A comparative study of AI-based deep learning techniques in 5G and 6G wireless networks: applications, challenges, and future directions
A comparative study of AI-based deep learning techniques in 5G and 6G wireless networks: applications, challenges, and future directions
Tina Karimifard1 Mehdi Yaqoubi2
1) Master s student, Gorgan Faculty of Engineering and Technology, Department of Computer Engineering, Golestan University
2) Assistant Professor, Gorgan Faculty of Engineering and Technology, Department of Computer Engineering, Golestan University
IOI Code
:
AAHD-ACZZGZ
Publication :
National Conference on Information Technology, Nanotechnology, Artificial Intelligence and Technological Futures Studies(itnaf.ir)
Abstract :
The aim of this paper is to comprehensively review and compare AI-based deep learning techniques in 5G and 6G wireless networks, with an emphasis on applications, capabilities, challenges, and future directions. In this regard, the application of various deep learning architectures, including deep neural networks, convolutional neural networks, recurrent neural networks, short-term and long-term memory networks, autoencoders, generative networks, graph neural networks, and deep reinforcement learning, in areas such as radio resource management, channel estimation, signal detection, spectrum allocation, mobility management, energy optimization, network security, edge computing, and device-to-device communications, is examined. Findings from recent studies indicate that in 5G networks, deep learning has been mainly used as a tool for optimizing performance and automating network management; while in the expected 6G architecture, AI will transform from an auxiliary technology to an intrinsic and fundamental component of the network. 6G networks, with concepts such as network-native AI, terahertz communications, massive MIMO, reconfigurable intelligent surfaces, semantic communications, digital twins, and the integration of communications, sensing, and computing, will require a new generation of deep learning algorithms. Despite significant benefits, the widespread deployment of deep learning in wireless networks faces challenges such as the need for large and high-quality data, high computational cost, energy consumption, limited model interpretability, privacy and security concerns, generalizability in dynamic environments, and difficulty in deploying models on resource-constrained devices. Accordingly, the future path of this field will move towards the development of hybrid intelligent architectures, federated learning, knowledge-based deep learning, transfer learning, lightweight models, generative AI, and the convergence of AI with edge computing. By providing a comparative framework between the application of deep learning in 5G and 6G, this article attempts to provide a comprehensive picture of the current status, research gaps, and future trends in wireless network intelligence.
Keywords : wireless networks deep reinforcement learning neural networks resource management network native artificial intelligence