Deep learning en la capa física de comunicaciones

Abstract

Abstract: Artificial Intelligence (AI) in recent years has had a great implication, with great advances in all fields of knowledge such as: medicine, education, industry, science, etc. One of the keys of AI is deep learning (Deep Learning DL), which has proven to be an effective solution for use in many areas, such as: computer vision, robotics, natural language processing due to its advanced algorithms and tools for learning complicated models. Similarly, DL has recently shown great power to revolutionize communication systems. That is why this research analyzes the differences between wireless communication systems based on OFDM (Orthogonal Frequency Division Multiplexing) and communication systems based on DL. Taking into account the criteria of performance, confidentiality, integrity, reduction of computer resources, information security and data transmission time. In addition, the potential use of DL techniques is identified and validated, through the use of convolutional neural networks (CNN). For the development of the investigation, the wireless communication system is considered as a black box. End-to-end DL-based communications replace the processing blocks of: receiver, channel estimation, equalization, demodulation, and signal detection; by the CNN neural network model. The results of the algorithm that DL integrates aim to reduce the sending bit error (BER) and improve the performance of the wireless communication system.