Hybrid neural coded modulation: Design and training methods

Hybrid neural coded modulation: Design and training methods

We propose a hybrid coded modulation scheme which composes of inner and outer codes. The outer-code can be any standard binary linear code with efficient soft decoding capability (e.g. low-density parity-check (LDPC) codes). The inner code is designed using a deep neural network (DNN) which takes th...

Full description

Bibliographic Details
Main Authors: Sung Hoon Lim, Jiyong Han, Wonjong Noh, Yujae Song, Sang-Woon Jeon
Format: Article
Language:English
Published: Elsevier 2022-03-01
Series:ICT Express
Subjects:
Machine learning
Neural networks
Modulation
Channel coding
Generalized mutual information
Online Access:http://www.sciencedirect.com/science/article/pii/S2405959522000182
Description
Summary:We propose a hybrid coded modulation scheme which composes of inner and outer codes. The outer-code can be any standard binary linear code with efficient soft decoding capability (e.g. low-density parity-check (LDPC) codes). The inner code is designed using a deep neural network (DNN) which takes the channel coded bits and outputs modulated symbols. For training the DNN, we propose to use a loss function that is inspired by the generalized mutual information. The resulting constellations are shown to outperform the conventional quadrature amplitude modulation (QAM) based coding scheme for modulation order 16 and 64 with 5G standard LDPC codes.
ISSN:2405-9595

Similar Items