The Role of AI and Machine Learning in the Evolution of 5G and Beyond Networks
Y.L. Malathi Latha, Saeed Rabbani, Chiranjeevi Manike, Sandip J. Gami, M. Kishore Kumar, Siva Koteswara Rao Katta · 2024
This article will examine the current capabilities of 5G and B5G networks in terms of artificial intelligence and machine learning. The research delves into several applications of AI and ML within the telecom industry. Neural networks are introduced as a class of tools for decision-making and statistical data modelling that are often nonlinear. Data pattern discovery and the representation of complex connections between input and output parameters in systems are common uses for these. This class of neural networks includes recurrent, convolutional, feed-forward, and deep neural networks. When improving a system property or maximising a collective reward, reinforcement learning focuses on the behaviours that intelligent agents must do. One advantage of deep reinforcement learning is its ability to handle unstructured input. This is achieved by combining deep neural networks. We offer hybrid solutions, like expert knowledge aided machine learning and mixed analytical and machine learning models. Lastly, further targeted approaches are showcased, including GANs, unsupervised learning, and clustering. CF M-MIMO technology, which stands for cell-free massive MIMO, is the most promising idea in B5G. The appealing aspects of CF M-MIMO systems, such as power distribution and channel estimation, are with their downsides. There have been numerous successful applications of deep learning (DL) across many scientific domains, including wireless communications. According to the research by Lazaros Alexios Iliadis et al., CF M-MIMO networks use cutting-edge DL techniques. We provide the most popular DL models and review cell-free network fundamentals here. The endnotes emphasise the potential avenues for further study.