Deep-Learning-Aided Design of LDPC Coding Scheme for Two-User Gaussian Multiple Access Channels

Abstract

We propose a deep-learning-aided low-density parity-check (LDPC) coding scheme for two-user Gaussian multiple access channels with equal rate and equal average power constraints. The proposed neural network model simultaneously optimizes the power allocations and the joint decoder for a given power constraint and LDPC codes. Numerical results show that the proposed scheme achieves a better bit error rate performance than existing schemes for shorter code lengths and fewer decoding iterations.