This award is funded under the American Recovery and Reinvestment Act of 2009 (Public Law 111-5).
Increasingly, the Internet is used to distribute content on massive scale. Massive content distribution causes shortage of network capacity, increases costs of service providers, and impairs quality of user experience. This project develops advanced content-distribution techniques for achieving complex cost-performance objectives. An interesting class of existing content-distribution techniques, known as ?swarming?, use ad-hoc methods designed to help file receivers exchange pieces of the file they have already received. However, little is known about how to develop suitable swarming techniques for infrastructure networks under complex objectives.
This project is unleashing the potential of swarming as the most advanced content-distribution technique. Unlike existing ad-hoc approaches, the new approach uses general optimization theory in developing swarming algorithms. The novelty is to conceptualize swarming as a technique for distributing content over multiple multicast trees (MMT). Under the new approach, content distribution is formulated as a problem of optimal swarming over MMT. The solution algorithms become the best way of conducting swarming under each performance objective. Even heuristic solutions have performance guarantees.
The project will contribute a family of optimized and robust swarming techniques, suitable for future networks and applications. These techniques can mitigate network congestion or increase distribution speed, leading to eventual outcomes in increasing economic value, accelerating the rate of innovation, and improving productivity. The algorithms can be applied generally to other content-intensive applications, including IPTV, or to scientific networks that routinely transfer massive data.
