Suboptimal quantizer design for outputs of discrete memoryless channels with a finite-input alphabet

摘要

Some quantization techniques are used in practice system to reduce the complexity of coding problem. Kurkoski, Yamaguchi, and Kobayashi proposed a suboptimal algorithm that design a channel output quantizer in the sense of maximizing mutual information between the channel input and the quantizer output for any finite-input discrete memoryless channels and fixed channel input distribution by using greedy algorithm. Moreover, Kurkoski and Yagi considered only binary-input case, and proposed an algorithm for finding the optimal quantizer design of channel output for any binary-input discrete memoryless channel and fixed channel input distribution by using dynamic programming. In this paper, we evaluate the precision of the former algorithm proposed by Kurkoski, Yamaguchi, and Kobayashi.