Dual-Function Radar and Communication (DFRC) system with full-duplex (FD) sensing and half-duplex (HD) communication is regarded as a promising way to improve safety and efficiency in vehicular networks, which have been widely used in vehicular networks. To achieve higher communication capacity, this paper extends the beamforming design of full-duplex sensing and communication DFRC systems to general scenarios within vehicular networks. Compared to the fully-connected hybrid beamforming architecture, the partially-connected hybrid beamforming architecture holds great potential of reducing hardware costs, thereby facilitating practical implementation in real-world wireless systems. However, the unique block-diagonal structure of its analog beamforming matrix introduces additional design challenges. This paper investigates an efficient partially-connected hybrid beamforming algorithm for FD-DFRC multi-input multi-output (MIMO) systems. Specifically, we formulate a multi-objective optimization problem to jointly optimize the dual-functional transmit beamformers for information receiving vehicles (IRVs) and the base station (BS), while also optimizing the received beamformers for information transmission vehicles (ITVs) at the full-duplex vehicle terminal (FD-VT). In particular, we decompose the original problem of maximizing the sum rate into several subproblems. We design an efficient algorithm based on lower bound maximization, Riemannian manifold optimization methods, and the alternating direction multiplier method (ADMM) algorithm. Numerical simulations demonstrate that the proposed scheme effectively optimizes the sum rate while maintaining radar performance thresholds, requiring fewer radio frequency chains, and adhering to energy offloading constraints.

The rapid growth of IoT applications has driven the demand for high-accuracy indoor localization. However, existing mainstream solutions have distinct limitations: while sensors such as vision, inertial, and LiDAR provide high-precision relative motion tracking, their inherent cumulative drift compromises long-term robustness; meanwhile, high-precision technologies like UWB rely on expensive, dedicated infrastructure deployment. The emergence of 5G networks offers a unique solution by utilizing its existing communication base stations as positioning anchors. Leveraging its wideband signals, 5G has the potential to provide the absolute position references needed to correct relative errors without significant additional hardware costs, thereby overcoming the limitations of the aforementioned techniques. However, 5G-based positioning technologies face challenges, such as multipath interference, clock synchronization errors, and degraded performance during terminal mobility. To address these limitations, we propose a novel, tightly-coupled 5G Positioning Reference Signal (PRS) and Inertial Measurement Unit (IMU) fusion framework. The core of our approach is a two-stage architecture. First, a front-end signal error estimator models the joint estimation of clock offset and integer ambiguity as a Mixed-Integer Nonlinear Programming (MINLP) problem. Crucially, it leverages IMU-predicted kinematics as a strong physical constraint to effectively identify and suppress measurements contaminated by multipath. Subsequently, a back-end, tightly-coupled Error-State Kalman Filter (ESKF) fuses the Time-Differenced Carrier Phase (TDCP) and Time Difference of Arrival (TDOA) measurements from the front-end with inertial navigation data. Simulation results conducted under 3GPP standard scenarios demonstrate that the proposed framework achieves a positioning accuracy improvement of approximately 40% in centimeter-level accuracy, in line-of-sight (LOS) environments with tactical IMU equipment. This provides a promising solution for 5G-based indoor localization systems. In real-world indoor field tests characterized by multipath effects, our proposed EE-ESKF algorithm achieved a three-dimensional root mean square error (RMSE) of 0.55 meters, which is 33.6% higher than the standard ESKF algorithm.

Zero-Touch Networks (ZTNs) represent a transformative paradigm toward fully automated and intelligent network management, providing the scalability and adaptability required for the complexity of sixth-generation (6G) networks. However, the distributed architecture, high openness, and deep heterogeneity of 6G networks expand the attack surface and pose unprecedented security challenges. To address this, security automation aims to enable intelligent security management across dynamic and complex environments, serving as a key capability for securing 6G ZTNs. Despite its promise, implementing security automation in 6G ZTNs presents two primary challenges: 1) automating the lifecycle from security strategy generation to validation and update under real-world, parallel, and adversarial conditions, and 2) adapting security strategies to evolving threats and dynamic environments. This motivates us to propose SecLoop and SA-GRPO. SecLoop constitutes the first fully automated framework that integrates large language models (LLMs) across the entire lifecycle of security strategy generation, orchestration, response, and feedback, enabling intelligent and adaptive defenses in dynamic network environments, thus tackling the first challenge. Furthermore, we propose SA-GRPO, a novel security-aware group relative policy optimization algorithm that iteratively refines security strategies by contrasting group feedback collected from parallel SecLoop executions, thereby addressing the second challenge. Extensive real-world experiments on five benchmarks, including 11 MITRE ATT&CK processes and over 20 types of attacks, demonstrate the superiority of the proposed SecLoop and SA-GRPO. We will release our platform to the community, facilitating the advancement of security automation towards next generation communications.

