Delay guarantee is essential for wireless communication technology, which is beneficial for real-time data processing. Such guarantee manifests as hard delay constraint, whose resolution facilitates reliable task fulfillment within strict deadline while enabling efficient communication resource scheduling. In this paper, we investigate the problem of transmitting a certain amount of data within a deadline and optimizing the expected total energy in an Orthogonal Frequency Division Multiplexing (OFDM) system. We represent the problem as a finite-horizon stochastic dynamic optimization problem, and aim to derive decision rule for allocating communication transmission rates across subcarriers. We propose a method named Multicarrier Deterministic Approximation Method (MDA). First, to address the complexity of expectation computation, we approximate the expected value function. Subsequently, for the resulting deterministic optimization problem, we introduce auxiliary variable, and adopt low-complexity two-layer optimization framework. For the proposed method, we derive performance upper bound for deterministic parameter with specific value and asymptotic performance upper bound for deterministic parameter with general value. The performance bound theoretically demonstrates the superiority of the proposed method over both the equal rate method and the first slot method. Furthermore, it proves that increasing the deterministic parameter under relaxed delay constraint can achieve enhanced theoretical performance guarantee. Finally, the effectiveness of the proposed method is validated through simulations.

In this paper, we investigate the problem of jamming detection and channel estimation during multi-user uplink beam training under random pilot jamming attacks in beamspace massive multi-input-multi-output (MIMO) systems. For jamming detection, we distinguish the signals from the jammer and the user by projecting the observation signals onto the pilot space. By using the multiple projected observation vectors corresponding to the unused pilots, we propose a jamming detection scheme based on the locally most powerful test (LMPT) for systems with general channel conditions. Analytical expressions for the probability of detection and false alarms are derived using the second-order statistics and likelihood functions of the projected observation vectors. For the detected jammer along with users, we propose a two-step minimum mean square error (MMSE) channel estimation using the projected observation vectors. As a part of the channel estimation, we develop schemes to estimate the norm and the phase of the inner-product of the legitimate pilot vector and the random jamming pilot vector, which can be obtained using linear MMSE estimation and a bilinear form of the multiple projected observation vectors. From simulations under different system parameters, we observe that the proposed scheme improves the detection probability by more than 25% compared to the baseline at medium-to-high channel correlation level. In addition, the proposed scheme effectively supports zero-forcing precoding, thus enabling reliable data transmission in jamming scenarios, specifically, achieving a symbol error rate (SER) of less than 10−3at a jamming-to-signal ratio of 10 dB and a signal-to-noise ratio of 5 dB.

Accurate acquisition of channel state information (CSI) is crucial for achieving high-rate communication, yet it introduces significant training overhead and latency, particularly in high-mobility cell-free massive multiple-input multipleoutput (CF-mMIMO) communication systems with unmanned aerial vehicles (UAVs). To address this challenge, we propose a predictive beamforming and resource allocation framework that significantly reduces training overhead while enhancing system throughput. Specifically, a novel frame structure is designed in which uplink training is performed only in the first time slot of each beam tracking frame, while distributed beam tracking and predictive beamforming are applied in all subsequent slots using the extended Kalman filter (EKF) at each access point (AP). Moreover, we develop a centralized information fusion algorithm that exploits cell-free multi-point cooperation to improve estimation accuracy with low fronthaul overhead. Then, we derive the theoretical posterior Cram´er-Rao bound (PCRB) for the fused estimation and show that, under the local linear-Gaussian approximation, the predicted PCRB coincides with the covariance of the fused estimate. We further establish an explicit analytical mapping between the predicted PCRB and the uplink pilot length. Leveraging this theoretical bridge, we formulate a prediction-aware joint optimization problem involving uplink pilot length, downlink AP-user association, and power allocation to actively adapt the training overhead and maximize the effective sum spectral efficiency (SE). A low-complexity iterative algorithm based on fractional programming is proposed to solve this problem. Numerical results demonstrate that the proposed framework achieves a favorable trade-off between signaling overhead and system throughput. Specifically, it reduces the training overhead by 95% with only a 3% decrease in positioning accuracy, incurs only modest additional fronthaul overhead, and improves the effective sum SE by 31% compared to traditional schemes. Furthermore, comprehensive evaluations show that the framework remains effective under multipath fading and higher UAV velocities, and continues to benefit from cooperative gains in expanded network deployments.

Deterministic communication within the wireless domain is essential for industrial applications. However, the stochastic nature of wireless communication introduces substantial challenges for time-sensitive networking (TSN) services, which require strict end-to-end latency bounds. This paper addresses the challenge of minimizing long-term delay jitter in downlink cell-free multi-user multi-input multi-output orthogonal frequency division multiple access (MU-MIMO OFDMA) systems, subject to heterogeneous delay upper bounds and satisfaction rates. The problem involves time-space-frequency precoding constrained by user-specific delay violation probabilities and transmit power limits. To overcome the limitations of model-driven methods in handling implicit system models and the inefficiency of data-driven approaches in large action spaces, we propose a hybrid solution. Specifically, we decompose the problem into two sub-problems: rate scheduling via a constrained Markov decision process (CMDP), and instantaneous precoding through weighted sum-rate (WSR) maximization. We develop a safety reinforcement learning-based algorithm to optimize rate scheduling by allocating user weights, and a weighted minimum mean squared error (WMMSE) algorithm to solve the WSR maximization. Simulation results demonstrate that our approach effectively reduces jitter while meeting stringent delay-related requirements. In diverse TSN scenarios with heavy loading ratio, our proposed co-driven scheme achieves about 45% reduction in delay jitter compared to earliest deadline first (EDF) scheduling, while realizing user-specific delay satisfactory ratios (99.9%-99.999%).

