A Novel Approach for On-line Encoding Algorithm Using A Generating Function

Abstract

An encoding technique is an interesting problem in information theory (i.e., how to reduce the size of the textual material when information is continually sent from one point to another.) The important factors to improve efficiency of the encoding process are probability distribution and dependency of the input data. Some encoding algorithms focused on these factors are Huffman algorithm, Lempel-Ziv algorithm and enhance versions of them. In some cases, dependency of the input data is not significant, but the probability distribution remains considerable. In this paper, we propose a novel approach for an on-line encoding algorithm using a generating function that the encoding process performs in an on-line manner. Our concept is that the generating function must be constructed up to the probability distribution. Some experimental results show that our technique can apply to the normal distribution of input data