Recent AI agents, such as ChatGPT and LLaMA, primarily rely on instruction tuning and reinforcement learning to calibrate the output of large language models (LLMs) with human intentions, ensuring the outputs are harmless and helpful. Existing methods heavily depend on the manual annotation of high-quality positive samples, while contending with issues such as noisy labels and minimal distinctions between preferred and dispreferred response data. However, readily available toxic samples with clear safety distinctions are often filtered out, removing valuable negative references that could aid LLMs in safety alignment. In response, we propose Positive–Toxic Self-Alignment (PT-ALIGN), a novel safety self-alignment approach that minimizes human supervision by automatically refining positive and toxic samples and performing fine-grained dual instruction tuning. Positive samples are harmless responses, while toxic samples deliberately contain extremely harmful content, serving as a new supervisory signal. Specifically, we utilize LLM itself to iteratively generate and refine training instances by only exploring fewer than 50 human annotations. We then employ two losses, i.e., maximum likelihood estimation (MLE) and fine-grained unlikelihood training (UT), to jointly learn to enhance the LLM’s safety. The MLE loss encourages an LLM to maximize the generation of harmless content based on positive samples. Conversely, the fine-grained UT loss guides the LLM to minimize the output of harmful words based on toxic samples at the token-level, thereby guiding the model to decouple safety from effectiveness, directing it toward safer fine-tuning objectives, and increasing the likelihood of generating helpful and reliable content. Experiments on 9 popular open-source LLMs demonstrate the effectiveness of our PT-ALIGN for safety alignment, while maintaining comparable levels of helpfulness and usefulness.

Energy efficiency is essential for federated learning (FL) over mobile devices and its potential prosperous applications. Different from existing communication efficient FL research efforts, which regard communication energy consumption as the bottleneck, we have observed that with ever increasing wireless transmission speed (e.g., Wi-Fi 5 or 5G), the energy consumption of wireless communications for model updates in FL is significantly reduced and sometimes is smaller than that of local on-device training. Motivated by such observations, in this paper, we propose a high-speed wireless communications inspired energy efficient federated learning over mobile devices (EEFL), whose goal is to reduce the overall energy consumption (computing + communication). In particular, we design a novel energy-aware adaptive local update policy for mobile devices by jointly considering FL performance and energy saving of high-speed wireless transmissions. Furthermore, given the device’s local update policy in each FL global round, we advance the dynamic voltage and frequency scaling (DVFS) strategy to minimize local training’s energy consumption by keeping GPU and CPU working at appropriate frequencies without triggering thermal throttling. Extensive experimental results with various learning models, datasets, and wireless transmission environments demonstrate the proposed EEFL’s superiority over the peer designs in terms of energy efficiency.

As a popular distributed learning paradigm, federated learning (FL) over mobile devices fosters numerous applications, while their practical deployment is hindered by participating devices’ computing and communication heterogeneity. Some pioneering research efforts proposed to extract subnetworks from the global model, and assign as large a subnetwork as possible to the device for local training based on its full computing and communications capacity. Although such fixed size subnetwork assignment enables FL training over heterogeneous mobile devices, it is unaware of (i) the dynamic changes of devices’ communication and computing conditions and (ii) FL training progress and its dynamic requirements of local training contributions, both of which may cause very long FL training delay. Motivated by those dynamics, in this paper, we develop a wireless and heterogeneity aware latency efficient FL (WHALE-FL) approach to accelerate FL training through adaptive subnetwork scheduling. Instead of sticking to the fixed size subnetwork, WHALE-FL introduces a novel subnetwork selection utility function to capture device and FL training dynamics, and guides the mobile device to adaptively select the subnetwork size for local training based on (a) its computing and communication capacity, (b) its dynamic computing and/or communication conditions, and (c) FL training status and its corresponding requirements for local training contributions. We provide a theoretical convergence analysis for WHALE-FL with heterogeneous subnetwork assignment, based on which subnetwork structures can be dynamically optimized to reduce the resulting gap to standard full-model FL. Our evaluation shows that, compared with peer designs, WHALE-FL effectively accelerates FL training without sacrificing learning accuracy.

