Multiple-input multiple-output (MIMO) orthogonal frequency-division multiplexing (OFDM) systems rely on digital or hybrid digital and analog designs for beamforming against frequency-selective fading, which suffer from high hardware complexity and energy consumption. To address this, this work introduces a fully-analog stacked intelligent metasurfaces (SIM) architecture that directly performs wave-domain beamforming, enabling diagonalization of the end-to-end channel matrix and inherently eliminating inter-antenna interference (IAI) for MIMO OFDM transmission. By leveraging cascaded programmable metasurface layers, the proposed system establishes multiple parallel subchannels, significantly improving multi-carrier transmission efficiency while reducing hardware complexity. To optimize the SIM phase shift matrices, a block coordinate descent and penalty convex-concave procedure (BCD-PCCP) algorithm is developed to iteratively minimize the channel fitting error across subcarriers. Simulation results validate the proposed approach, determining the maximum effective bandwidth and demonstrating substantial performance improvements. Moreover, for a MIMO OFDM system operating at 28 GHz with 16 subcarriers, the proposed SIM configuration method achieves over 300% enhancement in channel capacity compared to conventional SIM configuration that only accounts for the center frequency.

Stacked intelligent metasurfaces (SIM) enable fine-grained wave-domain signal processing, but their wideband deployment is impeded by two structural factors: (i) a single, quasi-static SIM phase tensor must adapt to all subcarriers, and (ii) multiuser scheduling changes the subcarrier activation pattern frame by frame, requiring rapid reconfiguration. To address these, we propose a SIM-enhanced wideband multiuser transceiver built on orthogonal frequency-division multiplexing with index modulation (OFDM-IM). The sparse activation of OFDM-IM confines high-fidelity equalization to the active tones, effectively widening the usable bandwidth. To make the design reliability-aware, we directly target the worst-link bit-error rate (BER) and adopt a max-min per-tone signal-to-interference-plus-noise ratio (SINR) as a principled surrogate, turning the reliability optimization tractable. For frame-rate inference and interpretability, we propose an unfolding projected-gradient-descent network (UPGD-Net) that unrolls across the SIM's layers and algorithmic iterations with a learnable per-iteration step size. Simulations demonstrate that the proposed framework achieves fast convergence and significant BER gains over fully-digital baselines. Notably, the design exhibits superior robustness against errors and outperforms large-aperture hybrid precoding benchmarks in both sum rate and energy efficiency. By combining structural sparsity with a BER-driven, deep-unfolded optimization backbone, the proposed framework effectively resolves the key wideband deficiencies of SIM.

Conventional digital beamforming for wideband multiuser orthogonal frequency-division multiplexing (OFDM) demands numerous power-hungry components, increasing hardware costs and complexity. By contrast, the stacked intelligent metasurfaces (SIM) can perform wave-based precoding at near-light speed, drastically reducing baseband overhead. However, realizing SIM-enhanced fully-analog beamforming for wideband multiuser transmissions remains challenging, as the SIM configuration has to handle interference across all subcarriers. To address this, this paper proposes a flexible subcarrier allocation strategy to fully reap the SIM-assisted fully-analog beamforming capability in an orthogonal frequency-division multiple access (OFDMA) system, where each subcarrier selectively serves one or more users to balance interference mitigation and resource utilization of SIM. We propose an iterative algorithm to jointly optimize the subcarrier assignment matrix and SIM transmission coefficients, approximating an interference-free channel for those selected subcarriers. Results show that the proposed system has low fitting errors yet allows each user to exploit more subcarriers. Further comparisons highlight a fundamental trade-off: our system achieves near-zero interference and robust data reliability without incurring the hardware burdens of digital precoding.

To provide customized high-quality services for all users in the sixth-generation (6G) wireless communication systems, it is fundamental to study all 6G channel scenarios and establish accurate channel models for these scenarios correspondingly. However, the absence of comprehensive 6G scenario categorization and the difficulties of modeling the channels for all scenarios bring huge challenges. In this article, we aim to give a thorough overview of channel scenarios, identification algorithms, and intelligent channel modeling theories. First, different standardized scenario categorization principles are reviewed. A unified and exclusive scenario categorization method is elaborated with detailed 6G scenario definitions. Second, scenario features, feature selection principles, ML-based identification algorithms, as well as data preprocessing methods are surveyed for the benefit of accurate scenario identification. Third, the intelligent scenario adaptive channel modeling theory based on 6GPCM is specified. Statistical properties for industrial IoT and HST scenarios are simulated and compared with those from measurements. Finally, future research directions and challenges are addressed.

