions are very 'sharp', in a very precise and well defined way. This could possibly lead to connectivity statements of finite systems, as opposed to asymptotic results.

Invited Speakers

SpaSWiN'06 programme will also feature two invited speakers,

• D. Aldous (University of California at Berkeley)
  Flows through random networks
  Consider highly abstracted models of transportation or communication networks, in which we need to simultaneously route flow between each source-destination pair, in an optimal way subject to various cost and capacity conditions. We take a statistical physics viewpoint: study properties of the optimal flow instead of algorithms for finding it. In particular, consider curves giving some quantitative measure of network performance as a function of overall traffic demand. Can one systematically relate qualitative properties of such curves to structural properties of the network? Such questions are much easier to ask than to answer: I will describe some work-in-progress.
• F. Baccelli (Ecole Normale Supérieure / INRIA, Paris)
  Self Organization of Interfering 802.11 Wireless Access Networks
  This paper proposes a set of distributed algorithms that allow (i) multiple interfering 802.11 Access Points to select their frequency in a way that minimizes co-channel interference, and (ii) clients to choose their Access Point so that the bandwidth of the whole network is shared optimally, where optimality is understood in the minimal potential delay sense. The proposed algorithms rely on Gibbs' sampler and optimize global network performance based on local information and decisions. They do not require explicit coordination among the wireless devices. We prove the convergence of the proposed algorithms and study their performance using event-driven simulations. [Joint work with A. Chaintreau and C. Diot (Thomson), B. Kauffmann (ENS) and D. Papagiannaki (Intel)].