|
I am working in the following funded project:
The members of our group are:
SNF project 102819 (10/2004 - 09/2008)
Peer-to-peer (P2P) systems, in which peer computers form a cooperative network
and share their resources (storage, CPU, bandwidth), have attracted a lot of
interest lately. After the apparition of the first truly successful P2P
systems, and the significant amount of research conducted in Academia and in
the Industry, most researchers now agree that P2P systems are more than just a
fashion phenomenon. They offer great potential for building cooperative
networks that are self-organizing, efficient, scalable, and reliable.
Research in P2P networks has so far mainly focused on content storage and
lookup, but little work has been done about its actual distribution. By
capitalizing the bandwidth of peer nodes, P2P architectures offer great
potential for addressing some of the most challenging issue of today's
Internet: the cost-effective distribution of bandwidth-intensive content to
thousands of simultaneous users and the resilience to "flash crowds" (a huge
and sudden surge of request traffic that usually leads to the collapse of the
affected server).
In this project, we propose to address the problem of cooperative distribution
of streaming media and large content from a networking perspective. We plan to
specifically focus on three complementary research directions: first, the
design of topology-aware P2P substrates specialized for efficient content
distribution; second, the push-based distribution of streaming content with
timing constraints, such as streaming media (TV, radio); third, the pull-based
distribution of large, but time-insensitive stored content. Each of these
research topics will be supported upstream by extensive analysis and modeling,
and downstream by prototype implementations and experimental validation. This
project is expected to yield not only novel research contributions, but also
practical techniques for cooperative content distribution applicable both
Internet-wide and in large private networks. These techniques can be of great
interest to media and infrastructure providers, as well as to medium and large
companies that wish to avoid the cost of dedicated Content Delivery Networks
(CDNs) for reliable content distribution.
|
