| 8:45am |
Opening Remarks |
9:00am
- 10:20 am
|
V.S. Anil Kumar, Virginia Tech.
On the Capacity of Wireless Networks
Abstract
A fundamental question in the design and analysis of
wireless networks is: Given a wireless ad-hoc network and a collection
of source-destination pairs, what is the maximum throughput capacity of
the network, i.e., the rate at which data from the sources to their
corresponding destinations can be transferred in the network?
Formulating the notion of capacity combinatorially is hard because of
its dependence on the protocols used at different layers and their
interaction. This has been an area of active research and in this talk,
we will discuss some of the main results and techniques for approximating
the capacity.
|
|
| 10:20-10:40am
| Break |
10:40am
- 12 pm
|
Alessandro Panconesi , Universita di Roma La Sapienza,
Act Locally, Think Globally
Abstract
Several problems for ad hoc networking are of the following
type. Each node in the network is to set up links to a subset of the
nodes within transmission radius in order to ensure the emergence of
good global properties. Examples of such properties are global
connectivity, expansion (which gives robustness to node failures),
low-stretch (which ensures short paths for routing) and so on. Due to
severe resource constraints of ad hoc networks, the choices made by
every node should be very simple algorithmically and based on local
information alone. We will discuss some basic problems of this kind
and see that the key to success is given by randomness. The discussion
will allow us to introduce some non-trivial ideas and tools from
probability theory that have much wider import. Specifically, the
problems that we will tackle are: power control; fast computation of
sparse overlay networks (with application to broadcasting) and, if time
permits, cryptographically secure sensor networks.
|
|
| 12-1:30pm | Lunch |
1:30pm
- 2:50 pm
|
Bobby Bhattacharjee,
University of Maryland
An Overview of Decentralized Applications
Abstract
A decentralized application has no single or central point for
coordination or application of policy. Decentralized applications
have potential benefits of being more resilient against component
failures, and more robust against malicious attacks. However,
distributed coordination of possibly faulty (or worse adversarial)
nodes is difficult. We will introduce different paradigms for
implementing decentralized applications (peer-to-peer, service
overlays, etc.), and discuss different user models within these
paradigms (cooperating versus competing versus adversarial users).
Using a data lookup, search, and multicast data distribution as
examples, we will survey the current state of the art in distributed
applications in different user models, and discuss open questions in
these areas.
|
|
| 2:50-3:10pm | Break
|
3:10 pm
- 4:30 pm
|
Amit Kumar
IIT Delhi and Aravind Srinivasan University of
Maryland
Game-Theoretic Issues in Networks
Abstract
Since large networks such as the Internet are shared
by users with competing (selfish) interests, game-theoretic notions
naturally arise in such contexts. We will present the relevant
basics of game theory including various notions of equilibiria,
mechanism design, cost sharing, and complexity issues. We will
also present a few applications in contexts such as multicast
and network design.
|
|
| | Concluding remarks |