New Results on Optimistic Source Coding

Naoki SATO  Hiroki KOGA  

Publication
IEICE TRANSACTIONS on Fundamentals of Electronics, Communications and Computer Sciences   Vol.E87-A   No.10   pp.2577-2580
Publication Date: 2004/10/01
Online ISSN: 
DOI: 
Print ISSN: 0916-8508
Type of Manuscript: Special Section LETTER (Special Section on Information Theory and Its Applications)
Category: Information Theory
Keyword: 
source coding,  fixed-length coding,  general source,  information-spectrum,  optimistic coding,  

Full Text: PDF>>
Buy this Article




Summary: 
Optimistic coding is a coding in which we require the existence of reliable codes for infinitely many block length. In this letter we consider the optimistic source coding theorems for a general source Z from the information-spectrum approach. We first formulate the problem to be considered clearly. We obtain the optimistic infimum achievable source coding rate Tε (Z) for the case where decoding error probability εn is asymptotically less than or equal to an arbitrarily given ε [0,1). In fact, Tε (Z) turns out to be expressed in a form similar to the ordinary infimum achievable source coding rate. A new expression for Tε (Z) is also given. In addition, we investigate the case where εn = 0 for infinitely many n and obtain the infimum achievable coding rate.