Learning (FL) has emerged as a promising paradigm for enabling artificial intelligence (AI) applications with stringent latency requirements in the Internet of Vehicles (IoV). However, the inherent mobility of vehicles, leading to limited residence time, dynamic wireless transmissions, and a time-varying candidate set, degrades the FL training efficiency. Moreover, the convergence rate of FL model suffers from the heterogeneity issues among vehicles. Due to constrained spectrum resources and high connecting density, client scheduling is essential for FL to mitigate network burdens and enhance learning performance. In this paper, we propose a client scheduling scheme based on the Multi-Armed Bandit (MAB) framework that selects effective and high-quality clients for the FL within IoV. In detail, we consider a FL-assisted IoV scenario and construct a candidate client pool based on a mobility model, representing moving vehicles on a unidirectional lane. In this context, we formulate an optimization problem to minimize the overall FL training latency while ensuring model accuracy under constrained resources. Considering the low complexity and scalability of MAB, we reformulate the optimization problem and design a reward function that jointly considers the training latency per round and the contribution of local data for reducing the overall latency. To estimate the contribution of local data on global aggregation, we introduce a utility function based on training loss of local datasets in the MAB reward function. Finally, an is an element of-greedy algorithm is proposed to balance exploration and exploitation within the dynamic and heterogeneous environment. Simulation results demonstrate that our lightweight and effective scheme outperforms existing approaches in reducing training time and accelerating convergence to the target accuracy. The results confirm that the proposed scheme can effectively address dynamicity and heterogeneity challenges under constrained resources of FL within IoV.

Simultaneously transmitting and reflecting reconfigurable intelligent surfaces (STAR-RISs) enable full-space coverage but also expose wireless transmissions to security from multiple spatial directions. This letter investigates a STAR-RIS-assisted secure RSMA system where both internal and external eavesdroppers may coexist in the transmission and reflection regions. In such a scenario, the RSMA common stream simultaneously serves legitimate users, impairs external eavesdroppers, and avoids assisting internal eavesdroppers, leading to a challenging trade-off between spectral efficiency and confidentiality. To address this issue, we formulate a max-min fairness problem under secrecy constraints and develop an iterative algorithm to jointly optimize transmit beamforming and STAR-RIS phase shifts. Simulation results demonstrate that the proposed scheme improves spectral efficiency while maintaining confidentiality.

Cell-free massive multiple-input multiple-output (CF-mMIMO) has emerged as one of the key technologies in the future wireless communications. Most existing CF-mMIMO systems employ wired fronthaul to connect access points (APs) with the central processing unit (CPU), but this will bring high cost on laying the fronthaul fiber. In this paper, we propose a heterogeneous CF-mMIMO system with part of APs using the millimeter wave (mmWave) to serve as the wireless fronthaul. Due to the short transmission distance of mmWave, mmWave APs (MAPs) can only distribute within a limited area around the CPU and serve nearby users. APs and users are distributed according to the Poisson point process. With tools of stochastic geometry, we derive a tight approximation of the system cost efficiency (CE) after evaluating the hardware cost on both the APs and fronthauls. Based on it, optimal MAP distribution range and spatial density that maximize the CE are obtained. Via simulation, we find that our proposed system can significantly improve the CE compared to the CF-mMIMO system with conventional APs and wired fronthauls. In addition, the results reveal that when each MAP is equipped with a small number of antennas, deploying more MAPs is beneficial to the system CE. In contrast, when each MAP has a large antenna array, deploying fewer MAPs turns better. Moreover, as user density increases, more MAPs should be deployed closer to the CPU.

