ASYMPTOTIC ANALYSIS OF MULTIPLE DESCRIPTION QUANTIZERS

Authors
VAISHAMPAYAN VA BATLLO JC
Citation
Va. Vaishampayan et Jc. Batllo, ASYMPTOTIC ANALYSIS OF MULTIPLE DESCRIPTION QUANTIZERS, IEEE transactions on information theory, 44(1), 1998, pp. 278-284
Citations number
25
Categorie Soggetti
Computer Science Information Systems","Engineering, Eletrical & Electronic","Computer Science Information Systems
Journal title
IEEE transactions on information theory → ACNP
ISSN journal
00189448
Volume
44
Issue
1
Year of publication
1998
Pages
278 - 284
Database
ISI
SICI code
0018-9448(1998)44:1<278:AAOMDQ>2.0.ZU;2-T
Abstract
A high-rate analysis of multiple description quantizers is presented f or rth-power distortions and general source densities. Both, fixed-len gth and variable-length encoding of the quantizer indices are consider ed. Optimal companding functions are shown to be the same as for singl e-channel quantizers. As compared to the bound of Ozarow, a gap of 8.6 9 dB and 3.07 dB exists between the entropy-constrained and level-cons trained eases, respectively, for a memoryless Gaussian source and r = 2.