Deep Communication: Exploring End-to-End Wireless with Convolutional Neural Network

Abstract

In recent years, end-to-end wireless communication has gained significant attention in the field of wireless communication. In this paper, we propose a new approach to achieving end-to-end wireless communication using convolutional neural networks (CNNs) in the presence of Nakagami fading, Additive white Gaussian noise (AWGN), and multiple-input multiple-output (MIMO) fading channels. Further, we have applied the bursty noise to the AWGN channel. We first develop a CNN-based transmitter architecture that can efficiently encode information bits into signals, followed by a CNN-based receiver architecture that can accurately decode the received signals. Our proposed method leverages the strengths of CNNs in learning and extracting features from raw data and applies them to wireless communication. We then evaluate the performance of our proposed method by extensive sets of simulations in different AWGN, Nakagami fading, and MIMO fading channel scenarios. The simulation results show that the method proposed shows superior performance compared to existing state-of-the-art techniques at low Signal to Noise ratio(SNR) in terms of bit error rate (BER) and Binary cross-entropy loss. Our proposed method can be a promising solution for achieving end-to-end wireless communication in various practical applications.