Terahertz (THz) communication technology is regarded as a promising enabler for achieving ultra-high data rate transmission in next-generation communication systems. To mitigate the high path loss in THz systems, the transmitting beams are typically narrow and highly directional, which makes it difficult for a single beam to serve multiple users simultaneously. To address this challenge, reconfigurable intelligent surfaces (RIS), which can dynamically steer the incident electromagnetic waves, have been integrated into THz communication systems to extend coverage. Existing works mostly remain theoretical analysis and simulation, while prototype validation of RIS-assisted THz systems is scarce. In this paper, we designed a kind of liquid crystal-based RIS operating at 220 GHz supporting both single-user and multi-user communication scenarios, followed by a 220 GHz THz-RIS communication system prototype. To support simultaneous multi-user transmission, we designed an OFDM-based resource allocation scheme. Beamforming optimization is performed in both single and muti-user scenarios. Specifically, to enhance the system performance of multi-user scenario, we developed a received power based feedback control method. In our experiments, the received power gain with RIS is no less than 10 dB in the single-beam mode, and no less than 5 dB in the multi-beam mode, compared to the case without RIS. With the assistance of RIS, the achievable rate of the system could reach 2.341 Gbps with 3 users sharing 2 GHz bandwidth and the bit error rate (BER) of the system decreased sharply. Finally, an image transmission experiment was conducted to vividly show that the receiver could recover the transmitted information correctly with the help of RIS. The experimental results also demonstrated that the received signal quality was enhanced through power feedback adjustments.

Simultaneous wireless information and power transfer (SWIPT) has been regarded as a highly promising technology for delivering reliable energy supply for low-powered Internet-of-Thing devices (IoTDs). In this paper, we originally investigate the performance of SWIPT systems in multi-Reconfigurable Intelligent Surface (RIS) aided cell-free IoT networks with fluid antenna (FA).Our objective is to simultaneously maximize the sum transmission rate and harvested energy by optimizing the selection of FA ports, transmit beamforming vectors at access points (APs), and reflect beamforming at RISs, which is a notorious trade-off and is impossible to simultaneously maximize at the same IoTD under the traditional Power splitting (PS) or Time switching (TS) based SWIPT techniques. By converting the harvested energy into throughput, we formulate a novel equivalent-sum-rate (ESR) maximization problem. To handle this sum-of-logarithmic function and nested multiple-ratio fractional nonconvex optimization problem, we develop a Lagrangian dual transform (LDT) and quadratic transform (QT)-based alternating optimization (AO) framework to tackle it. Simulation results demonstrate that the proposed method outperforms benchmarks, highlighting the enhanced system performance in both information and energy reception facilitated by RISs and FA. Moreover, our research also indicates that the deployment of FA at IoTDs can achieve better performance than the traditional fixed-position antennas in such a network, as long as it can provide an adequate number of ports and FA size.

This paper advocates a fluid antenna system (FAS)-assisted long-range communication (LoRa-FAS) for Internet-of-Things (IoT) applications. In the proposed system, FAS provides spatial diversity gains for LoRa, eliminating the necessity for integrating multiple-input multiple-output (MIMO) technologies into the system. It consists of a traditional LoRa transmitter with a fixed-position antenna and a LoRa receiver employing the FAS (Rx-FAS). The pilot sequence overhead and placement for FAS are also considered. Specifically, we consider embedding pilot sequences within symbols to reduce the impact of pilot overhead on system throughput and the physical layer (PHY) frame structure, leveraging the fact that the pilot sequences do not convey source information and correlation detection at the LoRa receiver need not be performed across the entire symbol. The achievable performance of LoRa-FAS is thoroughly analyzed under both coherent and non-coherent detection schemes. We obtain new closed-form approximations for the probability density function (PDF) and cumulative distribution function (CDF) of the FAS channel under the block-correlation model. Furthermore, the approximate SER, equivalently the bit error rate (BER), of the proposed LoRa-FAS is also derived in closed form. Simulation results indicate that substantial SER gains can be achieved by FAS within the LoRa framework, even with a limited size of FAS. In addition, our analytical results align well with Clarke’s exact spatial correlation model. Finally, when utilizing the block-correlation model, we suggest that the correlation factor should be selected as the proportion of the eigenvalues of the exact correlation matrix greater than 1 for higher accuracy.

As a popular distributed learning paradigm, federated learning (FL) over mobile devices fosters numerous applications, while their practical deployment is hindered by participating devices' computing and communication heterogeneity. Some pioneering research efforts proposed to extract subnetworks from the global model, and assign as large a subnetwork as possible to the device for local training based on its full computing capacity. Although such fixed size subnetwork assignment enables FL training over heterogeneous mobile devices, it is unaware of (i) the dynamic changes of devices' communication and computing conditions and (ii) FL training progress and its dynamic requirements of local training contributions, both of which may cause very long FL training delay. Motivated by those dynamics, in this paper, we develop a wireless and heterogeneity aware latency efficient FL (WHALE-FL) approach to accelerate FL training through adaptive subnetwork scheduling. Instead of sticking to the fixed size subnetwork, WHALE-FL introduces a novel subnetwork selection utility function to capture device and FL training dynamics, and guides the mobile device to adaptively select the subnetwork size for local training based on (a) its computing and communication capacity, (b) its dynamic computing and/or communication conditions, and (c) FL training status and its corresponding requirements for local training contributions. Our evaluation shows that, compared with peer designs, WHALE-FL effectively accelerates FL training without sacrificing learning accuracy.