Cell-free (CF) networks have attracted increasing attention for their effectiveness in mitigating inter-cell interference through cooperative transmission among distributed access points (APs). However, conventional terrestrial CF networks often lack spatial flexibility and struggle to adapt to dynamic environments. To overcome these limitations, we propose a new CF network served by unmanned aerial vehicles (UAVs) equipped with a three-dimensional (3D) rotatable antenna array. Combined with the UAV’s controllable 3D position, the resulting six-dimensional (6D) spatial reconfigurability enables the active beam steering of such aerial APs, thereby enhancing interference mitigation and dynamic user association. However, this design, referred to as 6D aerial rotatable antenna arrays (6DARAs), faces several critical challenges, such as high-dimensional coupled control variables, time-varying user positions, and increased channel state information (CSI) estimation overhead. To address these issues, we develop a two-timescale optimization framework that separates large-timescale 6DARA control (i.e., clustering, position, and rotation) from small-timescale signal processing. At the small-timescale, a closed-form team minimum mean-squared error decoder is derived using local and statistical CSI. At the large-timescale, 6DARA clustering is modeled as a local altruistic game and solved via a concurrent update algorithm, while 6DARA mobility is managed by an enhanced multi-agent reinforcement learning algorithm for efficient position and rotation adaptation under partial observability. Simulation results demonstrate that the proposed network and optimization framework significantly outperform existing baselines in terms of throughput, scalability, and robustness in dynamic environments.

This paper proposes a joint communication and computation scheme (JCCS) for cell-free multiple-input multiple-output (CF-MIMO) network to support federated learning (FL). The JCCS allows users to choose between a model update process or a data offloading process, where the offloaded data and the uploaded model gradient are respectively sent to the distributed processing units (DPUs) deployed on the access points (APs) for further model updates. Moreover, we define a performance metric called iteration error gap as the difference between model errors of adjacent iterations and decouple the total training time minimization problem into the error gap maximization problem within fixed time limit. Base on common machine learning (ML) assumptions, we derive a lower bound of iteration error gap, which is determined by the minimum batch size among all DPUs. An optimization problem aiming to maximize the minimum batch size is then formulated to jointly optimize the time division, power control, and user selection. By employing the block coordinate descent approach, we develop a new algorithm to solve the formulated non-convex mixed integer programming problem. Our simulation results verify the convergence of proposed algorithm and show that it reduces the relative error of a single FL iteration by more than 1.5 dB compared with other baseline schemes. Furthermore, the CF-MIMO network integrated with JCCS achieves a relative error reduction exceeding 1 dB per iteration when compared to collocated MIMO network. In addition, an FL example of handwritten digits classification shows that the JCCS indeed accelerates the convergence of FL model.

Deterministic low-latency communication is critical for emerging applications such as industrial automation and autonomous driving. Cell-free (CF) is a promising network architecture for deterministic delay communications, owing to its user-centric cooperative transmission. In this paper, we address the joint optimization problem of resource allocation and beam selection for deterministic delay in the CF network. Specifically, we propose a delay-aware hierarchical distributed deep reinforcement learning (DRL) framework that enables distributed decision-making and improves scalability in the CF network. This framework incorporates a safe reinforcement learning (RL) algorithm to effectively address the delay constraint. Simulation results demonstrate that the proposed scheme improves spectral efficiency and reduces the delay violation ratio, achieving a delay violation ratio of 10−4at a system load of around 90%, an order of magnitude lower than the 10−3achieved by the modified largest weighted delay first (MLWDF) scheme.

In unmanned aerial vehicle (UAV) communications, beam tracking faces significant challenges due to persistent variations in mobile users’ positions and flight attitudes. Fixed tracking periods result in either unnecessary or insufficient overhead depending on whether the UAV’s relative motion is static or dynamic. To address these challenges, this paper proposes a two-level beam tracking scheme enabled by data-driven beam prediction. In the upper layer, a time-series prediction method is used to adaptively adjust the tracking interval by forecasting confidence intervals of future communication quality. In the lower layer, a threshold-based probing beam selection method is implemented, which selects probing beams whose conditional likelihoods exceeding a predefined threshold. Simulation results demonstrate that the total beam training overhead drop 43.37 %, while the average outage probability decreases from 6.229 % to 0.095 %.

In this paper, a two-stage intelligent scheduler is proposed to minimize the packet-level delay jitter while guaranteeing delay bound. Firstly, Lyapunov technology is employed to transform the delay-violation constraint into a sequential slot-level queue stability problem. Secondly, a hierarchical scheme is proposed to solve the resource allocation between multiple base stations and users, where the multi-agent reinforcement learning (MARL) gives the user priority and the number of scheduled packets, while the underlying scheduler allocates the resource. Our proposed scheme achieves lower delay jitter and delay violation rate than the Round-Robin Earliest Deadline First algorithm and MARL with delay violation penalty.

Cell-free massive multiple-input multiple-output (CF-mMIMO) networks improve spectral efficiency via coordinated transmission and flexible beam selection. However, the resource allocation in such networks presents a high-dimensional optimization challenge due to the distributed architecture with multiple access points and antennas. To address this, we first formulate a beam selection problem, and then propose an efficient Q-value mixing (QMIX)-based algorithm. Furthermore, recognizing the inherent limitations of deep reinforcement learning (DRL) in practical applications, such as costly training, risky exploration phases, and suboptimal convergence speeds, we design a data-driven digital twin (DT) framework to optimize the DRL training phase. Simulation results show that our approach achieves accelerated convergence and enhanced stability compared to conventional methods. DT-based pre-training establishes a robust performance lower bound prior to real-system deployment.

With the rapid development of 5 G, managing resources to achieve high throughput and low latency has become an increasingly prominent challenge. In this paper, we propose an alternating deep reinforcement learning (ADRL) framework for dynamic resource allocation in cell-free systems, aiming to enhance the sum-rate and reduce delay violation probability by optimizing the association between access points (APs) and user equipments (UEs), along with subcarrier allocation, beamforming, and power control. To achieve efficient resource management while simplifying decision-making, the ADRL framework employs diverse deep reinforcement learning (DRL) networks to alternately optimize specific resources at different time granularities. Simulation results demonstrate improvements in both throughput and delay performance, highlighting the potential of the proposed framework in advancing industrial wireless communication systems.

In this paper, we study the joint detection and angle estimation problem for beamspace multiple-input multipleoutput (MIMO) systems with multiple random jammers. An iterative low-complexity generalized likelihood ratio test (GLRT) is proposed by transforming the composite multiple hypothesis test on the projected vector into a series of binary hypothesis tests based on the spatial covariance matrix. In each iteration, the detector implicitly inhibits the mainlobe effects of the previously detected jammers by utilizing the estimated angles and average jamming-to-signal ratios. This enables the detection of a new potential jammer and the identification of its corresponding spatial covariance. Simulation results demonstrate that the proposed method outperforms existing benchmarks by suppressing sidelobes of the detected jammers and interference from irrelevant angles, especially in medium-to-high jamming-to-noise ratio scenarios.