III-nitride quantum well (QW) diodes have attracted considerable attention in the fields of display and optical communication due to their high brightness, fast response, and broad spectral characteristics. However, existing technologies primarily focus on single functionalities such as display or sensing, lacking multimodal integrated design. This study leverages the dual functionality of III-nitride QW diodes in both emission and detection and proposes a multifunctional optoelectronic platform based on a 4 & times; 4 III-nitride QW diode array, which operates within a responsive wavelength range from 360 to 470 nm. Multiple functions including noncontact sensing, dynamic display control, and parallel multichannel visible light communication (VLC) are integrated onto a single platform, with convolutional neural networks employed to accurately process the sensing results. Experimental results show that the system achieves single-pixel transmission rates up to 106 bps and reception rates up to 104 bps, with the overall channel capacity exhibiting significant theoretical scaling potential with the number of pixels. This solution is clean, contactless, and interference resistant, offering a highly integrated and innovative proof of concept for next-generation optoelectronic systems, particularly for interactive display and Internet of Things applications.

InGaN/GaN systems are commonly employed for designing near-ultraviolet (UV) light-emitting diodes (LEDs) due to their excellent quantum efficiency and tunable emission wavelengths. However, existing studies predominantly focus on the emission-related functionalities of near-UV LEDs, such as photopolymerization and anti-counterfeiting. This work breaks through the conventional limitation of using these diodes solely as emitters by innovatively designing and demonstrating a 4 × 4 array integrated with 16 near-UV multiple quantum well (MQW) diodes operating at 395-405 nm, exhibiting multifunctional capabilities. When in emission mode, the array performs traditional anti-counterfeiting and UV curing tasks. When in detection mode, the array acts as a photodetector, detecting externally stimulated pixels and enabling optical image input and character recognition via a convolutional neural network (CNN) with an accuracy exceeding 96%. During content display, the non-emitting pixels are repurposed for multiple-input multiple-output (MIMO) near-UV wireless optical communication (WLC), achieving per-pixel data rates up to 40 kHz. Moreover, the array's robustness under strong ambient sunlight was experimentally validated. The integration of these functionalities offers new perspectives and technological solutions for near-UV MQW diode applications, not only in conventional roles, but also in advanced areas such as covert optical input and anti-interference WLC.

This paper presents a reconfigurable intelligent surface (RIS)-enhanced backscatter communication system. In this system, the primary receiver (PRx) employs joint decoding to receive information from both the primary transmitter (PTx) and the RIS-backscatter device (RIS-BDx), akin to symbiotic radio (SR) systems. Concurrently, the backscatter receiver (BRx) utilizes an energy detector for demodulating information from the RIS-BDx, aligning with the methodology of ambient backscatter communication (AmBC) systems. We initially address the joint design of the PTx's transmit beamforming and the RIS's reflection coefficients (TBF-RC) within the system, with the objective of minimizing transmit power while adhering to transmission performance constraints. The challenge of this problem lies in the multitude of constant modulus (CM) constraints and fourth-order constraints. By exploiting the problem's structure, we break it down into two sub-problems featuring rate-balanced constraints and propose an alternating optimization (AO) algorithm with linear complexity. Furthermore, we extend the algorithm to tackle the TBF-RC design problem incorporating secrecy rate constraints, which safeguard against the BRx demodulating the primary information. This secure transmission issue underscores a key distinction between the proposed hybrid system and traditional SR systems, where the BRx is typically required to demodulate the primary information. Finally, simulation results are presented to demonstrate the efficacy, security, and computational efficiency of the proposed system and algorithm.

The surface wave (SW) and free-space coexisting propagation mechanisms in industrial internet-ofthings (IIoT) is investigated. A novel single-input multiple-output channel under the hybrid propagation mechanism is measured, modeled and analyzed. In the proposed channel model, one of the receiving antennas is coupled with a long SW transmission line, with the rest ones distributed in free-space mode. A path loss (PL) model with three degrees of freedom is established, including d(T-RSW), d(T-RFS) , and the offset angle theta. The d(T-RSW) dominate the interference degree of the SW propagation mode on the free-space propagation mode. Validation results show that the PL in the free-space propagation increases by 2-8 dB under the interference of the SW propagation mode. When d(T-RSW) is fixed and the offset angle theta is varied in the free-space mode, the PL increase is negligible (approximately 2 dB) for theta less than 30 degrees. When it is larger than 30 degrees, the PL escalates significantly. When theta is fixed and the d(T-RSW) is varied, the fading severity is weakened as the d(T-RSW) increasing. It indicates SW and free-space propagation can exist in IIoT scenarios, and the reliable deployment of the SW propagation can improve the transmission performance in IIoT scenarios.

