With the exponential growth of smart devices connected to wireless networks, data production is increasing rapidly, requiring machine learning (ML) techniques to unlock its value. However, the centralized ML paradigm raises concerns over communication overhead and privacy. Federated learning (FL) offers an alternative at the network edge, but practical deployment in wireless networks remains challenging. This paper proposes a lightweight FL (LTFL) framework integrating wireless transmission power control, model pruning, and gradient quantization. We derive a closed-form expression of the FL convergence gap, considering transmission error, model pruning error, and gradient quantization error. Based on these insights, we formulate an optimization problem to minimize the convergence gap while meeting delay and energy constraints. To solve the non-convex problem efficiently, we derive closed-form solutions for the optimal model pruning ratio and gradient quantization level, and employ Bayesian optimization for transmission power control. Extensive experiments on real-world datasets show that LTFL outperforms state-of-the-art schemes.

Generative AI (GenAI) has emerged as a transformative technology, enabling customized and personalized AI-generated content (AIGC) services. In this paper, we address challenges of edge-enabled AIGC service provisioning, which remain underexplored in the literature. These services require executing GenAI models with billions of parameters, posing significant obstacles to resource-limited wireless edge. We subsequently introduce the formulation of joint model caching and resource allocation for AIGC services to balance a trade-off between AIGC quality and latency metrics. We obtain mathematical relationships of these metrics with the computational resources required by GenAI models via experimentation. Afterward, we decompose the formulation into a model caching subproblem on a long-timescale and a resource allocation subproblem on a short-timescale. Since the variables to be solved are discrete and continuous, respectively, we leverage a double deep Q-network (DDQN) algorithm to solve the former subproblem and propose a diffusion-based deep deterministic policy gradient (D3PG) algorithm to solve the latter. The proposed D3PG algorithm makes an innovative use of diffusion models as the actor network to determine optimal resource allocation decisions. Consequently, we integrate these two learning methods within the overarching two-timescale deep reinforcement learning (T2DRL) algorithm, the performance of which is studied through comparative numerical simulations.

Autonomous aerial vehicle (AAV) target tracking technology is an essential component for enabling diverse low-altitude activities. Due to the constraints on energy and computing resources of AAVs, current approaches face challenges in balancing prolonged flight duration with precise tracking while avoiding high computational complexity. Therefore, this paper proposes an energy-aware formation control algorithm for multiple AAVs to cooperatively track a target while retaining a desired formation pattern. Firstly, to achieve a balanced outcome in terms of tracking performance and control effort, an actor-critic based learning predictive rule is explored to develop a near-optimal control protocol that stabilizes error dynamics and minimizes value functions for discrete-time AAV systems. By decomposing the infinite-horizon target tracking problem into a sequence of finite-horizon sub-problems, the reinforcement learning (RL)-based predictive control algorithm can achieve fast convergence in approximating the solution of Hamilton-Jacobi-Bellman (HJB) equation. Furthermore, by employing a delicately designed asynchronous policy iteration mechanism with adjustable learning intervals in RL, the cumbersome learning process can be effectively mitigated, thereby attaining both high learning efficiency and a reduced computational burden simultaneously. The involved errors are proven to be convergent and simulation results validate the optimality of our method.

It is significant to employ multiple autonomous underwater vehicles (AUVs) to execute the underwater target tracking task collaboratively. However, it's pretty challenging to meet various prerequisites utilizing traditional control methods. Therefore, we propose an effective two-stage learning from demonstrations training framework, FISHER, to highlight the adaptability of reinforcement learning (RL) methods in the multi-AUV underwater target tracking task, while addressing its limitations such as extensive requirements for environmental interactions and the challenges in designing reward functions. The first stage utilizes imitation learning (IL) to realize policy improvement and generate offline datasets. To be specific, we introduce multi-agent discriminator-actor-critic based on improvements of the generative adversarial IL algorithm and multi-agent IL optimization objective derived from the Nash equilibrium condition. Then in the second stage, we develop multi-agent independent generalized decision transformer, which analyzes the latent representation to match the future states of high-quality samples rather than reward function, attaining further enhanced policies capable of handling various scenarios. Besides, we propose a simulation to simulation demonstration generation procedure to facilitate the generation of expert demonstrations in underwater environments, which capitalizes on traditional control methods and can easily accomplish the domain transfer to obtain demonstrations. Extensive simulation experiments from multiple scenarios showcase that FISHER possesses strong stability, multi-task performance and capability of generalization.

With the integrated development of global marine exploitation and 6G technology, building an all-domain marine wireless network has become crucial for supporting marine activities. However, the unique communication environment, varying collaboration of heterogeneous devices, and dynamic network changes pose technical bottlenecks for balancing real-time and efficient resource competition. To overcome those challenges, this paper proposes a novel integrated marine wireless network with multi-type unmanned device clusters across space-surface-submarine media. To address heterogeneous resource allocation, we consider channel capacity and device connection, modeling it as a three-sided matching framework with size constraints and cyclic preferences (TMSC). Building on this, we propose the satellite-prioritized restricted double-TMSC (SPR-DT) algorithm to solve optimal matching in quasi-static networks, aiming to maximize total backhaul revenue. To handle rapid dynamic network changes, we initialize the proximal policy optimization (PPO) with the stable solution of SPR-DT, thus addressing the challenge of acquiring real training data while accelerating algorithm convergence. Then, we propose a PPO-assisted multi-slot matching algorithm to enhance solution efficiency in large-scale dynamic scenarios. The simulation results show that the proposed algorithm achieves an optimal effect of 94.6% in quasistatic scenarios, with a complexity reduced to 3.2%. In dynamic scenarios, the results are 87.2% and 28.7%, respectively.