Cloud-edge collaboration has become a pivotal approach in edge computing, especially for applications needing real-time inference. This holds particular significance in scenarios where continuous and concurrent inference tasks have stringent latency demands. This paper considers a scenario where dynamically arriving tasks can be partially offloaded to a server with parallel computing for inference. We build a dynamic model for each task's inference process and formulate the problem of minimizing the long-term delay violation rate. To address the strategy problem of joint splitting and batch scheduling for multiple users, we adopt a sequential decision-making approach which effectively lowers the computational complexity. Meanwhile, we innovatively devise a wireless resource pool and a batch scheduling linked list to characterize the system's overall computing and communication resources. On this basis, we can evaluate different task splitting and batch scheduling strategies to make more effective decisions. Simulation results demonstrate a significant improvement in latency guarantees with millisecondlevel execution efficiency and robust scalability across diverse user scales, outperforming baseline methods under heterogeneous workloads.

Radio map, or pathloss map prediction, is a crucial method for wireless network modeling and management. By leveraging deep learning to construct pathloss patterns from geographical maps, an accurate digital replica of the transmission environment could be established with less computational overhead and lower prediction error compared to traditional model-driven techniques. While existing state-of-the-art (SOTA) methods predominantly rely on convolutional architectures, this paper introduces a hybrid transformer-convolution model, termed RM-Transformer, to enhance the accuracy of radio map prediction. The proposed model features a multi-scale transformer-based encoder for efficient feature extraction and a convolution-based decoder for precise pixel-level image reconstruction. Simulation results demonstrate that the proposed scheme significantly improves prediction accuracy, and over a 30% reduction in root mean square error (RMSE) is achieved compared to typical SOTA approaches.

Multi-timeslot multi-user communication scenarios necessitate achieving a balance among system efficiency, user fairness, and reliability under jamming attacks. Millimeter-wave (mmWave) technology can achieve high data rates, but its weak penetration capability and security vulnerabilities make the practical anti-jamming schemes critical to mitigate the attacks. Due to the time-varying nature of wireless channels, statistical quality of service (QoS) provisioning is critical for supporting real-time wireless communication. In this paper, we design a practical anti-jamming strategy for a multi-timeslot multi-user mm Wave system. By jointly designing beamforming and user scheduling, we formulate a cumulative sum-rate maximization problem subject to statistical QoS provisioning. To address the system causality and channel state uncertainty, we introduce residual performance vectors and a penalty function, which transforms the original problem into finite time domain dynamic programming. The problem is subsequently discretized to reduce computational complexity. The proposed algorithm achieves 15% and 22% performance gains over the proportional fair and greedy subcarrier allocation schemes, and approaches the ideal programming algorithm.

Ultra-reliable and low-latency communication (URLLC) with statistical quality-of-service (QoS) guarantees has garnered increasing attention. Cell-free massive multiple-input multiple-output (CF-mMIMO) emerges as a promising network architecture for such applications. This paper analyzes the reliability and latency-constrained performance in CF-mMIMO uplink transmission. First, we derive the closed-form expression for the signal-to-plus-noise ratio (SNR) distribution with maximum ratio combining (MRC) under independent non-identically distributed (i.n.i.d.) Rayleigh fading channels, precisely characterizing large-scale fading effects. Subsequently, we develop closed-form approximations for both decoding error probability and effective capacity. Finally, the QoS exponent is obtained by solving the effective bandwidth-effective capacity equation under Poisson traffic arrivals, thereby providing the delay violation probability expression. Monte Carlo simulations validate the theoretical results and demonstrate that the derived results achieve accurate approximations for both decoding error probabilities and delay violation probabilities. Furthermore, they highlight significant enhancements in reliability and latency assurance enabled by CF-mMIMO architectures.

In this paper, we investigate the hybrid traffic scheduling problem in 5G-TSN integrated systems, to achieve the endogenous deterministic communication for 5G. A novel strategy namely time-slot orchestration is proposed, and a latencyand jitter-aware traffic scheduling problem with the objective of maximizing the resource utilization is further formulated. To the best of our knowledge, it is the first time for such a problem being studied in the context of 5G-TSN integration. To solve this complex combinatorial optimization problem, a low-complexity heuristic scheduling algorithm is elaborately designed and extensively evaluated. Experimental results show that, compared with the optimal method derived from the genetic algorithm (GA) and the method based on the classical earliest deadline first (EDF) scheduling, the proposed method can efficiently enhance the endogenous deterministic communication capability of 5G by orders of magnitude, especially under high loads.

Network slicing is a key enabler for 6G mobile networks. Guaranteeing the service level agreement with the smallest amount of radio resources is a challenging problem in network slicing scenarios due to random traffic patterns and the channel environment. To this end, we propose a novel deep reinforcement learning algorithm named CGDSAC based on the conditional generative diffusion model to achieve the optimization objective while capturing the underlying environment distribution. Subsequently, we further design a digital twin (DT) enhanced version of CGDSAC named CGDSAC-DT, to address issues that CGDSAC is unsafe or has lower performance than the default strategy in the early training stages, and converges slowly. Numerical results show that our proposed method can solve the issues encountered and outperform the baseline algorithm regarding performance metrics.

