Extremely large-scale MIMO (XL-MIMO) is regarded as a pivotal enabler for achieving ultra-high spectral efficiency in 6G communications. Near-field channel models, which integrate both line-of-sight (LoS) and non-line-of-sight (NLoS) components, provide accurate characterizations of near-field XL-MIMO channels. However, existing channel estimation schemes encounter severe performance bottlenecks due to the high-dimensional nature of near-field XL-MIMO channels and their structured angular sparsity compared to far-field MIMO systems. To address these challenges, we propose a variance exploding stochastic differential equation (VE-SDE) generator based on an improved diffusion transformer (IDiT) network. The VE-SDE progressively maps the complex XL-MIMO channel distribution to a tractable prior distribution by gradually injecting noise. We utilize the patchify technique to decompose the perturbed angular domain channels into token sequences, which are then processed with diffusion transformer (DiT) blocks, substantially reducing floating-point operations (FLOPs). Additionally, a sparse self-attention mechanism is employed to enhance structured sparsity characterization learning, thereby improving estimation accuracy. Theoretical analysis and numerical experiments show that the VE-SDE generator exhibits strong generalizability and robustness across diverse channel distributions without requiring retraining. Simulation results reveal that the proposed method outperforms state-of-the-art estimation approaches, achieving high-fidelity channel estimation with only 20% pilot density.

Recent advancements in deep learning have shown considerable potential to enhance radar target detection, particularly in improving detection probability under complex environmental conditions. However, existing deep learning approaches largely operate in the real number domain, neglecting the complex-valued nature of radar data, and often inherit vision-oriented architectures that fail to address radar-specific challenges-such as sparse target echoes, the necessity for phase preservation, and constraints imposed by scanning radar systems. Meanwhile, conventional radar signal processing methods, including CA-CFAR, are limited by their dependence on idealized statistical models and often underperform in dynamic and cluttered electromagnetic environments.To overcome these issues, this paper proposes Radar Transformer for Detection (RaTDet), an end-to-end detection network that integrates complex-valued convolutional neural networks (CNNs) and Transformers. RaTDet fully leverages complex-valued data to preserve critical phase and amplitude information, enabling automated feature learning directly from raw radar signals. The model operates effectively with very few pulses, making it suitable for resource-constrained scenarios, and can serve as a pre-trained foundation model for various radar downstream tasks. Experimental results demonstrate that RaTDet achieves excellent detection performance, characterized by high detection probability (Pd) and low false alarm rate (Pfa), outperforming both traditional signal processing and conventional deep learning methods. This work bridges the gap between deep learning and radar signal processing, offering a flexible and powerful network for next-generation radar systems.

We investigate power-efficient multiuser integrated sensing and communication (ISAC) assisted by an element-grouping extremely large-scale intelligent reflecting surface (EG-XL-IRS). The grouping pattern is designed using slowly varying statistical channel state information (S-CSI), so that both IRS-related channel acquisition and online passive beamforming operate in the group domain rather than the element domain. We reveal a fundamental gain-rank tradeoff induced by element grouping: phase-consistent grouping can coherently enhance selected deterministic propagation components, while excessive concentration on a common deterministic mode can reduce the effective spatial rank of the multiuser channel and, for extended targets, the diversity of desired-scatterer responses. Motivated by this observation, we develop a task-adaptive rank-aware grouping strategy that balances weak-user enhancement and target-scatterer illumination while preserving task-relevant spatial dimensions. For each candidate grouping pattern, the transmit covariances and group-wise reflection phases are jointly optimized under communication and sensing quality-of-service constraints, followed by physical phase recovery and feasibility verification. Numerical results show that the proposed design substantially reduces the required transmit power compared with representative grouping benchmarks under the same grouping dimension and online optimization budget.