Integrated sensing and communication (ISAC) technology is crucial for next-generation underwater networks. However, covering multiple users and targets and balancing sensing and communication performance in complex under-water acoustic (UWA) environments remains challenging. This paper proposes an interleaved orthogonal frequency division multiplexing-based MIMO UWA-ISAC system, which employs a horizontal array to simultaneously transmit adaptive waveforms for downlink multi-user communication and omnidirectional target sensing. A multi-objective optimization framework is formulated to maximize the product of communication rate and range (PRR) while ensuring sensing performance and peak-to-average power ratio (PAPR) constraints. To solve this mixed-integer nonconvex problem, a two-dimensional grouped random search algorithm is developed, efficiently exploring subcarrier interleaved patterns and resource allocation schemes. Numerical simulations under real-world UWA channels demonstrate the designed system’s superiority and effectiveness: our algorithm achieves 90% faster convergence than conventional exhaustive search with only a marginal 0.5 kbps•km PRR degradation. Furthermore, the proposed resource allocation scheme maintains robustness beyond the baseline allocation schemes under stringent PRR and PAPR constraints.

Underwater acoustic sensor networks (UASNs) play a key role in ocean resource exploration and complex underwater tasks. However, the open acoustic channel makes them vulnerable to malicious node attacks. Therefore, accurately identifying attack nodes in harsh channels and adapting to their mobility presents a significant challenge. To address these issues, we adopt a meandering ocean current mobility model to describe node movement and construct a hidden Markov model (HMM) along with link transmission loss to characterize the unstable underwater acoustic channel. By monitoring the forwarding behavior of neighboring nodes and combining HMM state inference, we propose a trust model based on a subjective logic framework with dynamic topology updates to detect malicious nodes. It considers variable weights to assess improper node behavior and dynamically updates trustworthiness based on both historical trust and arrival strategies of new and old nodes. Simulation results indicate that the proposed method effectively identifies malicious nodes with attack intensities exceeding 0.38, and for intensities above 0.6, it achieves over 90% identification accuracy and adapts well to dynamic environmental mobility.

The rapid advancement of technologies such as the Internet of Things (IoT), autonomous driving, and smart manufacturing has led to a massive increase in data generation at the edge of networks. This necessitates effective machine learning (ML) methods that address challenges like communication overhead and privacy concerns. Federated learning (FL) has emerged as a promising solution for distributed model training, but the increasing complexity of ML models limits its communication efficiency. To address these challenges, we propose an ultra energy-efficient FL framework (FedUEE). FedUEE utilizes model pruning-based compressive sensing, outage compensation, and joint optimization of learning and resource configurations to comprehensively reduce energy consumption. We develop analytical models that quantify the energy impact of each proposed mechanism, ultimately providing an optimized solution for communication efficiency in edge FL environments.

Collaborative underwater target hunting, facilitated by multiple autonomous underwater vehicles (AUVs), plays a significant role in various domains, especially military missions. Existing research predominantly focuses on designing efficient and high-success-rate hunting policy, particularly addressing the target's evasion capabilities. However, in real-world scenarios, the target can not only adjust its evasion policy based on its observations and predictions but also possess eavesdropping capabilities. If communication among hunter AUVs, such as hunting policy exchanges, is intercepted by the target, it can adapt its escape policy accordingly, significantly reducing the success rate of the hunting mission. To address this challenge, we propose a covert communication-guaranteed collaborative target hunting framework, which ensures efficient hunting in complex underwater environments while defending against the target's eavesdropping. To the best of our knowledge, this is the first study to incorporate the confidentiality of inter-agent communication into the design of target hunting policy. Furthermore, given the complexity of coordinating multiple AUVs in dynamic and unpredictable environments, we propose an adaptive multi-agent diffusion policy (AMADP), which incorporates the strong generative ability of diffusion models into the multi-agent reinforcement learning (MARL) algorithm. Experimental results demonstrate that AMADP achieves faster convergence and higher hunting success rates while maintaining covertness constraints.

Unmanned aerial vehicles (UAVs), which can leverage their integrated sensing, computation, and communication (ISCC) capabilities to enable distributed intelligence at the network edge, play a critical role in next-generation wireless networks. However, UAVs face significant challenges in decentralized model training, including limited computational resources, energy constraints, and inefficient communication. This paper introduces SFL-ISCC, a novel framework that combines model splitting with federated learning (FL)-supported UAV ISCC systems, designed to train a global machine learning (ML) model with minimal energy consumption across multiple UAVs. To our knowledge, this is the first attempt to integrate model splitting into FL-supported UAV ISCC systems. Within the framework, we first theoretically explore the effects of UAV deployment strategies, split layer selection, and client-side aggregation frequency on model convergence performance. Then, we formulate a joint optimization problem to minimize UAV energy consumption while guaranteeing target model convergence accuracy, and propose a low-complexity solution. Moreover, we incorporate semisupervised learning into the SFL-ISCC framework to handle the sparsity of labeled data in UAV networks. Experimental results show that our proposed scheme significantly outperforms baseline schemes in both energy efficiency and model convergence.