Cell-free massive multiple-input multiple-output (CF-mMIMO) systems and simultaneous transmitting and reflecting reconfigurable intelligent surfaces (STAR-RIS) are considered as promising technologies for enhancing the performance of wireless communication systems. In this paper, we investigate the performance of a STAR-RIS-aided CF-mMIMO system under channel aging, which has been ignored in previous studies. Firstly, we propose a linear minimum mean squared error (LMMSE) aggregated channel estimator and formulate statistical channel state information (CSI) properties for the subsequent system performance analyses. Then, closed-form expressions for the uplink and downlink spectral efficiencies (SEs) of the STAR-RIS-aided CF-mMIMO system under channel aging are explored, where for the uplink, the two-layer large-scale fading decoding (LSFD) and the simple centralized decoding (SCD) are utilized, respectively, and for the downlink, the maximal ratio (MR) precoding and fractional power control (FPC) are adopted. Moreover, the optimal LSFD coefficients that maximize the uplink SE is presented. Afterwards, for further enhancement of SEs, a novel optimization scheme is presented, which optimizes the passive beamforming (PB) of the STAR-RIS to minimize the normalized mean square error (NMSE) of the aggregated channel estimation. The simulation results reveal that the STAR-RIS-aided CF-mMIMO system achieves superior uplink and downlink performance compared to both the RIS-aided CF-mMIMO system and the conventional CF-mMIMO system without RIS over aging channel. Furthermore, the results show that the PB optimization can significantly reduce the NMSE of channel estimation, thereby improving the estimation accuracy and SEs under channel aging.

As a cornerstone of the internet of things, massive grant-free random multiple access (MGFRMA) has attracted wide attention in recent years. However, due to lack of coordination, it is difficult to remove access collisions among randomly activated users, which makes the existing solutions unsatisfactory. To solve the issue, this paper introduces cognitive radio (CR) and preamble delay (PD) to reduce access collisions, together with non-orthogonal multiple access (NOMA) to improve spectrum efficiency, and then develops a CR-NOMA-PD-based MGFRMA scheme. First, a three-step MGFRMA protocol is advanced. It helps users to obtain the channel occupancy state before uplink transmission with spectrum sensing so as to consciously avoid access conflicts. Then, an optimisation strategy for jointly selecting access channel and power level is designed according to the sensing results. The competitive transmission scheme with PD is formulated, based on which uplink signals are well modelled. Finally, a multi-user detection algorithm involving channel filtering, power level detection, preamble detection, and data recovery is proposed. The performance of the CR-NOMA-PD-based MGFRMA scheme is also analysed and simulated. Simulation results indicate that the proposed scheme improves the access ratio of users and system overload capacity significantly compared to the existing schemes. It also has better robustness and scalability.

A fully mapped light communication network is demonstrated on gallium nitride (GaN) chips, wherein closed-loop communication units are enabled by GaN multiquantum-well diodes with light emission and detection functionalities. Bidirectional light communication between two nodes is achieved over a single waveguide under a time division duplexing scheme, while on-chip relay light communication is realized among three nodes. Three interconnected closed-loop units form a nine-node light communication network, enabling on-chip seamless video transmission. The decentralized and reconfigurable on-chip light communication network architecture is experimentally verified, providing a feasible route toward next-generation all-light information processing and system-on-chip computation.