As a key technology in beyond 5G and future 6G communications, RAN slicing can realize differentiated service level agreement (SLA) guarantees. In this paper, we investigate the multi-slice multi-user RAN slicing. A dynamic bandwidth allocation scheme for RAN slicing is proposed based on hierarchical intelligence, where bandwidth pre-allocation and hyper-parameter tuning for MAC layer schedulers are jointly optimized to maximize the system utility, i.e., the weighted sum of spectrum efficiency (SE) and SLA satisfaction ratio (SSR) of different slices. The problem is formulated as a twin-time scale Markov decision process (MDP), where the bandwidth pre-allocation and the scheduler parameter tuning are performed on a long-term scale (e.g., seconds) and a short-term scale (e.g., 100 ms), respectively, for which we propose a hierarchical twin-time scale Dueling deep Q learning network (TTS-DDQN) algorithm. A new reward-clipping mechanism is proposed to get a better trade-off between stabilized training and higher system utility. In order to improve the robustness to time-varying traffic patterns and non-stationary dynamic environments, we further propose a traffic-aware module for more efficient sampling of the experience pool, and a variational adversarial inverse reinforcement learning (VAIRL) module for reward automation design. Extensive simulations show that the traffic-aware TTS-DDQN in stationary scenarios and the VAIRL module embedded TTS-DDQN in non-stationary scenarios outperform existing typical DQN-based algorithms, hard slicing and non-slicing, etc.

Broadcast beam design and power control are essential for enhancing the network coverage, improving the quality of service (QoS) and reducing the energy consumption in massive multiple-input-multiple-output (MIMO) communications. To improve the broadcasting performance and decrease the power consumption in dynamic scenarios with varying user numbers and distributions, we leverage deep reinforcement learning (DRL) to jointly optimize the beam selection and power control policy, and propose a multi-agent DRL (MA-DRL) framework to address the extremely high action dimension brought by the non-convex combinational multi-objective optimization problem. To reduce the cost of performance fluctuations during the exploration of DRL, we construct a data-driven surrogate model (SM) as a virtual environment for the initial training phase, while using an empirical baseline scheme to ensure acceptable real-time performance. Simulation results demonstrate that the SM-enabled MA-DRL approach not only enhances the coverage and reduces the power consumption, but also enables safe exploration and rapid adaptation to varying user numbers and distributions. Moreover, since the optimization algorithm can interact with the SM much more quickly than the real network, a faster convergence speed can be achieved with the help of the SM.

Network slicing is essential for the sixth-generation mobile networks. Minimizing radio resources while guaranteeing service level agreement (SLA) remains challenging due to random traffic patterns and channel conditions, making policy security enforcement and underlying traffic distribution inference critical research goals. In this paper, to facilitate the design of policy agents for radio resource management, we first design a high-fidelity conditional generative diffusion model (CGDM)-driven digital twin network (DTN) to provide closed-loop interaction and pre-validation capabilities. The DTN consists of a safety-bound coarse correction method to enhance strategy SLA compliance, a model market for agent warm-up and decision-level pre-validation, and a virtual interaction environment for high-fidelity agent pre-optimization. Then, a CGDM-driven safe reinforcement learning agent based on constrained multi-agent Markov decision process, termed CGD safe actor-critic (CGDSAC), is proposed to manage inter-slice radio resources. CGDSAC balances safety and strategy quality via Lagrangian primal-dual optimization and a behavior cloning objective targeting DTN-corrected strategies, while capturing latent traffic patterns. Furthermore, CGDSAC comprehensively leverages policy warm-up, decision-level pre-validation, and policy-level pre-optimization capabilities of DTN to resolve early inferior performance, SLA jitter, and slow convergence. Numerical results confirm that the built DTN exhibits good fidelity. Under fixed slices and stable traffic pattern, the DTN-enhanced approaches outperform the best baseline with an average SLA violation relative reduction of 71.3% and an average resource block utilization relative degradation of 10.9%, achieve about convergence speed enhancement of 80% compared to native CGDSAC, and is capable of adapting to the scenarios of dynamic number of slices and varying traffic patterns through knowledge transfer and pre-validation.

The antenna weight optimization plays a vital role in improving the key performance indicators (KPIs) of massive multi-input multi-output (MIMO) systems. Due to complicated channel characteristics and numerous parameter combinations, conventional methods suffer from suboptimal performance and unaffordable complexity. Additionally, the implementation of existing intelligent algorithms in the practical system is limited by heavy overhead and potential instability associated with environmental interactions. In this paper, we propose a digital twin assisted optimization framework comprising a conditional generative adversarial network (C-GAN) aided digital twin and a deep reinforcement learning (DRL) based massive MIMO antenna weight optimization algorithm to maximize the KPI of coverage. The C-GAN aided digital twin is built to augment system performance data and offer high-precision KPI pre-validation by fitting the mapping between beamforming schemes and system performance along with the distribution of system performance over user positions. The DRL based optimization algorithm that achieves a better complexity-performance tradeoff is combined with digital twin for reduced training overhead and safe exploration. Simulation results based on quasi deterministic radio channel generator (QuaDRiGa) verify that compared to intelligent optimization methods directly conducted in practical systems, our proposed framework can reduce the overhead while achieving comparable performance by leveraging high-precision pre-validation capabilities of the proposed digital twin.

In this paper, we propose a distributed solution to address the beam selection problem in millimeter-wave (mmWave) Cell-Free Massive multiple-input multiple-output (CF-mMIMO) systems. To realize the real-time optimization under dynamic environments with high performance and low complexity, we formulate the sum-rate maximization problem in mmWave CF-mMIMO as a Markov decision process and leverage deep reinforcement learning (DRL) to optimize the beam decision. Furthermore, we propose a distributed solution via multi-agent DRL (MA-DRL) framework to handle the extremely high action dimension and reduce the fronthaul requirements for communications between the access points (APs). Numerical simulations demonstrate the superiority of the proposed solution in the real-time sum-rate performance for mobile users as well as the fronthaul requirements reduction brought by the distributed architecture.

In millimeter wave (mmWave) systems with the typical two-stage hybrid precoding structure, it is necessary to determine the optimal transmit (TX)-and-receive (RX) beam pairs through beam training. However, existing beam training methods are time-consuming and costly, making them unsuitable for cell-free mmWave systems with a large number of TX-RX pairs. In this paper, we propose a fast and efficient beam training method that adopts a hierarchical codebook design and requires only a small number of wide-sweeping beams to achieve the optimal beam pairs based on deep learning and compressed sensing. Additionally, it simultaneously supports the user-centric cell-free access point (AP) clustering and the corresponding AP beam training process. The experimental results demonstrate that our approach can accurately predict the effective beams with high accuracy while significantly reducing the beam training time and overhead.