This paper investigates an intelligent reflecting surface (IRS)-aided wireless-powered communication network (WPCN) for maximizing the weighted sum rate (WSR). To reduce the complexity of traditional model-driven algorithms and improve convergence in data-driven deep learning approaches, a novel hybrid block coordinate descent (BCD) algorithm motivated by the dilation extraction and context attention (DECA) neural network (NN) is proposed. Specifically, the WSR maximization problem is firstly reformulated as a more tractable form, enabling the BCD algorithm to efficiently optimize the decoupled variables within the constraints. Meanwhile, at each BCD iteration, the DECA NN accelerates IRS phase shift optimization by facilitating the majorization-minimization (MM) algorithm to solve the computationally intensive fractional programming problem. Moreover, by leveraging dilation convolution and high-speed attention mechanisms, the DECA NN significantly outperforms existing deep learning benchmarks in both precision and convergence speed. Numerical results show that the proposed hybrid framework delivers performance comparable to the traditional BCD algorithm with dramatically reduced time consumption, while consistently maintaining robust performance under imperfect CSI and exhibiting strong transferability across diverse communication scenarios.

This letter investigates uplink power allocation in cell-free massive multiple-input multiple-output (CFmMIMO) systems supporting ultra-reliable low-latency communication (URLLC), enabled by federated learning (FL). Building upon our prior theoretical work, which established a provably convergent power allocation framework based on fractional programming (FP) that accounts for wireless transmission errors, we further extend the analysis to a data-driven paradigm. Specifically, we leverage a closed-form finite blocklength uplink rate expression to formulate a multi-objective optimization that jointly incorporates FL convergence performance, uplink latency, and energy consumption. To tackle the resulting non-convex and tightly coupled power allocation problem with multiple constraints, we design a graph convolutional network (GCN)-based power allocation algorithm. Distinct from topology-agnostic deep neural networks and computationally intensive iterative optimization methods, the proposed algorithm explicitly leverages the spatial topology of access points (APs) through graph-based hierarchical feature aggregation. Simulation results demonstrate that GCN-PANet significantly enhances FL convergence with URLLC requirements, validating both theoretical soundness and practicality.

As a revolutionary paradigm for intelligently controlling wireless channels, intelligent reflecting surface (IRS) has emerged as a promising technology for future sixth-generation (6G) wireless communications. While IRS-aided communication systems can achieve attractive high performance gain, existing schemes require plenty of IRS elements to mitigate the "multiplicative fading" effect in cascaded channels, leading to high complexity for real-time beamforming and high signaling overhead for channel estimation. In this paper, the concept of sustainable intelligent element-grouping IRS (IEG-IRS) is proposed to overcome those fundamental bottlenecks. Specifically, based on the statistical channel state information (S-CSI), the proposed grouping strategy intelligently pre-divide the IEG-IRS elements into multiple groups based on the beam-domain grouping method, with each group sharing the common reflection coefficient and being optimized in real time using the instantaneous channel state information (I-CSI). Then, we further analyze the asymptotic performance of the IEG-IRS to reveal the substantial capacity gain in an extremely large-scale IRS (XL-IRS) aided single-user single-input single-output (SU-SISO) system. In particular, when a line-of-sight (LoS) component exists, it demonstrates that the combined cascaded link can be considered as a "deterministic virtual LoS" channel, resulting in a sustainable squared array gain achieved by the IEG-IRS. Finally, we formulate a weighted-sum-rate (WSR) maximization problem for an IEG-IRS-aided multiuser multiple-input single-output (MU-MISO) system and a two-stage algorithm for optimizing the beam-domain grouping strategy and the multi-user active-passive beamforming is proposed. Simulation results validate the superiority of our proposed two-stage algorithm in low pilot overhead conditions and show that in the context of an XL-IRS aided MU-MISO system, the proposed IEG-IRS can achieve a significant WSR gain, thus overcoming this performance drawback associated with high complexity and signaling overhead.