Low-earth orbit (LEO) satellite-assisted marine communication networks have become a research focus with the growth of marine activities. However, establishing communication links between underwater devices and maritime satellites is a challenge. Additionally, dynamic environments and multidomain media pose significant challenges in allocating resources effectively within this network. To address these issues, this paper constructs a Space-Surface-Submarine Unmanned Network (3SUN) incorporating LEO satellites, unmanned surface vehicles (USVs), and unmanned underwater vehicles (UUVs). We formulate the resource allocation problem in the 3SUN as a satellite revenue maximization problem. We propose a satelliteprioritized restricted three-sided matching algorithm to solve the match within a single time slot. Additionally, we incorporate deep reinforcement learning (DPL), using the previous stable matching results as training initialization to tackle dynamic connections across multiple slots. Simulation results show that our algorithm achieves satellite revenue closer to the optimal solution compared to other methods while maintaining lower time complexity.

Semantic communication (SC) is considered a critical technology for breaking through the Shannon limit and achieving low-latency, high-capacity 6 G transmission. However, previous SC systems have typically employed a symmetrical architecture to enhance data recovery capabilities, resulting in a strong coupling between the encoder and decoder. In this paper, we introduce a novel asymmetric SC system, termed masked joint source-channel coding (M-JSCC), which significantly enhances the encoder's versatility by allowing it to adapt to different decoder models tailored to specific task requirements. Moreover, we abandon traditional convolutional neural networks and adopt the innovative transformer to increase model capacity further. Additionally, we empower the model with data generation capabilities to combat interference and distortion during wireless transmission, achieving robust semantic transmission. As a result, extensive experiments verify that our M-JSCC achieves better semantic understanding and performance across various tasks and different channel conditions.

Flying ad hoc networks (FANETs) composed of small unmanned aerial vehicles (UAVs) are flexible, inexpensive, and fast to deploy, which have been used in an increasing number of mission scenarios. However, unstable link quality and frequently changing network topology pose significant challenges for adopting existing routing protocols in mobile ad hoc networks (MANETs). In this paper, we propose a time-slotted on-demand predictive (TSDP) routing protocol designed specifically for UAV networks. The TSDP protocol introduces a novel approach to route selection by incorporating multiple criteria, including delivery ratio, adjacent degree, and mobility prediction factor, to ensure reliable and efficient data transmission. By addressing high latency in route discovery and excessive broadcast overhead, TSDP employs a time-slotted communication mechanism that reduces packet drop rates and enhances route stability. Simulation results demonstrate that TSDP consistently outperforms ad hoc on-demand distance vector (AODV) and dynamic source routing (DSR) protocols in terms of throughput, packet delivery ratio, end-to-end delay, and overhead, particularly in highly dynamic network environments.

Federated learning (FL) is a promising approach that effectively and securely harnesses the vast amounts of data generated by the rapid proliferation of internet-connected devices. In FL, the transmission of model parameters over wireless channels plays a pivotal role in determining system performance. To optimize the wireless environment and boost communication efficiency, we present a novel FL beamforming design scheme that integrates multi-sector beyond diagonal reconfigurable intelligent surfaces (BD-RIS) with over-the-air computation (AirComp). The scheme leverages the waveform superposition property of wireless signals, using AirComp to rapidly aggregate the global model in FL. Additionally, the scheme utilizes BD- RIS to flexibly manip-ulate communication beams, improving user channel conditions and further reducing model aggregation errors. Specifically, we evaluate the impact of this design on FL systems and derive an upper limit on the gap between training loss and optimal loss. To minimize this gap, we formulate a joint optimization problem of BD- RIS passive beamforming and base station receive beamforming, and we propose an optimization algorithm based on successive convex approximation (SCA) and block coordinate descent (BCD) to solve it. Simulation results confirm that our de-sign significantly enhances user channel conditions and improves FL performance, with the benefits becoming more pronounced as the number of BD- RIS reflecting elements increases.

The integration of federated learning (FL) with the Internet of Things (IoT) represents an advanced technological trend, combining the extensive connectivity of IoT with the powerful processing capabilities of FL to drive innovation and optimization across multiple domains. Given the heterogeneity of IoT devices and the variability in data distribution, developing strategies to enhance FL performance without overly burdening resource-constrained devices is crucial. This article proposes an FL algorithm based on generative artificial intelligence (GAI) for IoT devices with extreme heterogeneity in data and resources. The algorithm utilizes pretrained GAI models to generate new data, aligning the data distributions of individual IoT devices closer to independent and identically distributed (i.i.d.), thereby effectively reducing the heterogeneity of local data. Additionally, the proposed algorithm incorporates data synthesis and resource scheduling strategies to mitigate the heterogeneity of local device resources. Finally, we formulate a joint optimization problem aimed at minimizing total energy consumption while maximizing FL performance. Experimental results demonstrate that, under significant resource and data distribution disparities, most existing solutions struggle to converge, whereas the proposed method converges and achieves superior performance. Compared to existing GAI-based approaches, our method significantly reduces latency and energy consumption.