In the context of Ultra-Reliable Low Latency Communication (URLLC) scenarios, 5G incorporates numerous enhancements, with link adaptation (LA) being one of them. In the pursuit of reliability, a measurement-prediction-decision approach can be considered to enhance the accuracy of Modulation and Coding Scheme (MCS) decisions during LA, specifically by forecasting interference. In this paper, a two-stage uplink interference prediction algorithm is proposed. In the first stage, complex uplink interference values are decomposed to extract inherent patterns. In the second stage, leveraging the prior knowledge provided by the first stage, which enhances the algorithm's robustness and accuracy, inference is made. The experimental results demonstrate that the proposed interference prediction algorithm not only exhibits a significant improvement in accuracy but also contributes to a substantial enhancement in the performance of the communication system.

Currently, there is a growing trend in deploying ubiquitous artificial intelligence (AI) applications at the network edge. As a promising framework that enables secure edge intelligence, federated learning (FL) has been paid attention, where the over-the-air computing technique has been adopted to enhance the communication efficiency. In this study, we focus on over-the-air FL over a large-scale network with numerous edge devices. Joint device selection and aggregate beamforming design is investigated under two different objectives, i.e., minimizing the aggregate error and maximizing the number of selected devices. Two combinatorial problems are formulated, which are demanding to solve especially in the large-scale network. To reduce the computational complexity, a random aggregate beamforming scheme is proposed, which employs random sampling instead of optimization to determine the aggregator beamforming vector. Notably, the implementation of the proposed scheme does not necessitate the full channel estimation. Asymptotic analysis reveals that the aggregate error asymptotically follows a Gaussian distribution, and the number of selected devices approximates a symmetrical distribution. The distribution parameters are explicitly expressed by the transmit power, the numbers of devices and selected devices. Simulation results confirm the theoretical analysis and demonstrate the effectiveness of the proposed random aggregate beamforming scheme.

Beam selection for joint transmission in cell-free massive multi-input multi-output systems faces the problem of extremely high training overhead and computational complexity. The traffic-aware quality of service additionally complicates the beam selection problem. To address this issue, we propose a traffic-aware hierarchical beam selection scheme performed in a dual timescale. In the long-timescale, the central processing unit collects wide beam responses from base stations (BSs) to predict the power profile in the narrow beam space with a convolutional neural network, based on which the cascaded multiple-BS beam space is carefully pruned. In the short-timescale, we introduce a centralized reinforcement learning (RL) algorithm to maximize the satisfaction rate of delay w.r.t. beam selection within multiple consecutive time slots. Moreover, we put forward three scalable distributed algorithms including hierarchical distributed Lyapunov optimization, fully distributed RL, and centralized training with decentralized execution of RL to achieve better scalability and better tradeoff between the performance and the execution signal overhead. Numerical results demonstrate that the proposed schemes significantly reduce both model training cost and beam training overhead and are easier to meet the user-specific delay requirement, compared to existing methods.

In frequency division duplexing (FDD) cell-free massive MIMO, the acquisition of the channel state information (CSI) is very challenging because of the large overhead required for the training and feedback of the downlink channels of multiple cooperating base stations (BSs). In this paper, for systems with partial uplink-downlink channel reciprocity, and a general spatial domain channel model with variations in the average port power and correlation among port coefficients, we propose a joint-port-selection-based CSI acquisition and feedback scheme for the downlink transmission with zero-forcing precoding. The scheme uses an eigenvalue-decomposition-based transformation to reduce the feedback overhead by exploring the port correlation. We derive the sum-rate of the system for any port selection. Based on the sum-rate result, we propose a low-complexity greedy-search-based joint port selection (GS-JPS) algorithm. Moreover, to adapt to fast time-varying scenarios, a supervised deep learning-enhanced joint port selection (DL-JPS) algorithm is proposed. Simulations verify the effectiveness of our proposed schemes and their advantage over existing port-selection channel acquisition schemes.

Network slicing-based communication systems can dynamically and efficiently allocate resources for diversified services. However, due to the limitation of the network interface on channel access and the complexity of the resource allocation, it is challenging to achieve an acceptable solution in the practical system without precise prior knowledge of the dynamics probability model of the service requests. Existing work attempts to solve this problem using deep reinforcement learning (DRL). However, such methods usually require a lot of interaction with the real environment to achieve good results. In this paper, a framework consisting of a digital twin and reinforcement learning agents is present to handle the issue. Specifically, we propose to use the historical data and the neural networks to build a digital twin model to simulate the state variation law of the real environment. Then, we use the data generated by the network slicing environment to calibrate the digital twin so that it is in sync with the real environment. Finally, DRL for slice optimization optimizes its performance in this virtual pre-verification environment. We conducted an exhaustive verification of the proposed digital twin framework to confirm its scalability. Specifically, we propose to use loss landscapes to visualize the generalization of DRL solutions. We explore a distillation-based optimization scheme for lightweight slicing strategies. In addition, we also extend the framework to offline reinforcement learning, where solutions can be used to obtain intelligent decisions based solely on historical data. Numerical simulation experiments show that the proposed digital twin can significantly improve the performance of the slice optimization strategy.

Network slicing (NS) can enable diverse communication services for vertical industries. In radio access network slicing, differential service level agreement (SLA) guaranteeing is an essential resource management task. Benefit from powerful data analysis capabilities, deep learning (DL) is suitable for intelligent resource management under the cases of complex constraints and time-varying states. Thus, DL has been used to manage resources for NS recently. However, the training of these DL-assisted methods is time-consuming and it is difficult to keep high SLA satisfaction rates dynamically. To address this problem, we propose a hierarchical intelligent NS resource configuration method via organically integrating NS preconfiguration models based on deep neural networks and NS reconfiguration models using multiarm bandits. A factory automation system is established to evaluate our proposed methods on different industrial services. Simulation and experimental results demonstrate that our proposed methods outperform benchmarks comprehensively.

