On tree-structured Vector quantization over noisy channels

Nam Phamdo, Nariman Farvardin

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

Vector quantization (VQ) has been known as an efficient method for data compression [1]. However, a drawback that limits the applicability of VQ is the large encoding complexity associated with the codebook search. Among various techniques used for VQ complexity reduction, tree-structured VQ (TSVQ) [2] has received special attention. Various TSVQ-based schemes have been used successfully in speech and image coding applications (e.g., [3]-[5]). The attractiveness of TSVQ resides in its structured codebook which, in turn, leads to a simple encoding procedure. The penalty paid for using the structured codebook is a modest decrease in performance (higher average distortion for the same bit rate) and a higher memory requirement. To date, there has been little research done on the performance of TSVQ when the data is to be transmitted over a noisy channel. Since in most practical situations, some sort of channel error occurs, the natural questions to ask are: (i) How do TSVQs perform in the presence of channel noise? and (ii) What can be done to improve this performance? This paper will attempt to answer these two questions.

Original languageEnglish
Title of host publicationProceedings - 1991 IEEE International Symposium on Information Theory, ISIT 1991
Pages248
Number of pages1
ISBN (Electronic)0780300564
DOIs
StatePublished - 1991
Event1991 IEEE International Symposium on Information Theory, ISIT 1991 - Budapest, Hungary
Duration: 24 Jun 199128 Jun 1991

Publication series

NameIEEE International Symposium on Information Theory - Proceedings
ISSN (Print)2157-8095

Conference

Conference1991 IEEE International Symposium on Information Theory, ISIT 1991
Country/TerritoryHungary
CityBudapest
Period24/06/9128/06/91

Fingerprint

Dive into the research topics of 'On tree-structured Vector quantization over noisy channels'. Together they form a unique fingerprint.

Cite this