کنفرانس مهندسی برق ایران

صفحه اصلی / سی و چهارمین کنفرانس بین المللی مهندسی برق

Machine Learning for Terahertz Communications: State of the Art, Limitations, and Future Directions

نویسندگان :

Kimia Norouz¹ Ahmad Cheldavi² Ali Abdolali³

1- دانشگاه علم و صنعت ایران 2- دانشگاه علم و صنعت ایران 3- دانشگاه علم و صنعت ایران

کلمات کلیدی :

Terahertz communications،Machine learning،6G networks،Beamforming،Resource allocation،artificial intelligence

چکیده :

This review paper focuses on the integration of machine learning (ML) methods in terahertz (THz) communications, a crucial area for the achievement of the challenging tasks of 6G communications. The paper describes the problem areas in THz communications, including high path loss, increased vulnerability to environmental effects, and the limitations of conventional signal-processing techniques. The paper carries out a critical analysis of different ML categories, including supervised learning, unsupervised learning, reinforcement learning, and deep learning, applied to tasks such as channel estimation and allocation in the context of the aforementioned area. The paper highlights the benefit of integrating knowledge of electromagnetics in overcoming the limitations of usual ML techniques, which often involve deficiencies in terms of generality, complexity, and interpretability. Furthermore, it highlights the essential open issues, such as the need for data from real-world experiments, improved robustness with respect to different environments, and approaches capable of meeting the strict latency and computational budget requirements, typical of the THz communication scenario. Finally, the article concludes with future research directions, including the proposal of hybrid physics-ML models, lightweight approaches for edge devices, and the construction of experimental testbeds, with the goal of finally enabling the use of THz communication in 6G networks.