Rate-Splitting Multiple Access (RSMA)is a powerful technology to enhance spectral efficiency in short packet Ultra-Reliable and Low-Latency Communication (URLLC) systems with Finite Blocklength (FBL) codes.Enhancing system performance requires larger antenna arrays to meet increasingly stringent QoS requirements, which is very costly. To overcome this limitation, we propose a novel movable antennas (MAs)-assisted RSMA short packet transmission schemes to realize more flexible beamforming and higher spatial multiplexing gains with a small number of MAs. We formulate a sum rate maximization problem for the joint optimization of beamforming and antenna position under the specified URLLC requirements. An alternating optimization (AO) algorithmis developed to address this challenging high-dimensional non-convex optimization problem.Specifically, the Successive Convex Approximation (SCA) methodis utilized to design the beamforming matrix. Additionally, we proposeda novel worst position neighborhood variation-oriented particle swarm optimization (WPNVPSO) to optimize the MA positions efficiently.Numerical results demonstrate our proposed design outperforms other benchmarks by achieving high data rates with lower latency and higher reliability, while utilizing fewer antennas.

Interactive multimodal applications (IMAs), such as route planning in the Internet of Vehicles, enrich users’ personalized experiences by integrating various forms of data over wireless networks. Recent advances in large language models (LLMs) utilize mixture-of-experts (MoE) mechanisms to empower multiple IMAs, with each LLM trained individually for a specific task that presents different business workflows. In contrast to existing approaches that rely on multiple LLMs for IMAs, this paper presents a novel paradigm that accomplishes various IMAs using a single compositional LLM over wireless networks. The two primary challenges include 1) guiding a single LLM to adapt to diverse IMA objectives and 2) ensuring the flexibility and efficiency of the LLM in resource-constrained mobile environments. To tackle the first challenge, we propose ContextLoRA, a novel method that guides an LLM to learn the rich structured context among IMAs by constructing a task dependency graph. We partition the learnable parameter matrix of neural layers for each IMA to facilitate LLM composition. Then, we develop a step-by-step fine-tuning procedure guided by task relations, including training, freezing, and masking phases. This allows the LLM to learn to reason among tasks for better adaptation, capturing the latent dependencies between tasks. For the second challenge, we introduce ContextGear, a scheduling strategy to optimize the training procedure of ContextLoRA, aiming to minimize computational and communication costs through a strategic grouping mechanism. Experiments on three benchmarks show the superiority of the proposed ContextLoRA and ContextGear. Furthermore, we prototype our proposed paradigm on a real-world wireless testbed, demonstrating its practical applicability for various IMAs. We will release our code to the community.

Training latency is critical for the success of numerous intrigued applications ignited by federated learning (FL) over heterogeneous mobile devices. By revolutionarily overlapping local gradient transmission with continuous local computing, FL can remarkably reduce its training latency over homogeneous clients, yet encounter severe model staleness, model drifts, memory cost and straggler issues in heterogeneous environments. To unleash the full potential of overlapping, we propose, FedEx, a novelfederated learning approach toexpedite FL training over mobile devices under data, computing and wireless heterogeneity. FedEx redefines the overlapping procedure with staleness ceilings to constrain memory consumption and make overlapping compatible with participation selection (PS) designs. Then, FedEx characterizes the PS utility function by considering the latency reduced by overlapping, and provides a holistic PS solution to address the straggler issue. FedEx also introduces a simple but effective metric to trigger overlapping, in order to avoid model drifts. Experimental results show that compared with its peer designs, FedEx demonstrates substantial reductions in FL training latency over heterogeneous mobile devices with limited memory cost.

As 6G networks actively advance edge intelligence, Federated Learning (FL) emerges as a key technology that enables data sharing while preserving data privacy and fostering collaboration among edge devices for intelligent service learning. However, the multi-dimensional heterogeneous and hierarchical network architecture brings many challenges to FL deployment, including selecting appropriate nodes for model training and designing effective methods for model aggregation. Compared with most studies that focus on solving individual problems within 6G, this paper proposes an efficient deployment scheme named hierarchical heterogeneous FL (HHFL), which comprehensively considers various influencing factors. First, the deployment of HHFL over 6G is modeled amid the heterogeneity of communications, computation, and data. An optimization problem is then formulated, aiming to minimize deployment costs in terms of latency and energy consumption. Subsequently, to tackle this optimization challenge, we design an intelligent FL deployment framework, consisting of a hierarchical aggregation deployment (HAD) component for hierarchical FL aggregation structure construction and an adaptive node selection (ANS) component for selecting diverse clients based on multi-dimensional discrepancy criteria. Experimental results demonstrate that our proposed framework not only adapts to various application requirements but also outperforms existing technologies by achieving superior learning performance, reduced latency, and lower energy consumption.