Interleaved training has been studied for single-user and multi-user massive MIMO downlink with either fully-digital or hybrid beamforming. However, the impact of channel correlation on its average training overhead is rarely addressed. In this paper, we explore the channel correlation to improve the interleaved training for single-user massive MIMO downlink. For the beam-domain interleaved training, we propose a modified scheme by optimizing the beam training codebook. The basic antenna-domain interleaved training is also improved by dynamically adjusting the training order of the base station (BS) antennas during the training process based on the values of the already trained channels. Exact and simplified approximate expressions of the average training length are derived in closed-form for the basic and modified beam-domain schemes and the basic antenna-domain scheme in correlated channels. For the modified antenna-domain scheme, a deep neural network (DNN)-based approximation is provided for fast performance evaluation. Analytical results and simulations verify the accuracy of our derived training length expressions and explicitly reveal the impact of system parameters on the average training length. In addition, the modified beam/antenna-domain schemes are shown to have a shorter average training length compared to the basic schemes.

RAN slicing is a key technology in 5G communications for realizing differentiated service level agreement (SLA) guarantee. In this paper, we investigate the multi-slice multi-user RAN slicing. A dynamic bandwidth allocation scheme for RAN slicing is proposed based on hierarchical intelligence, where bandwidth pre-allocation and hyper-parameter tuning for MAC-layer schedulers are jointly optimized to maximize the system utility, i.e., the weighted sum of spectrum efficiency (SE) and SLA satisfaction ratio (SSR) of different slices. The problem is formulated as a twin-time scale Markov decision process (MDP), where the bandwidth pre-allocation and the scheduler parameter tuning are performed on a long-term scale (e.g., seconds) and a short-term scale (e.g., 100 ms), respectively, for which we propose a hierarchical twin-time scale Dueling deep Q learning network (TTS-DDQN) algorithm. And a new reward-clipping mechanism is proposed to get a better trade-off between stabilized training and higher system utility. In order to improve the algorithm robustness to time-varying traffic pattern, we further propose a traffic-aware module for more efficient sampling of the experience pool. Extensive simulations show that the traffic-aware TTS-DDQN outperforms existing typical DQN based algorithms, hard slicing and non slicing, etc.

With the widespread use of massive multi-input-multi-output (MIMO) technology in current wireless networks, network optimization faces much higher costs due to the significantly increased angular space. Digital twin (DT), as a promising tool to enhance the effectiveness and efficiency of performance evaluation, still faces many challenges for massive MIMO optimization, where the complex channel characteristics and the system performance uncertainty over randomly distributed user equipment (UE) position both make it difficult to obtain an explicit relationship expression between the beamforming parameters at the base station (BS) and the system performance. In this article, we propose a conditional generative adversarial network (C-GAN) based digital twin network (DTN), which can fit the mapping from the beamforming to the system performance and match the distribution of system performance under a certain beamforming configuration over different UE position simultaneously. Moreover, it provides a generalized way for pre-validation of different key performance indicators (KPIs) and further raises the accuracy via data augmentation. QuaDRiGa based simulations validate the effectiveness of our proposed method in system performance modeling and KPI prediction.

Millimeter wave (mmWave) cell-free MIMO achieves an extremely high rate while its beam alignment (BA) suffers from excessive overhead due to a large number of transceivers. Recently, user location and probing measurements are utilized for BA based on machine learning (ML) models, e.g., deep neural network (DNN). However, most of these ML models are centralized with high communication and computational overhead and give no specific consideration to practical issues, e.g., limited training data and real-time model updates. In this paper, we study the probing beam-based BA for mmWave cell-free MIMO downlink with the help of broad learning (BL). For channels without and with uplink-downlink reciprocity, we propose the user-side and base station (BS)-side BL-aided incremental collaborative BA approaches. Via transforming the centralized BL into a distributed learning with data and feature splitting respectively, the user-side and BS-side schemes realize implicit sharing of multiple user data and multiple BS features. Simulations confirm that the user-side scheme is applicable to fast time-varying and/or non-stationary channels, while the BS-side scheme is suitable for systems with low-bandwidth fronthaul links and a central unit with limited computing power. The advantages of proposed schemes are also demonstrated compared to traditional and DNN-aided BA schemes.

This paper studies the performance of multi-antenna proactive eavesdropping in 5G uplink systems with spatially correlated Rayleigh fadings, where the base station (BS) serves an illegal user and a multi-antenna legitimate monitor eavesdrops the suspicious link with the help of jamming attack. Based on a practical assumption of channel state information (CSI) in 5G uplink, i.e., imperfect instantaneous CSI of the suspicious link at the BS and that of eavesdropping link at the monitor, and only jamming channel statistics at the monitor, we first give a statistical jamming beamforming design. Then, semi-closed form expressions of eavesdropping non-outage probability and relative average eavesdropping rate are, respectively, derived for delay-sensitive and delay-tolerant scenarios. Via reasonable approximations and asymptotic analysis, we gain many insights on the effect of key system parameters, e.g., location-dependent channel path loss and angular spread, jamming energy and the number of antennas. Further, we provide the optimal energy allocation between pilot and data jamming and the optimal antenna allocation between jamming and eavesdropping under total energy constraint and total antenna number, respectively, which are both explicit functions of system parameters. Finally, simulation results validate our analytical results.

Fifth-generation (5G) wireless communications confront explosive growth in mobile data demand and massive intensive user equipment (UE) connections. The optimization of base station (BS) array orientations in large-scale networks can provide high potential performance gain to meet these requirements. However, traditional schemes are highly dependent on experience and difficult in implementation due to their demand on repeated drive tests and large amounts of data samples. In this paper, via exploiting UE locations and channel direction information, we propose an intelligent network optimization framework to maximize the long-term network rate performance. For the augmentation of limited drive test data, a deep Gaussian process regression (DGPR) model is designed to construct a scenario-specific channel modeling. In addition, via a domain-knowledge driven fusion of convolutional neural network (CNN) and multi-layer perceptron (MLP), we propose a multi-branch deep neural network (DNN) to accurately map BS array orientations and concise channel measurements to the network performance. Finally, based on the scenario-specific modeling, an efficient gradient search approach is proposed to optimize BS array orientations via neural network (NN) backpropagation. Both simulations and field tests in 5G experimental networks validate the effectiveness and efficiency of our proposed framework, especially with limited drive test data.

