[1] N. González-Prelcic, M. F. Keskin, O. Kaltiokallio, M. Valkama, D. Dardari, X. Shen, Y. Shen, M. Bayraktar, and H. Wymeersch, "The integrated sensing and communication revolution for 6G: Vision, techniques, and applications," Proceedings of the IEEE, 2024.

Artificial intelligence (AI) and machine learning (ML) have transformed numerous fields over the past decade, with their impact on wireless communications accelerating in recent years, particularly with the development of 6G.

One of the most significant applications of AI in wireless communications is in channel estimation and prediction. Traditional channel estimation methods leverage channel correlations across antennas, time samples, and subcarriers in multicarrier systems; however, these approaches often entail high computational complexity. Additionally, accurate channel statistics are typically required, especially for channel prediction in highly dynamic environments. Recent studies indicate that AI-driven methods can effectively balance computational efficiency and estimation accuracy. Furthermore, neural receivers have the potential to replace multiple signal processing blocks, such as channel estimation and equalization, further enhancing computational efficiency. 

In precoder and combiner design for MIMO systems, conventional approaches rely on optimization frameworks that maximize an objective function (e.g., spectral efficiency) or minimize a cost function (e.g., symbol detection error). More complex systems, such as the ones supporting ISAC functionalities, impose additional constraints, often leading to solutions with high computational demands or slow convergence rates. Model-based deep learning is an emerging paradigm that enhances traditional optimization techniques by integrating neural architectures as complementary tools rather than replacing established blocks with black-box AI models. This hybrid approach offers a promising avenue for improving the performance and efficiency of wireless communication systems.