Unmanned underwater vehicles (UUVs) have been widely used in various ocean applications, such as underwater exploration and data collection. And the underwater pursuit-evasion game (UPEG) is the key to efficient implementation of other tasks, holding significant research value. However, testing the UPEG task in real ocean environment is both costly and risky, and currently, UUV control algorithms that rely on specific environmental models struggle to complete the complicated UPEG task. To address above challenge, we propose UPEGSim, an UUV simulator specifically designed for the UPEG task. Built through Gazebo and robot operating system, UPEGSim provides a reinforcement learning (RL) environment to train UUVs for improving the intelligent performance in the UPEG task. Furthermore, we propose an efficient UPEG training framework (ETFDU), which includes multiagent decentralized training and execution techniques, scene transfer training methods, and offline RL techniques based on decision transformer, to facilitate efficient UUV training. Through training on the UPEG task in UPEGSim, we validate the effectiveness and feasibility of the proposed UPEGSim simulator and the ETFDU training framework.

Emerging real-time computer vision (CV) applications on wireless edge devices demand energy-efficient and privacy-preserving learning. Federated learning (FL) enables on-device training without raw data sharing, yet remains challenging in resource-constrained environments due to energy-intensive computation and communication, as well as limited and non-i.i.d. local data. We propose FedDPQ, an ultra energy-efficient FL framework for real-time CV over unreliable wireless networks. FedDPQ integrates diffusion-based data augmentation, model pruning, communication quantization, and transmission power control to enhance training efficiency. It expands local datasets using synthetic data, reduces computation through pruning, compresses updates via quantization, and mitigates transmission outages with adaptive power control. We further derive a closed-form energy–convergence model capturing the coupled impact of these components, and develop a Bayesian optimization (BO)-based algorithm to jointly tune data augmentation strategy, pruning ratio, quantization level, and power control. To the best of our knowledge, this is the first work to jointly optimize FL performance from the perspectives of data, computation, and communication under unreliable wireless conditions. Experiments on representative CV tasks show that FedDPQ achieves superior convergence speed and energy efficiency.

Integrated sensing and communication (ISAC) technology is a promising contender for the future Internet of Underwater Things (IoUT). However, the complexity of underwater acoustic (UWA) channels and the randomness of ISAC signals may pose challenges to underwater communication and sensing. To address this issue, this paper investigates a novel communication-assisted bi-static sensing scheme capable of facilitating underwater multi-node collaboration using orthogonal frequency division multiplexing (OFDM), which refers to as UWA-OFDM-ISAC. Moreover, two efficient receivers are designed based on compressed sensing to enhance communication and sensing performance. In this paper, we first portray the UWA-OFDM-ISAC system model and emphasize that the Doppler and symbols estimated at the communication side are beneficial in enhancing the bi-static sensing performance. To estimate doubly spread UWA channels, an orthogonal matching pursuit-based interference cancellation channel estimation method is developed, which decouples Doppler and delay in OFDM signals and significantly reduces the parameter search dimension. Furthermore, we propose an enhanced detection algorithm based on matching pursuit, which can exploit sparse multipath information of echoes to improve target detection performance under doubly spread channels. The detection probability is improved by more than 30% at the 10−2bit error rate level compared with the energy detector. Finally, simulation results illustrate the effectiveness of the proposed UWA-OFDM-ISAC and demonstrate that the designed receivers have significant advantages relative to various existing algorithms.

Underwater target hunting (UTH) is a critical and complex mission involving the search, monitoring, and hunting of targets in an underwater environment. However, the unpredictable trajectories and flexibility of these targets, along with complex underwater environments, significantly impede the efficiency and success of traditional schemes that depend solely on unmanned underwater vehicles (UUVs). Consequently, this paper presents the “3U network”, a novel heterogeneous framework integrating unmanned aerial vehicles (UAVs), unmanned surface vehicles (USVs), and UUVs for UTH. Within this framework, a UAV searches and monitors the target, a USV acts as a communication relay, and a swarm of UUVs hunts the target. Moreover, to improve the timeliness of target search, we propose the age of information (AoI)-based UAV search strategy. Additionally, we construct a constrained multi-objective optimization problem aiming to minimize energy consumption and mission duration by optimizing vehicles' trajectories, considering mobility limitations, safety, and connectivity constraints. To tackle this problem, we design an AoI- and energy-aware deep reinforcement learning (DRL) algorithm to optimize control policies for heterogeneous vehicles. The experimental results demonstrate that the proposed scheme outperforms the baseline schemes in terms of energy consumption, and mission duration and success rates.

