A Recursive Algorithm for Quantizer Design for Binary-Input Discrete Memoryless Channels
Mehdi Dabirnia, Alfonso García Martínez, Albert Guillén i Fàbregas · Repository for Publications and Research Data (ETH Zurich) · 2020
The quantization of the outputs of a binary-input discrete memoryless channel is considered.A new recursive method for finding all optimal quantizers for all output cardinalities is proposed.Two different versions of the newly proposed method for top-down and bottom-up approaches are developed which provide an improved understanding of the quantization problem under consideration.Also, an efficient algorithm based on dynamic programing is proposed and shown to have a comparable complexity with the state of the art.