Improving Performance Through Channel Diversity in the Presence of Bursty Losses

Loading...
Thumbnail Image

Related Collections

Degree type

Discipline

Subject

Diversity
Losses
Performance
Bursty

Funder

Grant number

License

Copyright date

Distributor

Related resources

Author

Vergetis, Evangelos

Contributor

Abstract

As more applications migrate to IP networks, ensuring a consistent level of service is increasingly important. One option is for the network to offer service guarantees. Another is to leverage the path diversity that the Internet intrinsically offers. Our focus is on understanding if and when one can indeed take advantage of multiple disjoint paths to improve performance. We consider an environment where loss patterns are bursty and where coding is used to provide robustness against packet losses. We assume that only long-term loss statistics are known about each path, and we seek to identify the best strategy for sending packets over the available paths. Our contributions are two-fold. First we demonstrate that even with minimal knowledge of channel characteristics and using simple transmission policies, path diversity can help significantly improve performance. Second, we derive an efficient method for identifying optimal policies, and more importantly characterize when having access to multiple paths can be of benefit.

Advisor

Date of presentation

2005-08-29

Conference name

Departmental Papers (ESE)

Conference dates

2023-05-16T22:33:07.000

Conference location

Date Range for Data Collection (Start Date)

Date Range for Data Collection (End Date)

Digital Object Identifier

Series name and number

Volume number

Issue number

Publisher

Publisher DOI

relationships.isJournalIssueOf

Comments

Proceedings of the 19th International Teletraffic Congress 2005 (ITC19), Performance Challenges for Efficient Next Generation Networks, August 29 - September 2, 2005, Beijing, China.


Proceedings of the 19th International Teletraffic Congress 2005 (ITC19), Performance Challenges for Efficient Next Generation Networks, August 29 - September 2, 2005, Beijing, China.

Recommended citation

Collection