A convolution neural network based QPSK and 16QAM modulations simulator for a multiuser MIMO-OFDM transmission simulation over a Nakagami-m Fading channel

Abstract

Wireless communication has been a widespread method of data transfer in the modern era. Various modulation techniques have been found and among them 16QAM has proven to be the most efficient. This study provides a Convolutional Neural Network (CNN) approach for simulating 16QAM and QPSK modulation schemes over a multiuser MIMO-OFDM using the Keras library over Nakagami-m fading. It then examines the two modulation schemes to determine if QPSK might be a viable alternative to 16QAM. In this paper two analysis have been done, a realistic MIMO OFDM system simulation and weighted belief propagation decoding.