Next Generation Traffic Control
Self-Similar and Multimedia Traffic Control
Principal Investigator: Kihong Park
Sponsor: NSF (ANI-9714707)
Students: Tsunyi Tuan, Ambal Balakrishnan, Wei Wang
In this project,
we study the issue of how to design network protocols
that effectively deal with the scale-invariant burstiness associated
with self-similar traffic.
Our framework - multiple time scale traffic control (MTSC) - exploits
long-range dependence present in self-similar traffic to facilitate
effective traffic control at several time scales.
This work builds on our earlier work on the characterization
of the causal roots of traffic self-similarity and its
performance impact which
showed that self-similarity is a ubiquitous phenomenon on the
Internet with detrimental effect on network performance.
Our work has resulted in enhanced, extensible protocols
for both bulk (e.g., TCP) and real-time traffic (e.g., UDP-based
MPEG video and audio) that achieve significant performance
improvement vis-a-vis existing protocols.
New solutions to Internet traffic control for bulk and
QoS-sensitive data transport under self-similar traffic conditions.
Some of the issues include:
- Self-similar burstiness.
- Transport of multimedia data with real-time QoS constraints using
adaptive redundancy control.
- Fairness, optimality, and efficiency.
- Game theoretic considerations.
Network Systems Lab
self-similar traffic references
Visit the Purdue Infobahn QoS Testbed
Outdated. Update may follow soon.