Mathematical Modeling and Simulation of Composite Wireless Channels for QAM-Based Image Transmission and Datasets Generation

Abstract

The advances towards 5G and 6G technologies in modern wireless communication systems have increased the demand for strong transmission and reception of signals over fading channels. The work provided in the contributions herein covers simulating a composite fading channel that incorporates the Rician and Rayleigh models into the AWGN noise and transmitting images modulated under the scheme of various QAMs including 16-, 64-, 1024and 4096-QAM at different SNR values. The aim is to generate modulation I/Q signals datasets suitable for deep learning applications, in particular, Automatic Modulation Classification using Convolutional Neural Networks. These datasets represent real-world wireless environments, modeling 4G to 6G technologies. Applying some preprocessing techniques and CNN models, in particular ResNet50 and LeNet, for classification of these modulation schemes yielded promising results for cognitive radio and spectrum management. It indicates that the most important role of CNN-based AMC is in enhancing the spectrum efficiency of wireless communication systems.