Federated Learning (FL) algorithms commonly sample a random subset of clients to address the straggler issue and improve communication efficiency. While recent works have proposed various client sampling methods, they have limitations in joint system and data heterogeneity design, which may not align with practical heterogeneous wireless networks. In this work, we advocate a new independent client sampling strategy to minimize the wall-clock training time of FL, while considering data heterogeneity and system heterogeneity in both communication and computation. We first derive a new convergence bound for non-convex loss functions with independent client sampling and then propose an adaptive bandwidth allocation scheme. Furthermore, we propose an efficient independent client sampling algorithm based on the upper bounds on the convergence rounds and the expected per-round training time, to minimize the wall-clock time of FL, while considering both the data and system heterogeneity. Experimental results under practical wireless network settings with real-world prototype demonstrate that the proposed independent sampling scheme substantially outperforms the current best sampling schemes under various training models and datasets.

The mobile communication network is gradually shifting from the deployment of dedicated physical facilities to the deployment of virtualized network functions on general infrastructure. Taking advantage of this transformation, operators can provide customized services to better meet the demands of vertical industries. However, the participation of vertical industry tenants introduces a new attack surface to the mobile networks. The centralized access control scheme currently employed in the 5G system has a high risk of single-point failure and privacy leakage. And it is inefficient when the scheme is applied in distributed core networks. To secure the mobile communication network and protect the privacy of vertical industry tenants, we propose a distributed policy-hidden (DPH) framework, which achieves flexible cross-trust-domain authorization with lower latency. Besides, Attribute-based Encryption is introduced to protect long-distance communications between network functions. Our evaluation demonstrates that DPH is suitable for virtualized 5G networks and effectively reduces the latency of authorization procedures.

The next-generation wireless communication sys-tems require available bandwidth, reliability, and mobility. How-ever, the wider bandwidth and higher carrier frequency pose challenges for communication systems, including stricter peak-to-average power ratio (PAPR) requirements and more severe Doppler effects. This paper introduces an enhanced discrete Fourier transform spread (ED-OTFS) waveform that utilizes head-tail insertion sequences and frequency domain spectrum shaping (FDSS) methods to mitigate multipath delay variations and reduce high PAPR. Additionally, a reliable channel equalization algorithm for the ED-OTFS receiver is designed to mitigate the impact of the Doppler shift. Simulation results demonstrate that the proposed waveform significantly improves the bit error rate (BER) and PAPR performance compared with the existing schemes.

By exploiting the entropy of wireless channels, physical layer key generation (PLKG) has gained considerable attention in recent years. Due to the characteristic of dynamic controlling of wireless channels, reconfigurable intelligent surface (RIS) can improve the key generation rate (KGR). Simultaneous transmission and reflection reconfigurable intelligent surface (STAR- RIS) can extend the half-space coverage to full-space coverage. This paper proposes a STAR-RIS-assisted physical layer key generation (SRKG) scheme in a multi-user case. The objective is to maximize the sum KGR by jointly designing the transmitting and reflecting coefficients (TARCs). We propose an iterative algorithm based on alternating optimization (AO), utilizing successive convex approximation (SCA) and semi-definite relaxation (SDR) methods, to solve the non-convex optimization problem in the presence of correlated channels. We also propose a low-complexity algorithm based on Lagrange multipliers to jointly optimize TARCs considering independent fading. Finally, the simulation results validate that, compared with the PLKG scheme assisted by traditional RIS (T-RIS), the proposed SRKG scheme can achieve higher sum KGR in a wider range under correlated and independent channel conditions.

Dual-function radar-communication (DFRC) system has been recognized as a potential technology to address the issues of radio frequency spectrum congestion. Despite the advan-tages of the existing orthogonal frequency-division multiplexing (OFDM) chirp waveform, such as its high range resolution and low peak-to-average ratio, it is still plagued by issues related to ghost targets in complex communication environments with multiple targets. In this paper, a novel waveform, leveraging trapezoidal frequency modulation OFDM, is introduced to address the challenge of multi-target detection scenarios. Addition-ally, a power allocation and subcarrier assignment algorithm has been developed to maximize communication performance while adhering to the radar performance threshold, thereby achieving overall system optimization while enhancing multi-target detection capabilities. Finally, the effectiveness of the proposed algorithm is demonstrated through simulation experiments, showcasing its ability to handle multi-target detection scenarios and achieve superior system performance while maintaining a delicate equilibrium between communication and radar considerations.