Emerging applications are placing increasing demands on wireless networks, particularly in terms of ensuring reliable communication for control-related information. However, the complexity of network architectures and the growing number of user devices present significant challenges in achieving reliable multiple access. In this paper, we present a framework that utilizes machine learning (ML) to meet the need for reliable access in unmanned aerial vehicle (UAV) networks. The K-means algorithm is employed to cluster users according to their communication reliability requirements, grouping together users with similar demands within each cluster. Each cluster adopts a different access strategy: clusters with lower reliability requirements utilize non-orthogonal multiple access to enhance spectrum efficiency, while clusters with higher reliability requirements employ orthogonal multiple access to ensure reliability. Taking into account the impact of UAV altitude and power allocation schemes on reliability, we propose an iterative algorithm to optimize the UAV altitude and power allocation factors, aiming to maximize UAV coverage while meeting the users’ reliability requirements. The simulation results validate the effectiveness of the proposed ML-based reliable access scheme, highlighting its potential to enhance the design and deployment of reliable communication in future UAV networks.

Crowdsourcing is an effective method to alleviate the burden of conducting a site-survey procedure for localization tasks. However, crowdsourced data is typically inaccurately and scarcely annotated, rendering accurate localization a rather challenging problem. To alleviate this problem, we propose VRLoc, a deep reinforcement learning (DRL)-based unsupervised wireless localization framework using crowdsourced trajectory data. The proposed VRLoc primarily encompasses three components, i.e., a robust K-means (RKM) clustering method for generating a series of virtual reference points (VRPs), DRL for determining the physical layout for VRPs, and online localization based on VRPs. Specifically, the proposed RKM method employs a density-based approach for the initialization of cluster centers, rather than the commonly used random solution, yielding repeatable and reliable VRP generation results. To accurately determine the physical locations for VRPs, we develop a modified soft actor-critic (SAC)- based VRP layout method with multiple objectives, i.e., the connection topology among VRPs, the floor-plan information, and the near-field condition. Then, we effectively predict locations of target users by utilizing classification models to match the online collected samples with the VRPs annotated by physical locations. The proposed framework is advantageous in achieving high-accuracy unsupervised localization, with the VRPs bridging the unlabeled crowdsourced data and physical location space. Both experimental and simulation results demonstrate the effectiveness and superiority of the proposed VRLoc framework as an accurate and practical solution for unsupervised localization.

Integrated sensing and communication (ISAC) technology is envisioned as a pivotal component for the next-generation communication networks. Similarly, underwater acoustic ISAC (UWA-ISAC) holds promising prospects for enhancing future UWA sensor networks due to its efficient communication and sensing capabilities. However, the design of UWA-ISAC waveforms and receivers suitable for multi-user scenarios faces formidable challenges, primarily arising from the complex UWA channel and multi-access interference (MAI). In this paper, we propose a UWA-ISAC waveform design scheme based on generalized sinusoidal frequency modulation (GSFM). The proposed waveform provides satisfactory communication and sensing performance and exhibits excellent orthogonality. Furthermore, we design a MAI suppression receiver, leveraging successive interference cancellation based on two-factor compensation and turbo equalization to improve interference suppression capabilities and enhance communication performance. Simulation results validate that the proposed UWA-ISAC waveform has an approximate thumbtack ambiguity function and comparable cross-correlation properties with GSFM under the defined parameters. Moreover, the designed receiver with low training sequence overhead is robust against Doppler, and can iteratively improve the MAI suppression performance.

Autonomous underwater vehicles (AUVs) are increasingly utilized across various domains, employing diverse machine learning (ML) algorithms to enhance functionality. However, the dynamic underwater environment, characterized by high temporal and spatial variability, makes offline-trained models on static datasets inadequate for practical AUV operations. This necessitates the integration of online learning capabilities that can adapt to changing conditions in real time. The quality of training data plays a crucial role in the performance of these models, and the ability to utilize distributed data from multiple AUVs can be beneficial. Nonetheless, the challenge lies in the limited communication resources available underwater. The typical acoustic communication rates, which are just in the tens of kilobits per second, pose a significant barrier to implementing centralized ML strategies that require extensive data sharing among AUVs. To overcome these hurdles, we propose an underwater federated learning (UFL) framework that incorporates model pruning and gradient quantization. This approach aims to establish a communication-efficient distributed learning paradigm. Furthermore, we have derived a closed-form expression to quantify the upper bound of the convergence error, which highlights the impact of pruning and quantization on the federated learning (FL) convergence. Additionally, we utilize a heuristic algorithm to optimize the pruning and quantization strategies, aiming to minimize the convergence error while adhering to delay constraints. The effectiveness of our proposed framework is demonstrated through its application in a cooperative navigation task involving multiple AUVs, showing significant resource conservation and enhanced operational efficiency.

