Alphabetic AIFV Codes Constructed from Hu- Tucker codes

An alphabetic code is a code such that the order of codeword sequences coincides with the alphabetic order of source sequences. If we use one code tree, the optimal alphabetic code attaining the minimum average code length can be constructed by the Hu-Tucker algorithm. In this paper, we show that an efficient alphabetic AIFV (almost instantaneous fixed-to-variable length) code can be constructed by using three code trees and allowing at most two-bit decoding delay, and we propose a simple method to construct an alphabetic AIFV code from the Hu- Tucker code tree so that the alphabetic AIFV code can attain better compression rate than the Hu- Tucker code.