Data-driven deep learning (DL) techniques have increasingly been employed to construct digital twin (DT) models for intelligent industrial Internet of Things (IIoT) systems.Within DTs, federated learning (FL) offers a decentralized framework that enables distributed entities to collaboratively update global models without sharing raw data. Clustered federated learning (CFL) further enhances training efficiency by grouping clients, thereby reducing communication overhead and accelerating model convergence. However, data heterogeneity arising from spatial distribution differences and system heterogeneity resulting in straggler clients jointly hinder the convergence and efficiency of CFL. The interplay of these factors introduces spatiotemporal coupling, which further degrades model training. To address these challenges, we propose a spatiotemporal coupling based CFL scheme that jointly optimizes client clustering and aggregation strategies to minimize overall training latency. A semi-synchronous aggregation mechanism is introduced, allowing clients to update at different frequencies based on their delay tiers. Furthermore, client clustering is performed according to location similarity to improve convergence, while clients with higher delay tiers and greater data diversity are prioritized for cluster head selection. To mitigate the impact of imbalanced cluster sizes under data heterogeneity, a balanced matching optimization is formulated to evenly distribute remaining clients to the nearest cluster heads. Within each cluster, adaptive bandwidth allocation is employed to satisfy delay-tier constraints and shorten communication rounds. Extensive simulations on CIFAR-10 and Fashion-MNIST with non-i.i.d. settings show that the proposed scheme can reduce the total training latency by up to 38.71% and 8.87%, respectively, to reach a fixed target accuracy, while achieving comparable model accuracy to existing baselines. These results confirm the effectiveness of the proposed scheme in heterogeneous IIoT environments.

Vertical-structure light-emitting diodes (VLEDs) offer dual functionality as emitters and photodetectors, leveraging spectral overlap between electroluminescence and photosensitivity in the visible range. Here, we present a large-area, structurally optimized green VLED (>0.8 mm(2)) fabricated via wafer-flipping, laser lift-off (LLO), metal bonding, and silicon substrate thinning. These steps integrate a reflective electrode, effectively improve the device's thermal management, and employ n-GaN surface roughening to boost light extraction. In conventional visible-light communication (VLC) transmitters, reducing the emission area improves modulation bandwidth by lowering parasitic capacitance but limits optical power and complicates integration. In contrast, the proposed large-area vertical architecture mitigates the RC bandwidth limitation through wafer-level bonding, substrate thinning, and distributed current injection, achieving a balanced trade-off between optical power and modulation speed. Despite its size, the VLED attains a peak transmission rate of 52.8 Mbps using 32-QAM discrete multitone (DMT) modulation, while serving as a self-powered photodetector achieving 3.5 Mbps optical signal reception under 405 nm illumination. Unlike micro-LEDs, which trade output power for speed, our design delivers higher optical power, enhanced heat dissipation, and simpler fabrication, demonstrating both high-power emission and self-powered detection within a single device. This work provides a cost-effective and scalable vertical architecture that bridges the gap between high-speed micro-LEDs and power-demanding VLC applications, offering a practical route toward integrated optical nodes for future Internet of Things (IoT) networks.

Reconfigurable intelligent surface (RIS)-enhanced secure wireless communication systems have attracted growing attention. However, RIS can also be misused to assist eavesdropping, introducing new security challenges. This article investigates the secrecy rate degradation caused by illegal RIS (IRIS) in a typical RIS-assisted integrated sensing and communication (ISAC) system, where a dual-functional base station performs multiuser communication and radar sensing simultaneously with the help of an RIS, while an eavesdropper (Eve) leverages an IRIS for malicious interception. Specifically, two types of sensing targets are considered, including point targets and extended targets. The detection probability is used as the sensing performance metric for a point target, while the Cram & eacute;r-Rao bound (CRB) of the complete target response matrix is used for an extended target. Based on these metrics, two different secrecy rate maximization problems are formulated under either a detection probability or a CRB threshold constraint. To solve the first nonconvex problem, we adopt a combination of closed-form fractional programming (CFFP), minorization-maximization (MM), successive convex approximation (SCA), and the alternating direction method of multipliers (ADMMs). For the second problem, semidefinite relaxation (SDR) and alternating optimization (AO) are employed to obtain suboptimal solutions. Simulation results demonstrate the great potential of ISAC systems and highlight the critical role of joint beamforming and reflection design, as well as strategic RIS deployment, in mitigating potential security threats posed by the IRIS.