Federated Learning (FL) enables distributed intelligence in Internet of Things (IoT) networks, facilitating decentralized machine learning without the need for exchanging raw data. However, the growing complexity of training models significantly hinders their deployment on resource-constrained IoT devices. To address this challenge, Split Federated Learning (SFL) has emerged as a promising solution by partitioning the entire model into client-side and server-side sub-models to alleviate the computational burden on IoT devices. Considering that the client-edge-cloud architecture can enhance data privacy, support connections to a wider range of devices, and reduce communication costs, we explore a hierarchical SFL (HierSFL) system. This system is supported by a HierSFL algorithm that allows for different aggregation frequencies between the client-side and server-side sub-models. Then, we present a convergence analysis of HierSFL that quantifies the effects of client-side and server-side model aggregation on learning performance, providing a theoretical foundation. Empirical experiments verify the theoretical analysis and demonstrate the superiority of the hierarchical architecture within a wireless IoT network. In particular, it is validated that adopting different aggregation frequencies can enhance the training performance. Moreover, the HierSFL algorithm outperforms traditional hierarchical FL algorithm, achieving superior test accuracy in a shorter time.

With recent advents of industrial Internet of Things (IIoT), the connectivity and data collection capabilities of industrial equipment have be significantly enhanced, yet bringing new challenges for the remaining useful life (RUL) prediction. To fulfill the RUL predicting demand in multivariate time series, this work proposes an encoder-decoder model termed as dual-scale transformer model (DSFormer), built upon the Transformer architecture. First, in the encoder part, a dual-attention module is designed for the weight feature extraction from both dimensions of the sensor and time series, aiming to compensate for the diverse impacts of different sensors on the prediction. Next, a temporal convolutional network (TCN) module is introduced to capture sequence features and alleviate the loss of positional information incurred by stacking blocks. Then, the feature decomposition module is integrated into the decoder for trend feature extraction from sequences, providing the model with additional sequence information. Finally, compared to existing models, the proposed method can obtain the superior performance in terms of the root mean square error (RMSE) and Score metrics on the FD001, FD002 and FD003 subsets of the C-MAPSS dataset, with an average improvement of 3.2% and 2.5% respectively. In particular, the ablation experiment further validates the effectiveness of proposed modules in handling multivariate time series and extracting features.

Unmanned aerial vehicles (UAVs) empowered Internet of things (IoT) networks have emerged as a burgeoning paradigm in the era of 6G. However, due to substantial data volume and privacy concerns, the conventional UAV backhaul to cloud center framework is not applicable to various latency and privacy-sensitive applications. Therefore, we propose a blockchain-based hierarchical federated learning (FL) framework for UAV-enabled IoT networks. Specifically, we utilize the total data distance-aware device association to mitigate model impairment arising from imbalanced data distribution. Besides, we introduce a lightweight blockchain into FL to tackle the trust deficit caused in decentralized global model aggregation. Furthermore, we design an optimization framework that jointly orchestrating device association, wireless resource allocation, and UAV deployment, aiming at a balance between the learning latency and model accuracy. To address the formulated optimization problem, we proposed a two-stage algorithm that integrates both greedy strategy and soft actor-critic algorithm. Extensive experiments show that our proposed scheme outperforms contemporary relative to state-of-the-art alternatives.

Metaverse is envisioned to merge the actual world with a virtual world to bring users unprecedented immersive feelings. To ensure user experience, federated learning (FL) has been expected as a critical enabler to provide metaverse users with high-quality sensing, communicating, and rendering. However, considering the limitation of wireless communication resources and the stringent requirements of users, collaborating with massive metaverse users to realize FL still has tremendous challenges. Most pioneer works on improving the performance of FL assume that the system states are static, which is unsuitable in the metaverse. Because the FL in the metaverse is always a complicated long-term iteration process, where the fluctuations of channel status and available computing resources of users are inevitable, a changeless strategy may lead to poor results. Therefore, this paper proposes an efficient FL scheme relying on dynamic user selection, gradient quantization, and resource allocation. Specifically, we derive the convergence error bound to reveal the impact of user selection, wireless transmission error, and gradient quantization error of each iteration on FL’s convergence. Based on the theoretical analysis, we jointly and dynamically optimize the user selection, gradient quantization, and resource allocation to minimize the error bound with time and energy consumption budgets. Furthermore, to make the formulated sequential decision-making problem tractable, we transform it into a Markov decision process and design a soft actor-critic-based solution. Extensive experiments validate that our proposed scheme has superior performance compared to conventional schemes in dynamic-changing network environments.

To support emerging environmentally-aware intelligent applications, a massive amount of data needs to be collected by sensor devices and transmitted to edge/cloud servers for further computation and analysis. However, due to the high deployment and operational cost, only depending on terrestrial infrastructures cannot satisfy the communication and computation requirements of sensor devices in the unexpected and emergency situations. To tackle this issue, this paper presents a digital twin-enabled space-air-ground integrated sensing, communication and computation network framework, where unmanned aerial vehicles (UAVs) serve as aerial edge access point to provide wireless access and edge computing services for ground sensor devices, and satellites provide access to cloud data center. In order to tackle the complex network environments and coupled multi-dimensional resources, the digital twin technique is utilized to realize real-time network monitoring and resource management, and the mapping deviation is also considered. To realize real-time data sensing and analysis, we formulate a maximum execution latency minimization problem while satisfying the energy consumption constraints and network resource restrictions. Based on the block coordinate descent method and successive convex approximation technique, we develop an efficient algorithm to obtain the optimal sensing time, transmit power, bandwidth allocation, UAV deployment position, data assignment strategy, and computation capability allocation scheme. Simulation results demonstrate that the proposed method outperforms several benchmark methods in terms of maximum execution latency among all sensor devices.