Reconfigurable intelligent surface (RIS) is highly envisaged as a promising technology in millimeter wave (mmWave) communication systems for its capability of restructuring the wireless communication environment and mitigating the severe signal blockage. However, passive RIS beamforming design is still a challenge due to the non-convex property of the problem. This work presents a deep learning (DL) based codebook classification beam search algorithm, comprising a convolutional neural network (CNN) based codebook searcher and a mapper. The searcher determines the optimal codeword position range index based on the channel state information (CSI), while the mapper selects the optimal codeword based on this index. The proposed network is trained both with the DeepMIMO dataset and the Saleh-Valenzuela channel model, respectively, and imperfect CSI is utilized to enhance the system robustness. Simulation results demonstrate that the proposed algorithm significantly improves the beam search efficiency.

Participant selection (PS) helps to accelerate federated learning (FL) convergence, which is essential for the practical deployment of FL over mobile devices. While most existing PS approaches focus on improving training accuracy and efficiency rather than residual energy of mobile devices, which fundamentally determines whether the selected devices can participate. Meanwhile, the impacts of mobile devices heterogeneous wireless transmission rates on PS and FL training efficiency are largely ignored. Moreover, PS causes the staleness issue. Prior research exploits isolated functions to force long-neglected devices to participate, which is decoupled from original PS designs. In this paper, we propose aresidualenergy andwirelessaware PS design for efficientFLtraining over mobile devices (REWAFL). REWAFL introduces a novel PS utility function that jointly considers global FL training utilities and local energy utility, which integrates energy consumption and residual battery energy of candidate mobile devices. Under the proposed PS utility function framework, REWAFL further presents a residual energy and wireless aware local computing policy. Besides, REWAFL buries the staleness solution into its utility function and local computing policy. The experimental results show that REWAFL is effective in improving training accuracy and efficiency, while avoiding flat battery of mobile devices.

Intelligent reflecting surfaces (IRSs) have recently attained growing popularity in wireless networks owning to their capability to customize the wireless channel via smartly configured passive reflections. In addition to optimizing IRS reflection patterns, the flexible deployment of IRSs offers another design degree of freedom (DoF) to reconfigure the wireless propagation environment in favour of signal transmission. To unveil the impact of IRS deployment on the system capacity, we investigate the capacity of a broadcast channel with a multi-antenna base station (BS) sending independent messages to multiple users, aided by IRSs with N elements. In particular, both the distributed and centralized IRS deployment architectures are considered. Regarding the distributed IRS, the N IRS elements form multiple IRSs and each of them is installed near a user cluster; while for the centralized IRS, all IRS elements are located in the vicinity of the BS. To draw essential insights, we first derive the maximum capacity achieved by the distributed IRS and centralized IRS, respectively, under the assumption of line-of-sight (LoS) propagation and homogeneous channel setups. By carefully capturing the fundamental tradeoff between the spatial multiplexing gain and passive beamforming gain, we rigourously prove that the capacity of the distributed IRS is higher than that of the centralized IRS provided that the total number of IRS elements is above a threshold. Motivated by the superiority of the distributed IRS, we then focus on the transmission and element allocation design under the distributed IRS. By exploiting the user channel correlation of intra-clusters and inter-clusters, an efficient hybrid multiple access scheme relying on both spatial and time domains is proposed to fully exploit both the passive beamforming gain and spatial DoF. Moreover, the IRS element allocation problem is investigated for the objectives of the sum-rate maximization and the minimum user rate maximization, respectively. Finally, extensive numerical results are provided to validate our theoretical finding and also to unveil the effectiveness of the distributed IRS for improving the system capacity under various system setups.

Energy efficiency is essential for federated learning (FL) over mobile devices and its potential prosperous applications. Different from existing communication efficient FL research efforts, which regard communication energy consumption as the bottleneck, we have observed that with ever increasing wireless transmission speed (e.g., Wi-Fi 5 or 5G), the energy consumption of wireless communications for model updates in FL is significantly reduced and sometimes is smaller than that of local on-device training. Motivated by such observations, in this paper, we propose a high-speed wireless communications inspired energy efficient federated learning over mobile devices (EEFL), whose goal is to reduce the overall energy consumption (computing + communication). In particular, we design a novel energy-aware adaptive local update policy for mobile devices by jointly considering FL performance and energy saving of high-speed wireless transmissions. Furthermore, given the device's local update policy in each FL global round, we advance the dynamic voltage and frequency scaling (DVFS) strategy to minimize local training's energy consumption by keeping GPU and CPU working at appropriate frequencies without triggering thermal throttling. Extensive experimental results with various learning models, datasets, and wireless transmission environments demonstrate the proposed EEFL's superiority over the peer designs in terms of energy efficiency.

