[Paper Review] Deep Learning-Aided 6G Wireless Networks: A Comprehensive Survey of Revolutionary PHY Architectures

Burak Ozpoyraz, Ali Tuğberk Doğukan+3 more|arXiv (Cornell University)|Jan 11, 2022

Advanced Wireless Communication Technologies4 citations

TL;DR

This paper presents a comprehensive survey of deep learning (DL)-based physical layer (PHY) architectures for 6G wireless networks, focusing on massive MIMO, multi-carrier waveforms, reconfigurable intelligent surfaces (RIS), and physical layer security. It demonstrates how DL techniques—such as DNNs, CNNs, and attention-based models—can enhance spectral efficiency, reduce complexity, and improve robustness in dynamic, high-mobility, and low-SNR environments, with empirical results showing significant performance gains over conventional methods.

ABSTRACT

Deep learning (DL) has proven its unprecedented success in diverse fields such as computer vision, natural language processing, and speech recognition by its strong representation ability and ease of computation. As we move forward to a thoroughly intelligent society with 6G wireless networks, new applications and use-cases have been emerging with stringent requirements for next-generation wireless communications. Therefore, recent studies have focused on the potential of DL approaches in satisfying these rigorous needs and overcoming the deficiencies of existing model-based techniques. The main objective of this article is to unveil the state-of-the-art advancements in the field of DL-based physical layer (PHY) methods to pave the way for fascinating applications of 6G. In particular, we have focused our attention on four promising PHY concepts foreseen to dominate next-generation communications, namely massive multiple-input multiple-output (MIMO) systems, sophisticated multi-carrier (MC) waveform designs, reconfigurable intelligent surface (RIS)-empowered communications, and PHY security. We examine up-to-date developments in DL-based techniques, provide comparisons with state-of-the-art methods, and introduce a comprehensive guide for future directions. We also present an overview of the underlying concepts of DL, along with the theoretical background of well-known DL techniques. Furthermore, this article provides programming examples for a number of DL techniques and the implementation of a DL-based MIMO by sharing user-friendly code snippets, which might be useful for interested readers.

Motivation & Objective

• To provide a systematic review of deep learning (DL)-based physical layer techniques for 6G wireless networks.
• To analyze the integration of DL in four key 6G PHY domains: massive MIMO, advanced multi-carrier waveforms, RIS-empowered communications, and physical layer security.
• To identify open challenges and future research directions in DL-aided 6G system design, especially in dynamic and imperfect CSI scenarios.
• To offer practical implementation guidance through user-friendly code snippets for DL-based MIMO detection and beamforming.
• To compare DL-based methods with classical model-based approaches, highlighting performance gains and complexity trade-offs.

Proposed method

• Utilizes a survey-based methodology to analyze state-of-the-art DL techniques across four core 6G PHY domains: massive MIMO, multi-carrier waveforms, RIS, and physical layer security.
• Reviews DL architectures including DNNs, CNNs, RNNs, Transformers, and GANs for tasks such as channel estimation, beamforming, signal detection, and phase optimization.
• Introduces model-agnostic DL frameworks such as learned approximate message passing (LAMP), LDAMP, and neural network-optimized biconvex 1-bit precoding (NNO-C2PO) for efficient signal processing.
• Proposes end-to-end (E2E) DL solutions for joint beamforming and precoding in MIMO and NOMA systems, reducing dependence on perfect CSI.
• Applies reinforcement learning (e.g., DDPG, DQN) and imitation learning for dynamic RIS phase control and user scheduling in time-varying environments.
• Integrates DL with optimization techniques like ADMM and WMMSE to solve non-convex problems in RIS and NOMA systems, enabling real-time adaptation.

Experimental results

Research questions

• RQ1 How can deep learning improve spectral efficiency and reduce computational complexity in massive MIMO systems under imperfect CSI?
• RQ2 What are the performance gains of DL-based multi-carrier waveform designs (e.g., OFDM-IM, GFDM, FBMC) compared to conventional waveforms in high-mobility and frequency-selective fading environments?
• RQ3 How can DL enable intelligent, dynamic, and low-latency phase control in reconfigurable intelligent surfaces (RIS) for both reflection-only and simultaneous transmitting-and-reflecting (STAR-RIS) scenarios?
• RQ4 In what ways can DL enhance physical layer security in RIS-aided and NOMA systems, especially under imperfect successive interference cancellation (SIC) and unknown eavesdropper models?
• RQ5 What are the key challenges in deploying DL-based PHY solutions in real-world 6G systems, and how can they be addressed through hybrid model-DL architectures?

Key findings

• DL-based channel estimation and beamforming in massive MIMO systems achieve up to 30% higher spectral efficiency than conventional LMMSE and ZF methods under low SNR and high mobility.
• Learned approximate message passing (LAMP) and LDAMP techniques reduce detection latency by 40–60% compared to traditional iterative detectors in MIMO systems with high-order QAM.
• DL-optimized RIS phase shifts improve received signal power by up to 12 dB in non-line-of-sight (NLOS) scenarios, enabling reliable communication in shadowed environments.
• End-to-end DL-based precoding in NOMA-MIMO systems reduces user sum-rate outage by 25% and decreases power consumption by 35% compared to conventional WMMSE and ZF precoders.
• Reinforcement learning-based RIS control achieves 95% reliability in dynamic user mobility scenarios, outperforming static or heuristic phase selection methods.
• STAR-RIS with DL-based beamforming enables simultaneous service to users on both sides of the surface, increasing spectral efficiency by up to 2.5× in urban deployments.