Unexploded Ordnance (UXO) buried in the seafloor poses a serious threat to the environment and human safety. Removing UXOs in the ocean presents a tricky challenge due to the greater difficulty in controlling the damage caused by explosions compared to those on land. Therefore, this study designs a method for detecting seafloor-buried UXO using the transient electromagnetic (TEM) approach and proposes a new forward model for UXO characteristic responses in the ocean by modeling the marine environments as a two-layer medium. We first numerically solve the time-harmonic equations of the primary magnetic field using Sommerfeld integrals and Hankel transforms. Then, we derive the characteristic responses of UXO in seawater based on the three-dimensional magnetic dipole model and the TEM method. Finally, we simulate the characteristic responses of six typical UXOs and two interfering targets, comparing them with those in free space and analyzing the effects of type, measured distance, and buried attitude on identification. The results show that the characteristic responses in seawater delay 2-12 ms with target size compared to free space. The errors of the characteristic response measurement depend not only on the measured distance but also on the buried attitude of the target. The errors reach the maximum when the target is vertical, the minimum when horizontal at the same distance, and disappear when the measured distance is longer than twice the target size. These findings establish a crucial foundation for accurately identifying the type, burial state, and location of UXO during marine demining.

With the booming development of Smart Healthcare Systems (SHSs), employing federated learning (FL) in SHS devices has become a research hotspot. FL, as a distributed learning framework, can train models without sharing the original data among users, and then protect the user privacy. Existing research has proposed many methods to improve the security and efficiency of FL, which may not fully consider the characteristics of SHSs. Specifically, the requirements of privacy protection and efficiency pose significant challenges to FL. Current studies have struggled to balance privacy security and efficiency, and the degradation of model training efficiency in SHSs can be critical to patient health. Therefore, to improve the privacy protection of healthcare data and ensure communication efficiency, this work proposes a novel personalized FL framework based on Communication quality and Adaptive Sparsification (pFedCAS). In order to achieve privacy protection, a control unit is proposed and introduced to adjust the sparsity of the local model adaptively. To further improve the training efficiency, a selection unit is added during global model aggregation to select suitable clients for parameter updates. Finally, we validate the proposed method operated on the HAM10000 dataset. Simulation results validate that pFedCAS can not only improve privacy protection, but also gain an improvement of 15% in training accuracy and a reduction of 30% in training costs based on communication quality. The simulation results also validate the excellent robustness of pFedCAS to non-iid data.

Unmanned aerial vehicles (UAVs) associated with federated learning (FL) have been deemed as a prospective framework by utilizing private data generated in the edge devices. However, despite various privacy-preserving and cryptography technologies adopted at the data level, FL still faces a range of security threats to raw data considering the broadcast nature of wireless channel. In this paper, to facilitate the communication-efficiency and privacy-preservation capability, we propose a UAV-enhanced covert federated learning architecture over mmWave massive multiple input multiple output (MIMO) channel, where we harness the covert communication technique in FL in order to avoid eavesdropping of illegal wardens. To achieve a trade-off between the security performance and training cost, we formulate a joint optimization problem involving the UAV’s trajectory, transmitting power, analog beamforming, and the required accuracy of FL. Furthermore, we propose the multi-agent deep deterministic policy gradient (MADDPG) algorithm to solve the above-mentioned problem. Numerous simulations have been performed to demonstrate both the effectiveness and convergence of the proposed algorithm.

The low earth orbit (LEO) satellite constellation is regarded as an effective complement to the terrestrial communication system due to its seamless coverage and ultra-low latency. Unfortunately, the highly dynamic traffic volume, as well as the inherent nature of dynamic topology changes caused by frequent link handover and uncertain hardware failures, pose severe challenges in the design of reliable routing. However, most existing reliable routing approaches with distributed schemes only focus on information exchange between adjacent nodes, which makes them fail to perceive real-time global network changes and make optimal decisions. In this paper, we propose a software defined networking (SDN) based intelligent satellite routing (SISR) method to increase the adaptivity and reliability during the packet transmission process. With the facilitation of SDN, we manage the network in a hierarchical and centralized paradigm, and further implement a more refined form of reinforcement learning (RL) to enhance the fault-tolerant ability of satellite network routing. Experimental results show that our solution can reduce latency and packet loss ratio by more than 42% and 29% compared to baselines.