Mobile-Edge Computing (MEC) is an emerging paradigm in the Internet of Vehicles (IoV) to meet the ever-increasing computation demands of smart applications. To provide satisfactory computation performance, it is of significant importance to conduct computation offloading in IoV. In this paper, we investigate a multi-vehicle IoV system assisted by MECs with limited computation resources, where vehicles with complex applications can offload their subtasks to MEC servers. Applications are modeled as interdependent subtasks with general random task graphs, different from existing works with independent ones. To maximize the average logarithmic data processing rate (LDPR), the computation offloading problem is formulated as a time-average optimization with long-term constraints, which results from variable vehicle number, various applications and time-varying communication channels. To reduce the cooperation overhead, we propose a multi-agent Proximal Policy Optimization algorithm (Ly-MAPPO) which requires local observation only to solve the subproblems achieved by Lyapunov optimization technique in real time. In addition, to improve the performance of the Ly-MAPPO algorithm, Graph Convolutional Neural Network (GCN) is introduced to extract inter-dependencies between subtasks. Extensive simulations show that the GCN embedded Ly-MAPPO outperforms other baseline algorithms, e.g., greedy algorithm and gene algorithm, etc., for different traffic loads and computation resources in MEC servers.

Hybrid heterogeneous cellular networks (HCNets) with sub-6G and millimeter-wave (mmWave) base stations (BSs) achieves good performance-and-cost tradeoffs. Via noticing the negligible retrieving latency and the potential joint transmission (JT) gain of sub-6G BSs, we propose a novel cache-enabled user association scheme for hybrid HCNets with limited local storage, in which partial sub-6G local storage caches contents with relatively low popularity for higher overall content diversity. To sufficiently exploit the large bandwidth of mmWave BSs, they are endowed with higher association priority. Then, the success probability and meta distribution of achievable rate are studied for the proposed scheme, respectively. We further provide the optimization of caching parameters and the condition for superior performance over the MPC-based scheme. Analytical and numerical results show that the network performance can be improved for larger sub-6G BS cooperation size and the array size at mmWave BSs, while the mmWave BS density has two adverse effect for low-to-medium and high rate threshold. For the proposed scheme, the content diversity in combination with JT gain benefits the success probability. And its performance advantage increases for larger local storage and/or higher association probability with mmWave BSs.

In this paper, the random iterative method is introduced to massive multiple-input multiple-output (MIMO) systems for the efficient downlink linear precoding. By adopting the random sampling into the traditional iterative methods, the matrix inversion within the linear precoding schemes can be approximated statistically, which not only achieves a faster exponential convergence with low complexity but also experiences a global convergence without suffering from the various convergence requirements. Specifically, based on the random iterative method, the randomized iterative precoding algorithm (RIPA) is firstly proposed and we show its approximation error decays exponentially and globally along with the number of iterations. Then, with respect to the derived convergence rate, the concept of conditional sampling is introduced, so that further optimization and enhancement are carried out to improve both the convergence and the efficiency of the randomized iterations. After that, based on the equivalent iteration transformation, the modified randomized iterative precoding algorithm (MRIPA) is presented, which achieves a better precoding performance with low-complexity for various scenarios of massive MIMO. Finally, simulation results based on downlink precoding in massive MIMO systems are given to show the system gains of RIPA and MRIPA in terms of performance and complexity.

This paper focuses on the outage performance of a downlink integrated satellite-terrestrial relay network (ISTRN), which includes a satellite (S) transmitting messages to a user (D) with the help of K relays ({Rk, k =1,⋯,K}) using a threshold-based decode-and-forward (DF) transmission protocol. In addition, we consider the case of relay selection based on out-dated channel state information (CSI) to better reflect the actual scenario. The shadowed-Rician (SR) fading and Nakagami-m fading are considered for the channel model of the S-Rklinks and Rk-D links, respectively. In order to reveal the impact of the parameters on the performance of the considered network, the outage probability is adopted as a performance criterion and the corresponding exact closed-form expressions are derived. Further, in order to investigate the influence of system parameters more intuitively, asymptotic outage probabilities are considered for the two cases: 1) high average SNR for S-Rklink and Rk-D link; 2) high average SNR for Rk-D link. The asymptotic analysis results exactly reveal the influence of the relevant system parameters on the decoding gain and the diversity gain. Finally, simulation and numerical results verify the correctness of the analysis.

Massive multiple-input-multiple-output (MIMO) system inevitably faces the hardware cost and energy efficiency problem due to its large number of antennas at the base station (BS). The use of low-resolution analog-to-digital converters (ADCs), e.g., typical 1-bit ADCs, can effectively reduce the system cost. In this paper, we consider a massive MIMO uplink with ternary/three level ADCs. The design of typical linear combiner based detectors is given along with their analytical symbol-error-rate (SER) performance results. Analytical and simulation results show that 1) ternary ADCs can effectively compensate the SER performance gap between 1-bit and full-resolution ADCs; 2) optimal design of ternary ADCs for SER minimization can be referred to the existing design for quantization error minimization; 3) ternary ADCs perform better than 2-bit ADCs in energy efficiency. Thus, for some low-cost scenarios where implementing 2-bit ADCs for each antenna in massive MIMO may be even unaffordable, ternary ADC can be a good choice.

In this letter, for outage performance orientated large intelligent surfaces (LISs)-assisted point to point wireless systems with severely blocked direct link and Rayleigh fading channels,we first propose a jointly interleaved training and transmission design. Then a semi-closed form expression is derived for the average training overhead. And it is shown to be upper bounded by the minimum between the LIS size and a value explicitly dependent on the target receiver signal-to-noise-ratio (SNR). The upper bound gives the condition on the target SNR for achieving overhead saving compared to the full CSI scheme. And the overhead saving increases linearly with the LIS size for constant target SNR. Non-negligible overhead saving is still available even though one increases the target SNR with larger LIS, e.g., as the square of the LIS size for fully exploiting the beamforming gain. Finally, we indicate the impact of practical phase quantization on the training and feedback overhead. Simulations verify these results and show that the proposed scheme can significantly reduce the training overhead without performance loss compared to the full CSI scheme.

In this paper, we study the beam-based training design for the single-user (SU) hybrid massive antenna system based on outage probability performance. First, an interleaved training design is proposed where the feedback is concatenated with the training procedure to monitor the training status and to have the training length adaptive to the channel realization. Then, the average training length and outage probability are derived for the proposed interleaved training and SU transmission. Analytical results and simulations show that the proposed interleaved scheme achieves the same outage performance as the traditional full-training scheme but with significant saving in the training overhead.