This study explores an industrial Internet of Things (IIoT) scenario characterized by the dense deployment of femtocells within a macrocell coverage area. To address the critical challenges of inter-device interference and spectral efficiency optimization, we propose a novel D2D group architecture that integrates non-orthogonal multiple access (NOMA) technology for efficient group communications. However, the complex cumulative interference caused by channel sharing in such a network severely impairs the system performance, which was seldom considered in the previous study. To mitigate this issue, our framework proposes an innovative hypergraph-based spectrum allocation algorithm, where femtocell access points (FAPs) are represented as vertices and interference relationships are captured as hyperedges, enabling the application of hypergraph theory to formulate the optimal resource allocation problem. Comprehensive experimental evaluations demonstrate that the proposed scheme significantly improves the spectral efficiency compared to traditional graph-based spectrum allocation schemes.

Vehicular edge computing (VEC) supports real-time applications by offloading tasks to nearby servers, but traditional single radio access technology (RAT) VEC systems fail to meet diverse QoS demands. Heterogeneous VEC integrating DSRC and C-V2X radio access technologies enhances capacity but faces resource allocation challenges for partial task offloading from the mutual influence between continuous offloading ratios and discrete communication modes, leading to increased latency and task costs. To resolve this, we propose a local decision-making offloading framework for Vehicle User Equipment (VUE) in DSRC and C-V2X dual-mode VEC servers. Further, we establish a joint optimization model minimizing latency and task costs, then we propose a novel Hybrid Proximal Policy Optimization (H-PPO) algorithm. Unlike parallel action handling, our proposed H-PPO employs a dual-branch Actor network: first determining continuous offloading ratios under local constraints, then selecting discrete communication modes to decouple policy optimization. Experimental results show that our method achieves significant improvements in key performance metrics such as average delay, system cost and task completion rate compared to existing offloading schemes, and its performance is superior to the existing offloading schemes under different system environments and different vehicle densities.

The space-air-ground integrated network (SAGIN) can provide a promising architecture for computation-intensive mobile applications through the complementary advantages of unmanned aerial vehicles (UAVs) and low Earth orbit (LEO) satellites. However, efficient task offloading remains challenging due to the strong coupling among UAV trajectories, user association, and resource allocation, especially with dynamic and uncertain task workloads. In this paper, we aim to minimize the average task completion delay by enabling UAVs and satellites to cooperatively assist users in executing complex computational tasks. We formulate the joint optimization problem as a decentralized partially observable Markov decision process (DEC-POMDP), where task workload prediction is incorporated to enhance foresight and coordination among agents. Then, a task-prediction-augmented multi-agent collaborative offloading (TAMACO) framework is proposed, which integrates task-prediction-augmented multiagent proximal policy optimization (TA-MAPPO) for UAV trajectory control. Meanwhile, a prediction-augmented coalition formation game (PA-CFG) algorithm embedded into TA-MAPPO is proposed to solve the joint problem of user association and computing resource allocation. Simulation results demonstrate that the proposed TAMACO framework achieves up to 19.3% reduction in average task completion delay compared to nonpredictive baselines, validating its effectiveness for real-time task offloading in dynamic SAGIN.

This paper investigates a simultaneously transmitting and reflecting reconfigurable intelligent surface (STAR-RIS) assisted backscatter communication system. Our objective is to minimize the transmit power by jointly optimizing the transmit beamforming vector, the tags' reflection coefficients, the STAR-RIS coefficient matrices and energy splitting coefficients at the STAR-RIS, subject to the signal-to-interference-plus-noise ratio constraints for decoding backscatter signals and the power constraints of the tags. We propose an alternating optimization algorithm to find near-optimal solution. Numerical results show that our proposed scheme can significantly reduce the transmit power even under the assumption that all reflection/transmission elements have the same amplitude coefficients.