Reconfigurable intelligent surface (RIS) plays an essential role in alleviating severe path loss in millimeter wave communication systems. Its performance hinges on the precise modeling of high-dimensional cascaded channels. However, traditional modeling approaches require extensive experience in radio propagation, resulting in complex and inefficient processes. To overcome these limitations, we transform the RIS channel modeling into a channel distribution transport mapping problem and introduce a generative model based on rectified flow. Our approach integrates distance information into a diffusion transformer (DiT) architecture through cross-attention mechanisms, resulting in a conditional DiT capable of synthesizing target channels from distance inputs. We further optimize the rectified flow into a single-step generator via reflow techniques. Building on this framework, we design a generative digital twin (DT) channel model that serves as a high fidelity data generator for downstream tasks. The proposed model acts as a virtual replica of the propagation environment, enabling efficient channel data synthesis for training communication algorithms such as channel state information feedback and channel estimation. Simulation results show that our approach generates channels with minimal distribution discrepancy compared to real channels (a maximum mean discrepancy < 0.01), outperforming existing generative methods. Furthermore, the reflow-driven DT channel model achieves the shortest generation time among all evaluated benchmarks.

Joint radar and communication beamforming in near-field extremely large-scale multiple-input multiple-output (XL-MIMO) integrated sensing and communication (ISAC) systems requires simultaneous optimization of communication spectral efficiency and sensing detection performance. Existing optimization-based approaches typically rely on accurate channel estimation and convex relaxation, which impose high computa tional complexity. In this paper, we propose a deep learning (DL) based transceiver beamforming framework that jointly optimizes sensing and communication beamformers without explicit chan nel estimation. To better achieve the dual performance, we first develop a generic beamforming formulation that enables unified optimization of the sensing and communication beamforming matrices. Based on this structure, we design an end-to-end joint radar and communication beamforming network (JRCnet), which integrates the derived algorithmic structure with data driven learning to enhance beamforming inference. JRCnet employs a multi-head cross-covariance attention mechanism and depth-wise convolution layers to significantly reduce inference complexity while preserving beamforming performance for both communication and sensing. Furthermore, the framework is extended to an enhanced spatio-temporal JRCnet (STJRC) to accommodate user mobility scenarios. Extensive simulations and complexity analysis validate that the proposed framework achieves excellent performance, particularly in large-scale an tenna array scenarios.

Beamforming enhances signal strength and quality by focusing energy in specific directions. This capability is particularly crucial in cell-free integrated sensing and communication (ISAC) systems, where multiple distributed access points (APs) collaborate to provide both communication and sensing services. In this work, we first derive the distribution of joint target detection probabilities across multiple receiving APs under false alarm rate constraints, and then formulate the beam selection procedure as a Markov decision process (MDP). We establish a deep reinforcement learning (DRL) framework, in which reward shaping and sinusoidal embedding are introduced to facilitate agent learning. To eliminate the high costs and associated risks of real-time agent-environment interactions, we further propose a novel digital twin (DT)-assisted offline DRL approach. Different from traditional online DRL, a conditional generative adversarial network (cGAN)-based DT module, operating as a replica of the real world, is meticulously designed to generate virtual state-action transition pairs and enrich data diversity, enabling offline adjustment of the agent's policy. Additionally, we address the out-of-distribution issue by incorporating an extra penalty term into the loss function design. The convergency of agent-DT interaction and the upper bound of the Q-error function are theoretically derived. Numerical results demonstrate the remarkable performance of our proposed approach, which significantly reduces online interaction overhead while maintaining effective beam selection across diverse conditions including strict false alarm control, low signal-to-noise ratios, and high target velocities.