Zero-forcing (ZF) precoding plays an important role for massive MIMO downlink due to its near optimal performance in high signal-to- noise (SNR) region. However, the high computation cost of the involved matrix inversion hinders its application in practical large-scale systems. In this paper, we adopt the first order Neumann series (NS) expansion for a low-complexity approximation of matrix inversion. Compared to existing NS based schemes, we introduce a relaxation parameter jointly with one user's channel interference to others into the precondition matrix and propose the identity-plus- column NS (ICNS) method. By further exploiting the multi-user diversity gain via choosing the user with largest interference to others, the ordered ICNS method is also proposed. Moreover, the closed-form sum-rate approximation of the ICNS method is derived. Simulations verify our analytical results and the advantage of the proposed schemes over other existing low-complexity ZF precodings for massive MIMO systems with correlated channels and not-so-small loading factor.

Zero-forcing (ZF) precoding plays an important role for massive MIMO downlink due to its near optimal performance. However, the high computation cost of the involved matrix inversion hinders its application. In this paper, we adopt the first order Neumann series (NS) for a low-complexity approximation. By introducing a relaxation parameter jointly with the channel non-orthogonality between one selected user and others into the precondition matrix, we propose the identity-plus-column NS (ICNS) method. By further choosing the user with the least channel orthogonality with others, the ordered ICNS method is also proposed. Moreover, the sum-rate approximations of the proposed ICNS method and the competitive existing identity matrix based NS (INS) method are derived in closed-form, based on which the performance loss of ICNS due to inversion approximation compared with ideal ZF and its performance gain over INS are explicitly analyzed for three typical massive MIMO scenarios. Finally, simulations verify our analytical results and also show that the proposed two designs achieve better performance-complexity tradeoff than ideal ZF and existing low-complexity ZF precodings for practical large antenna number, correlated channels, and not-so-small loading factor.

Large-scale antenna arrays which provide high beamforming gain are commonly used to combat the serious path-loss in millimeter wave (mmwave) systems. Traditionally, the beamforming is completely implemented at the baseband or the digital domain, which, however, causes high hardware cost and power consumption. The hybrid precoding method which can effectively avoids these problems is, therefore, more attractive. However, the design of hybrid precoder is challenging due to the constant modules of phase shifters. To solve this problem, we propose a novel hybrid precoding approach from the perspective of geometric construction in this paper. The new method can significantly reduce the number of RF chains meanwhile still achieve an almost optimal performance in terms of sum rate. The proposed algorithm is compared with the popular orthogonal matching pursuit algorithm (OMP) via numerical simulations, and shows that our proposal can increase the system spectral efficiency with reduced computational complexity.

This paper studies beam training design for a codebook- based beamforming millimeter wave (mmwave) system where multiple antenna arrays are employed and each array is capable of beamforming independently. To reduce the training overhead and the complexity of subsequent beam direction search, we propose a cooperative multi- subarray beam training method. Specifically, from the perspective of excluding noneffective beam direction combinations and thus reducing search space, method and criterion of beam superposition are proposed to construct a wide beam from multiple narrow beams corresponding to multiple subarrays. Then, a cooperative multisubarray beam training scheme is proposed based on the proposed criterion. Finally, simulation results show that the proposed scheme achieves a spectral efficiency close to that of the optimal exhaustive search scheme, while has greatly reduced training overhead and computational complexity.

In massive multiple‐input multiple‐output (MIMO) high‐speed train (HST) wireless communication systems (WCS), due to the large‐dimension and rapidly time‐varying property, channel tracking rather than conventional channel estimation is more feasible. Due to the non‐linear channel variation, the channel is first decomposed into sub‐channels with different angle of departure. By exploiting the property of near line of sight (LoS) and location information, Kalman filtering is used to track the channel by predicting and modifying the channel gain of the LoS sub‐channel. To further improve the tracking performance, a location‐based sequentially optimal beam pattern design is proposed. Since placing pilot beams for many receive antennas with different locations in time or frequency division manner costs much system resource, the authors further propose a low‐complexity grouping algorithm based on location. Finally, two transmission designs based on the proposed channel tracking scheme and the location only, respectively, are introduced. Simulations validate the efficiency of the authors' proposed tracking scheme and the corresponding transmission design, which shows the importance of location information and channel tracking in massive MIMO HST‐WCS.

Statistical beamforming is an important technique for multi-user massive MIMO downlink, since it depends on the downlink channel covariance only. In this paper, we first derive an explicit analytical sum-rate expression for generic channel covariance-based beamforming scheme. Then, a low-complexity joint statistical beamforming and user scheduling algorithm via greedy search is proposed, where the beamforming is based on the signal-to-leakage-and-noise-ratio (SLNR) for closed-form design and tractable analysis, while the user scheduling is based on the derived sum-rate expression. Further, with the help of large-scale asymptotic simplifications and the introduction of the interference user number parameter, a simple analytical sum-rate expression of the joint algorithm is derived for channels with flat power beam spectrum. The expression explicitly exhibits the sum-rate behavior with respect to different network parameters and captures the effect of sum-rate-based user scheduling. Finally, simulation results are provided to verify our analytical results and to show the advantage of the proposed joint design compared with existing schemes.

To fully exploit the spatial multiplexing gains and array gains of massive multiple-input-multiple-output (MIMO), the channel state information must be obtained accurately at the transmitter side (CSIT). However, conventional channel estimation solutions are not suitable for Frequency-Division Duplexing (FDD) multi-user massive MIMO systems, due to overwhelming pilot and feedback overhead. In this paper, We find that part of the user channels tend to exhibit an approximate beam-blocked sparsity. To exploit this property, we propose a novel blocked compressive channel estimation scheme based on user grouping to reduce the pilot and feedback overhead. More specifically, we adopt user grouping by making the users in one group have similar channel covariance, which makes the channels in one group exhibit beam block sparsity. Then users feed the compressed measurements back to BS and the BS performs the CSIT recovery. Using the beam block sparsity, an optimal block orthogonal matching pursuit algorithm (OBOMP) is developed which effectively recovers the channel parameters. Numerous simulation results demonstrate our proposed scheme outperforms conventional solutions.