With the development of UAV communications and aerial edge intelligence, Uncrewed Aerial Vehicles (UAVs) are increasingly used for inspection in high-risk environments. However, limited onboard energy and data-privacy constraints make efficient collaborative learning challenging. Hierarchical Federated Learning (HFL) provides a privacy-preserving paradigm, yet existing energy-optimization studies often treat computing, communication, and client selection separately, lacking a unified system perspective. Inspired by biological neural systems, where neurons compute in an event-driven manner, synapses transmit information sparsely, and organisms make reward-modulated decisions, we propose a Bio-Inspired Hierarchical Federated Learning (BIO-HFL) framework that takes UAV energy consumption as the central driving objective. In the computation module, BIO-HFL employs Spiking Neural Networks (SNNs) to realize neuron-like event-driven computing and reduce onboard energy; in the communication module, it integrates a Critical Tensor mechanism into Deep Gradient Compression (DGC-CT) to mimic synapse-like sparse but selective transmission and maintain stability under high compression ratios; and in the control module, it uses a Distributed Multi-Armed Bandit (DMAB) strategy as an organism-level decision module to select clients with minimal expected energy consumption. These three components are mutually coupled, SNN and DGC-CT energy statistics serve as inputs to DMAB, DMAB determines the optimization targets of SNN and DGC-CT through energy-aware scheduling, and CT selection in DGC-CT further depends on SNN-driven spike distributions. Experiments demonstrate that DGC-CT ensures stable training even at a 10% compression ratio, and DMAB reduces inefficient client participation. Across both CIFAR-10 and DAGM2007, BIO-HFL consistently improves energy efficiency while maintaining competitive macro-average F1, achieving up to 81.1% lower total energy consumption compared with a conventional SNN-HFL baseline.

Rate-splitting multiple access (RSMA) is a promising non-orthogonal transmission scheme capable of achieving higher data rates with massive connectivity compared to conventional orthogonal multiple access. However, the effectiveness of RSMA can be significantly hindered by imperfect successive interference cancellation (SIC), leading to severe interference and rate degradation. Existing research has either focused on hybrid RSMA systems, where users are grouped and assigned orthogonal resources, or incorporated improper Gaussian signaling (IGS) to enhance interference management. However, no prior work has explored the combination of these approaches to flexibly manage interference under imperfect SIC. To address this gap, we propose a novel downlink hybrid RSMA system with IGS under imperfect SIC, where users are grouped into pairs to form RSMA groups. Specifically, we introduce the use of IGS for the common message within each group to effectively mitigate interference caused by imperfect SIC. The problem of optimizing the user grouping, subcarrier allocation, rate allocation, power allocation and IGS circularity coefficient, aimed at maximizing the sum rate under the minimum rate requirement, is investigated. We first develop a block coordinate descent method combined with successive convex approximation to optimize all variables except user grouping and subcarrier allocation. Subsequently, a swapping-based algorithm is proposed to refine user grouping and subcarrier allocation iteratively. Extensive simulation results validate the effectiveness of the proposed hybrid RSMA with IGS scheme, demonstrating its superior performance compared to various benchmark schemes in the literature.

This article considers an integrated sensing and communication (ISAC) system, where two dual-functional base stations (BSs) serve their users and detect multiple targets. To improve detection accuracy while meeting communication quality of service (QoS), this article proposes a two-phase cooperative target detection algorithm that relies on Capon-based adaptive beamforming and maximum likelihood estimation (MLE)-based hypothesis testing. Specifically, based on Capon's detection results, the two BSs first scan targets with an omnidirectional beam and then track targets with a directional beam. Subsequently, multiple hypotheses regarding the locations of targets are established based on the detection results of the Capon method, and the MLE is used for hypothesis testing to filter out ghost targets. Finally, simulation results show that the proposed algorithm achieves more precise angle-of-arrival (AoA) estimation of multiple targets than conventional single-BS sensing and enables high-precision localization by eliminating ghost targets.