The Internet of Vehicles (IoV) is a promising paradigm for autonomous driving, where the sensing data from the onboard sensors can be disseminated and processed cooperatively via vehicle-to-vehicle links. Autonomous vehicles can share their local views for cooperative, reliable, and robust driving decisions. However, the limited wireless resources may become the bottleneck with the increasing number of vehicles. The technical challenges also arise from the decentralized control, the spatial couplings of decisions, and the complexity of combinatorial optimization. This paper proposes a novel fully distributed graph-based approach to jointly optimize multicast link establishment with data dissemination and processing decisions by only exchanging partial information among neighboring vehicles. The mixed-integer programming problem aims to maximize system energy efficiency while achieving maximum data throughput. We prove that maximizing data processing throughput is submodular optimization to find the local optimum efficiently. The optimization of data dissemination and processing is reformulated to a minimum-cost maximum-flow problem in a three-layer graph, and efficiently solved by exploiting the graphical interdependence. Both simulation-generated and trace-based datasets are evaluated to validate the effectiveness of the proposed approach in terms of data throughput and energy efficiency.

In recent years, vehicle-to-vehicle (V2V) communication underlaying cellular networks has attracted more and more attention both in academy and industry. In this paper, guaranteeing the latency and reliability of vehicle users (VUEs), we study the resource management problem to maximize the energy efficiency of VUEs while considering the QoS requirement of cellular users (CUEs). Based on the Lyapunov optimization theory, a novel two-layer power control and resource allocation scheme is proposed by exploiting the properties of fractional programming. Simulation results show that the proposed scheme can achieves remarkable improvements in terms of EE and ensure the latency and reliability requirements of V2V communication.

In this paper, spectrum resource allocation in vehicle-to-vehicle (V2V) communications network is studied. Multiple V2V links can share one resource block (RB) for maximizing con-current links with the satisfaction of quality of service (QoS). The resource allocation problem is transformed into an inferential problem on a factor graph model. To solve this problem, the Belief Propagation based on Real-time Update of Messages (BPRUM) algorithm is proposed. In this method, the information matrix is updated after each message passing procedure is completed rather than after once iteration. From the simulation results, the proposed method achieves a significant increment in spectrum utilization compared to the existing algorithms.

Providing reliable and efficient coverage for cell-edge mobile users (MUs) is a key issue in wireless communication networks. With non-uniform structure and heterogeneity of network topology in the 5G/B5G networks, performance improvement of cell-edge MUs becomes even more challenging. Unmanned aerial vehicle (UAV) exhibits a comparable advantage in enhancing cell edge performance due to its flexible mobility and line-of-sight air-to-ground links. In this paper, we study UAV-assisted cell-edge MU offloading in the non-uniform heterogeneous cellular networks. A base station (BS) coordination and ground-to-air offloading scheme is proposed to enhance the cell-edge MUs' performance, whereby cell-edge MUs are periodically scheduled between coordinated ground BSs and a flying UAV. Furthermore, a theoretical framework is developed to analyze the average spectral efficiency (SE) and average network throughput. Specifically, closed-form expressions for the average SE are derived for MUs associated with the ground BSs. Upper and lower bounds for the average SE are also obtained when the MU is offloaded to the flying UAV. Finally, numerical and simulation results are provided to validate the theoretical analysis and investigate the impact of key system parameters on the system performance, which also demonstrate the advantages of the UAV-assisted offloading scheme, compared with benchmark solutions.

Caching on the wireless edge is a promising way to alleviate the backhaul burden of the heterogeneous ultra-dense network (H-UDN) and reduce the transmission delay. However, user mobility makes the content distribution more challenging for the association to small base stations (SBSs) always changes. In this paper, we propose a mobility-aware proactive caching strategy named MoPC in H-UDNs where each SBS can perceive the mobility pattern of users in neighbor SBSs. Moreover, by taking advantage of the hierarchical network feature, files requested by different speed users will be cached in different cache layers by amending content popularity. The cache placement problem is formulated with the aim of effective capacity maximization and solved by a genetic algorithm based approach. Simulation results prove that the proposed MoPC strategy outperforms other existing caching strategies in terms of both the transmission delay and cache hit ratio.

