DocumentCode :
1078699
Title :
Optimal Prefix Codes for Infinite Alphabets With Nonlinear Costs
Author :
Baer, Michael B.
Author_Institution :
Ocarina Networks Inc., San Jose, CA
Volume :
54
Issue :
3
fYear :
2008
fDate :
3/1/2008 12:00:00 AM
Firstpage :
1273
Lastpage :
1286
Abstract :
Let P={p(i)} be a measure of strictly positive probabilities on the set of nonnegative integers. Although the countable number of inputs prevents usage of the Huffman algorithm, there are nontrivial P for which known methods find a source code that is optimal in the sense of minimizing expected codeword length. For some applications, however, a source code should instead minimize one of a family of nonlinear objective functions, beta-exponential means, those of the form loga Sigmaip(i)an(i), where n(i) is the length of the ith codeword and a is a positive constant. Applications of such minimizations include a novel problem of maximizing the chance of message receipt in single-shot communications (a<1) and a previously known problem of minimizing the chance of buffer overflow in a queueing system (a>1). This paper introduces methods for finding codes optimal for such exponential means. One method applies to geometric distributions, while another applies to distributions with lighter tails. The latter algorithm is applied to Poisson distributions and both are extended to alphabetic codes, as well as to minimizing maximum pointwise redundancy. The aforementioned application of minimizing the chance of buffer overflow is also considered.
Keywords :
Huffman codes; Poisson distribution; geometric codes; nonlinear codes; queueing theory; Huffman algorithm; Poisson distributions; alphabetic codes; beta-exponential means; buffer overflow; codeword length; geometric distributions; infinite alphabets; maximum pointwise redundancy; nonlinear objective functions; optimal prefix codes; queueing system; single-shot communications; Binary codes; Buffer overflow; Cost function; Electronic mail; Entropy; Information theory; Minimax techniques; Probability distribution; Research initiatives; Source coding; Communication networks; Golomb codes; Huffman algorithm; generalized entropies; generalized means; optimal prefix codes; queueing; worst case minimax redundancy;
fLanguage :
English
Journal_Title :
Information Theory, IEEE Transactions on
Publisher :
ieee
ISSN :
0018-9448
Type :
jour
DOI :
10.1109/TIT.2007.915696
Filename :
4455729
Link To Document :
بازگشت