The convergence of high-throughput communications and high-precision sensing in sixth-generation (6G) networks is a key driver for Integrated Sensing and Communication (ISAC) frameworks. A fundamental challenge in ISAC stems from a spatial trade-off: communication performance benefits from rich scattering and high spatial Degrees of Freedom (DoFs), whereas sensing often requires a dominant Line-of-Sight (LoS) path, treating non-target reflections as clutter. This paper proposes a novel element grouping-based extremely large-scale Intelligent Reflecting Surface (EG-XL-IRS) architecture to proactively manage this trade-off. The core idea involves strategically manipulating the inherent spatial DoFs of the IRS-reflected channel by partitioning its elements into groups, guided by statistical channel state information. This grouping enables the sensing channel to operate with lower effective DoFs while allocating higher DoFs to the communication link. We formulate a joint optimization problem to minimize the total transmit power at the ISAC base station (ISAC-BS) by co-designing the active beamformers and passive EG-XL-IRS phase shifts, subject to both communication signal-to-interference-plus-noise ratio (SINR) and sensing signal-to-clutter-plus-noise ratio (SCNR) constraints. To address this non-convex problem with coupled variables and constant modulus constraints, we develop a semidefinite relaxation (SDR)-based alternating optimization (AO) algorithm. Simulation results demonstrate that the proposed scheme achieves substantial energy efficiency gains with low resource overhead compared to benchmark methods, validating the efficacy of the EG-XL-IRS in harmonizing sensing and communication functionalities.

The multiple-input multiple-output dual functional radar communication (MIMO-DFRC) system is a promising platform for future integrated sensing and communication applications. Ensuring reliable performance of both radar and communication functions, the beam selection is a critical technology in MIMO-DFRC systems. However, the beam selection problem is known to be NP-hard, and efficiently addressing it remains an open issue, especially in distributed systems. In this paper, we address the beam selection problem for a MIMO-DFRC system by formulating it as a semi-Markov decision process and propose a novel hierarchical reinforcement learning (HRL) algorithm. In our approach, codebook-based beam selection for transmitting and receiving BS is controlled by an agent deployed in the cloud. Inspired by the mechanism of hierarchical codebook beam training, we employ an option-based policy that enables the agent to explore different layers of the codebook and extract context information across multiple discrete time steps. We utilize an invalid action masking technique to overcome the dynamic action space problem caused by the option-based policy. Simulation results demonstrate that the HRL-based algorithm outperforms existing beam selection methods and achieves remarkable performance even under conditions of a high probability of false alarm and low signal-to-noise ratio. Furthermore, we find that the proposed algorithm exhibits promising capabilities to learn a more efficient policy beyond the full hierarchical codebook training trajectory.

Federated learning (FL) has garnered significant attention as an emerging distributed learning paradigm due to its inherent privacy protection features. Cell-free massive MIMO (CF-mMIMO) is a key enabler for future wireless communication systems, offering high reliability, energy efficiency, low latency, and seamless coverage to meet the demands of FL. This paper investigates an FL realization in a CF-mMIMO system with multi-antenna access points and user equipments, employing multi-datastream transmission to leverage the advantages of multi-antenna transmission. The results of FL convergence analysis under two typical cooperative reception scenarios in the CF-mMIMO system—fully-centralized processing (FCP) and large-scale fading decoding (LSFD)—are presented without predefining the transmission and reception strategies. Based on the analysis, the optimization of FL convergence performance was carried out. Through decoupling, the FL convergence performance optimization problem is ultimately transformed into a series of weighted sum-MSE minimization problems. The alternative optimization approach is employed to design iterative optimization algorithms for both cooperative reception scenarios, thereby achieving optimal convergence performance. Simulation experiments validate the effectiveness of the convergence analysis and demonstrate the advantages of multi-antenna multi-datastream transmission. Moreover, the proposed alternating iterative algorithms improve the convergence performance of FL under most experimental setups.