In this paper, we study the secrecy performance in a hybrid wiretapping wireless system, where the half-duplex (HD) or full-duplex (FD) eavesdroppers may wiretap the confidential signal and/or transmit a jamming signal. To evaluate the secrecy performance, we derive the approximate closed-form results for the secrecy outage probability and mean secrecy rate by means of the random matrix theory (RMT). The RMT method can greatly simplify the complicated mathematical analysis with high accuracy, and can provide a useful analytical framework for other researches. The Monte Carlo simulations and numerical results are provided to validate the theoretical analysis and demonstrate the impacts of the system parameters. From the perspective of BS transmission, increasing BS transmission power can greatly improve the secrecy performance in the low transmit power region, but the secrecy performance is constant in the high BS transmit power region. Moreover, in terms of adversaries, FD eavesdroppers have better wiretapping performance than HD eavesdroppers when the jamming power is relatively low; otherwise, this result is reversed.

In this paper, a caching network is established based on OpenAirInterface (OAI) system. A caching server is introduced to locally cache the requested service, which is embedded into the core network. The user behavior as well as the user request profiles are collected and analyzed to dynamically update the cache list. In the demo system, we test the performance of different caching strategies according to different application scenarios using commercial smartphones. The latency performance and user throughput are demonstrated to show the caching influence in the limited storage environment and can be easily integrated in current cellular systems.

In recent years, convolutional neural network (CNN) has made a breakthrough development and been widely used in various fields, such as image recognition, target classification and natural language processing. However, with the continuous development of CNN, the complexity of CNN is gradually increasing. The ordinary hardware processors cannot meet the speed requirements of CNN. The hardware platform about CNN based on FPGA has gradually become the focus of research because of its parallel computing advantages. However, it is difficult and not friendly to implement CNN on FPGA for software developers. In this paper, we propose a FPGA-Cloud architecture for CNN. We implement CNN platform based on FPGA, and deploy the platform in the cloud. CNN resources based on FPGA can be utilized by local users through the network, which is language-friendly for software developers in hardware development and satisfies the user's requirement for CNN processing task.

In 5G ultra dense heterogeneous network, the scheme of Passive Reception Detection (PRD) can exploit more busy bands and enhance the small base station (SBS) access opportunity by identifying the distance between the active macro vehicle (MVE) and the macro base station (MBS). In this paper, with the assistance of the MBS, the MVE distribution area can be obtained by the SBS joint-PRD scheme. And the probability density function of the distance between the MVE and SBS is derived leveraging stochastic geometry. By combining the trajectory prediction of co-channel MVE location with levy flight mobile model, the opportunity of the SBS accessing the busy bands can be derived. It is useful to maximize the number of concurrent vehicle-to-vehicle transmissions. The simulation results show that by taking advantage of the proposed scheme in the mobile network, the overall system throughput can be improved more than 20% regardless of the co-channel MVE velocity.

In this paper, we propose a novel Adaptive Transmission Strategy to improve energy efficiency (EE) in a wireless point-to-point transmission system with Quality of Service (QoS) requirement, i.e., delay-outage probability considered. The transmitter is scheduled to use the channel that has better coefficients, and is forced into silent state when the channel suffers deep fading. The proposed strategy can be easily implemented by applying two-mode circuitry and is suitable for massive machine type communications (MTC) scenarios. In order to enhance EE, we formulate an EE maximization problem, which has a single optimum under a loose QoS constraint. We also show that the maximization problem can be solved efficiently by a binary search algorithm. Simulations demonstrate that our proposed strategy can obtain significant EE gains, hence confirming our theoretical analysis.

Device to Device (D2D) communication has been proved to be an effective way to enhance cellular network capacity, which enables direct data exchange of localized traffic of users in proximity. Current D2D links are mainly based on WiFi Direct technology. In this paper, we propose a 4G LTE-assisted distributed D2D communication network. Information of user devices will be uploaded to a D2D server periodically via commercial 4G LTE network. The D2D server then will initialize a D2D network after collecting the device information. A Token sharing strategy is proposed to control the process of D2D networking, based on the information of SINR, location, battery power, as well as service QoS demand. Finally, our proposed system is demonstrated by Android smartphones.

Recently, device-to-device (D2D) communication has been considered as a promising technology to implement vehicle-to-vehicle (V2V) communication. To ensure road safety and traffic efficiency, V2V has more strict requirements for latency, reliability and access availability. In this paper, we consider the problem of resource block (RB) scheduling for maximizing the number of concurrent V2V transmissions instead of sum rate, allowing multiple vehicles to access to one RB. Firstly, the reliability requirement is transformed into a constraint of a matrix spectral radius to limit the accumulated interference. Secondly, utilizing spectral radius estimation theory, a Minimizing the Increment of Spectral Radius (MISR) resource block sharing algorithm is proposed to accommodate as many users as possible. Finally, simulation results show that the proposed MISR algorithm can improve spectrum efficiency by 150% and 96% in rush hour compared with some other existing methods.