The development of sixth-generation (6G) wireless communication systems brings significant challenges in channel modeling. Conducting channel measurements for 6G communications is highly expensive and cannot cover all scenarios and frequency bands. Moreover, existing conventional channel models fail to accurately predict channel characteristics in unknown frequency band. As a result, predictive channel modeling has emerged as a promising solution for addressing these challenges in 6G channel modeling. In this study, we propose a frequency domain predictive channel model that combines an autoencoder with a coupling Convolution Gated Recurrent Unit (Conv-GRU) cells. The proposed model aims to predict channel characteristics in unknown frequency bands. The proposed predictive channel model is validated by using data collected from multiple frequency bands channel measurements. To evaluate its performance, several commonly used prediction networks, i.e., a general LSTM network, a GRU-based predictive network, and a Conv-LSTM-based predictive network, are conducted as benchmarks for comparison. Based on evaluation results, our proposed predictive channel model achieves the highest level of accuracy in predicting channels. Additionally, we provide a performance bound for extrapolation predictability using a Ray Tracing simulator.

Artificial intelligence (AI) is a promising solution to achieve channel prediction under limited channel data. In this paper, a novel scatterer density-based predictive channel model is proposed to predict channels in multiple scenarios. By exploring the graph attention networks (GAT) and gated recurrent unit (GRU), the proposed model captures multi-domain information in dynamic scenarios. Besides, it extracts highly space-time correlated data characteristics, captures channel dynamic evolutional patterns, and predicts channels in different scenarios. The space-time graph channel datasets are constructed based on the ray tracing (RT) simulation channels. In the prediction experiments, the proposed method is validated on the datasets to predict channels with good performance. Compared with the 3GPP TR 38.901 channel model, the proposed model obtains more accurate channel statistical properties in different scenarios.

Urban wireless communications are one of most important scenarios in the sixth generation (6G) global communication networks. To provide customized high-quality services for all users in various 6G urban wireless communication scenarios, it is necessary and fundamental to study all kinds of 6G urban wireless channel scenarios and establish corresponding channel models for each scenario. However, existing standardized channel models are insufficient to cover all 6G urban communication scenarios. This paper aims to extend the conventional urban communication scenarios with detailed definitions and environmental parameters to accurately establish corresponding channel models. Specifically, the statistical properties of high speed train (HST) scenario are simulated and compared with those of measurement data. The channel model simulation results match well with the channel measurement data, which demonstrates the correctness of the channel model. Then, by applying the proposed precise scenarios and the corresponding model parameters to the channel model, corresponding channel characteristics can be quickly provided. Finally, the channel characteristics of several urban scenarios are simulated and analyzed.

Compared with conventional passive channel modeling, artificial intelligence (AI) based channel models show great advantages in solving real-time prediction problems in wireless communications. In this paper, a generative adversarial network (GAN) and long short-term memory (LSTM) based channel prediction framework is proposed to model indoor wireless channels. By using GAN and LSTM, the model not only enriches the channel data but also achieves the sequence prediction, which can solve the problem of the shortage of training data and prediction channels in the space domain. The prediction performance is evaluated by comparing the root mean square error (RMSE) and mean absolute percentage error (MAPE) of measured data and predicted data. By comparing the statistical properties of the channel measurement data and of the synthetic data, it can be found that the proposed model can predict unknown information in the space domain.

In the military and civilian domains, the modulation classification in the communication system is an extremely important technology that needs to be constantly updated and improved. In this paper, we present an automatic modulation classification (AMC) model to do modulation classification in 5 typical types of signal modulation BPSK, QPSK, 8PSK, 16QAM, and 64QAM. The proposed algorithm uses an improved AlexNet with deep residual learning, regularization, global pooling, and PReLU activation function to extract features from constellation diagrams for better recognition performance. Compared with the original AlexNet, support vector machine (SVM), and the traditional maximum likelihood-based cumulant technique, experiment results indicate that the proposed AMC model with the improved AlexNet has achieved very good recognition results, with its robustness, generalization, and high efficiency. We also explore the identifiability and reliability of signal transmission under different SNR conditions for different modulation types.