This letter investigates a movable-antenna (MA)-enabled integrated sensing and communication (ISAC) system and formulates a joint optimization problem for minimizing the integrated sidelobe level (ISL). Specifically, we jointly optimize the transmit beamforming and the antenna position vector subject to multiuser quality-of-service (QoS) requirements, radar illumination power constraint, and a total transmit power budget. The resulting problem is a highly nonconvex optimization problem. To address it, we develop a two-layer particle swarm optimization (TL-PSO) algorithm that incorporates semidefinite programming (SDP) relaxation in the inner layer. Simulation results show that the proposed MA-enabled ISAC system achieves pronounced sidelobe suppression by exploiting the additional spatial degrees of freedom provided by antenna position optimization and consistently outperforms a baseline with fixed-position antennas (FPAs).

To address the multiple access issue in satellite communication systems under spectrum scarcity, a semi-grant-free (SGF) transmission strategy based on non-orthogonal multiple access (NOMA) was proposed. In this strategy, mobile terminals termed as grant-free (GF) users were equipped with a planar array, while the earth station termed as grant-based (GB) user was equipped with a high gain directional antenna, and they access the satellite network simultaneously. Firstly, to ensure the quality of service for GB users, the maximum tolerable interference threshold was calculated by the GB user under the condition of only known statistical channel state information (SCSI), and was broadcasted by the satellite to all GF users. GF users who satisfy the threshold condition shared the same spectrum resource with the GB user by utilizing NOMA technology. Then, corresponding low complexity beamforming methods were proposed for GF users employing both continuous and discrete phase beamforming. Furthermore, under the assumption that the satellite channel follows a Shadowed-Rician distribution and accounts for imperfect successive interference cancellation (SIC) at the receiver, a closed-form expression for the system throughput under the proposed SGF transmission strategy was derived. Finally, computer simulations were conducted to validate the correctness of the theoretical analysis and the superiority of the proposed transmission strategy, while quantitative analyses were performed to evaluate the impact of typical parameters such as imperfect SIC, discrete bit count, and number of antennas on system performance.The results showed that when the phase discrete bit number was 3 bits, the system performance approached the level achieved with continuous phase BF.

To address the spatially non-stationary (SnS) channel estimation problem in near-field extremely large-scale multiple-input multiple-output (XL-MIMO) systems, this paper proposes a hybrid-architecture-based multi-user integrated generalized approximate message passing (MUI-GAMP) framework under a more practical antenna-wise visibility region (VR) assumption. The proposed MUI-GAMP framework effectively exploits inter-user channel correlations and the block sparsity of SnS channels in both spatial and angular domains, achieving high-accuracy channel estimation with limited pilot resources. Specifically, a subchannel classification (SC) mechanism based on normalized cross-correlation is first developed to group signals from different users that share common scattering clusters, thereby providing collections of classified signals. Building upon these collections, a hierarchical-Bayesian-inference-based joint estimation algorithm is then developed to process the signals within each collection, where a first-order Markov chain is incorporated to capture the structured sparsity of SnS channels in both spatial and angular domains, and a multi-user joint factor graph is constructed to explicitly exploit inter-user correlations through shared variable nodes. Finally, a damping strategy is introduced to ensure the stability of message passing over the large-scale loopy factor graph. Simulation results demonstrate the superior estimation accuracy and robustness of the proposed algorithm compared with existing approaches.

The quality of channel state information (CSI) feedback is critical for maximizing the spectral efficiency of massive multiple-input multiple-output systems. With multiple antenna arrays, the overhead of direct CSI feedback in frequency division duplex mode is usually large, and many CSI compression techniques have been proposed to alleviate this problem. Deep learning (DL) has achieved tremendous strides in CSI feedback. However, most current DL-based CSI compression methods utilize fully connected layers to achieve dimensionality reduction, which may be suboptimal for network optimization and result in noteworthy information loss and reduced CSI reconstruction accuracy. In this paper, we propose a novel recursive discretization compression framework with a selective state space model for CSI feedback, namely CsiMamba-RDC. The framework employs improved residual vector quantization to recursively refine CSI representation, reducing information loss and storage overhead. Additionally, we present an encoder-decoder model leveraging a selective state space model to extract diverse channel features.

In this paper, we consider robust joint access point (AP) clustering and beamforming design with imperfect channel state information (CSI) in cell-free systems. Specifically, we jointly optimize AP clustering and beamforming with imperfect CSI to simultaneously maximize the worst-case sum rate and minimize the number of AP clustering under power constraint and the discrete constraint of AP clustering. Through transformations, the intractable simultaneous optimization of continuous and discrete variables is reduced to optimizing only the sparsity of the continuous variables, facilitating a computationally efficient unsupervised deep learning algorithm. In addition, to further reduce the computational complexity, a computationally effective unsupervised deep learning algorithm is proposed to implement robust joint AP clustering and beamforming design with imperfect CSI in cell-free systems. Numerical results demonstrate that the proposed unsupervised deep learning algorithm achieves a higher worst-case sum rate under a smaller number of AP clustering with computational efficiency.

This letter addresses the frequency division duplex (FDD) massive multiple-input multiple-output (MIMO) downlink (DL) channel estimation problem in dynamic scenarios. We propose a multipath angle of arrival (AoA) and angle of departure (AoD)-based class-specific dictionary learning (MACDL) algorithm, where the discriminative information of multipath AoA/AoD is exploited via supervised dictionary learning to enhance the model's generalization ability, which enables it working well in dynamic environments without constant retraining. Simulation results show that the proposed algorithm significantly reduces the pilot overhead and achieves better stability than other channel estimation schemes.

In order to address the data security and communication efficiency of vehicles during high-speed mobile communication, this paper investigates the problem of secure in-vehicle communication resource allocation based on slow-variable large-scale fading channel information, to meet the quality of service requirements of vehicular communication, i.e., to ensure the reliability of V2V communication and the time delay while maximizing the transmission rate of the cellular link. And an eavesdropping model is introduced to ensure the secure delivery of link information. Considering that the high mobility of vehicles causes rapid channel changes, we model the problem as a Markov decision process and propose a resource allocation optimization framework based on the Multi-Agent Reinforcement Learning Algorithm (MARL-DDQN), in which a large-scale neural network model is built to train vehicular to learn the optimal resource allocation strategy for optimal communication performance and security performance. Simulation results show that the load successful delivery rate and confidentiality performance of the vehicular communication network are effectively improved compared to the baseline and MADDPG strategies while ensuring link security. This study provides useful references and practical value for the optimization of secure communication resource allocation in vehicular networking.

Cell-free massive multiple-input multiple-output (MIMO) is a promising technology to address the requirements for higher spectral efficiency and energy efficiency in 6G networks. Downlink beamforming scheme, essential for mitigating multiuser interference and enhancing overall system performance, relies on the estimated uplink channel state information (CSI) in time-division duplex (TDD) mode exploiting channel reciprocity. However, hardware impairments render the bi-directional channel non-reciprocal. This paper focuses on channel calibration for cell-free massive MIMO systems, taking into account both radio frequency (RF) mismatches and nonlinear distortions. We derive the closed-form expression for downlink achievable rate within a specific calibration scheme. To address the calibration challenge, we introduce a novel conceptual model, in which the calibration vector is determined by optimizing the performance of the reference antenna. Expanding on this concept, we propose a novel digital twin (DT)-enabled approach to overcome the limitations in the conceptual model, where the DT model is established to perform calibration task by introducing DT services of virtual reference antennas. By exploiting this method, the calibration vector is computed utilizing the proposed alternating optimization algorithm within the DT model, obviating the need for deploying reference antennas in the real cell-free system, thereby reducing costs. The communication overheads and computation complexity for updating the calibration vector is proportional to the access point (AP) number. Simulation results demonstrate the significant improvement of system performance through channel calibration and verify the higher downlink throughput of our proposed DT-enabled calibration method compared to the existing calibration methods.