Wireless virtual reality (VR) is expected to be a technology that revolutionizes human interaction and perceived media, where the quality of experience (QoE) is an important indicator to measure user service perception. However, existing schemes only consider general and time-invariant QoE optimization, which may suffer performance degradation. Moreover, it is also necessary to ensure the fairness of the individual user’s performance in wireless VR. To address these challenges, we employ digital twin technology to investigate a max-min QoE-optimal problem for wireless VR systems in this paper. Specifically, we maximize the QoE of the worst-case head-mounted displays (HDMs) client, where the QoE model is the linear weighting combination of video quality, service delay, and energy efficiency. The formulated optimization problem is characterized by multidimensional control, which jointly optimizes model selection, transmit power, computation time, and GPU-cycle frequency. Due to the mixed combinatorial features of the optimization problem, we give a low-complexity algorithm design by decoupling the optimization variables. Notably, we first obtain the allocation of the transmit power by employing the generalized fractional programming theory and the Lagrangian dual decomposition, followed by attaining the optimal allocation of GPU-cycle frequency in VR mode is derived by the proposed adaptive modified harmony search algorithm, and finally achieve the computation time by the barrier method. Meanwhile, we devise a greedy-style heuristic algorithm for mode selection. In the simulation, three baseline schemes are established as comparisons to assess the effectiveness of the proposed scheme. Meanwhile, the simulation results manifest that the proposed algorithms have good convergence performance and better increase the QoE of the DT-enabled wireless VR system compared to benchmark solutions.

The Internet of underwater things (IoUT) is envisioned to be an essential part of maritime activities. Given the IoUT devices’ wide-area distribution and constrained transmit power, autonomous underwater vehicles (AUVs) have been widely adopted for collecting and forwarding the data sensed by IoUT devices to the surface-stations. In order to accommodate the diverse requirements of IoUT applications, it is imperative to conceive a multi-tier underwater computing (MTUC) framework by carefully harnessing both the computing and the communications as well as the storage resources of both the surface-station and of the AUVs as well as of the IoUT devices. Furthermore, to meet the stringent energy constraints of the IoUT devices and to reduce the operating cost of the MTUC framework, a joint environment-aware AUV trajectory design and resource management problem is formulated, which is a high-dimensional NP-hard problem. To tackle this challenge, we first transform the problem into a Markov decision process (MDP) and solve it with the aid of the asynchronous advantage actor-critic (A3C) algorithm. Our simulation results demonstrate the superiority of our scheme.

Integrating unmanned aerial vehicles (UAVs) with federated learning (FL) has been seen as a promising paradigm for dealing with the massive amounts of data generated by intelligent devices. Nevertheless, although FL has natural advantages in data security protection, eavesdroppers can also deduce the raw data according to the shared parameters. Existing works mainly focused on encrypting the content of uploaded parameters, but we believe that it can improve security further by hiding the presence of parameter updating. Therefore, in this paper, we conceive a UAV-enabled covert federated learning architecture, where the UAV is not only responsible for orchestrating the operation of FL but also for emitting artificial noise (AN) to interfere with the eavesdropping of unintended users. To strike a balance between the security level and the training cost (including time overhead and energy consumption), we propose a distributed proximal policy optimization-based strategy for the sake of jointly optimizing the trajectory and AN transmitting power of the UAV, the CPU frequency, the transmitting power and the bandwidth allocation of the participated devices, as well as the needed accuracy of the local model. Furthermore, a series of experiments have been conducted to validate the effectiveness of our proposed scheme.

Recently deep neural networks (DNNs) are widely used in various fields. These intelligence applications, such as target recognition, are often computation-intensive and latency-sensitive. Since a single UAV's computing resource is limited, it is difficult to complete the DNN inference task independently. Partitioning the deep neural work into numerals subtasks and distributing them to multiple UAVs for collaborative computing seems a better way to finish the task. However, UAV usually works in a harsh environment, such as battlefield, disaster area, etc., and the link interruption or node failure in the inference process caused by uncertain factors may lead to failure of inference task. Hence, the reliability of DNN inference is of high importance. In this paper, we propose a deep Q learning-based DNN partitioning strategy for minimizing the energy consumption of DNN collaborative inference among multi-UAVs within latency and reliability requirements. To validate the effectiveness of the proposed strategy, a series of experiments are conducted on four kinds of typical DNNs (i.e., AlexNet, VGG19, GoogleNet, and ResNet). The simulation results prove the proposed strategy can effectively reduce the DNN inference cost under constraints.

In the face of deeply exploring and exploiting marine resources, the Internet of Underwater Things (IoUT) networks have drawn great attention considering its widely distributed low-cost and easy-deployment smart sensing nodes. However, given the hostile underwater environment, it is critical to conceive energy-efficient information collection because of limited underwater energy supply and inefficient artificial recharge methods. Characterized by high flexibility and maneuverability, autonomous underwater vehicles (AUVs) are regarded as a promising solution for information collection in the IoUT relying upon delicate AUVs’ trajectory and information collection strategy design with the spirit of balancing their energy consumption and information processing capability. In this article, we propose a heterogeneous AUV-aided information collection system with the aim of maximizing the energy efficiency of IoUT nodes taking into account AUV trajectory, resource allocation, and the Age of Information (AoI). Moreover, based on the particle swarm optimization (PSO), we obtain the trajectory of AUVs with low time complexity. Additionally, a two-stage joint optimization algorithm based on the Lyapunov optimization is constructed to strike a tradeoff between energy efficiency and system queue backlog iteratively. Finally, simulation results validate the effectiveness and superiority of our proposed strategy.

Autonomous underwater vehicles (AUVs) have been envisaged as a key enabler for empowering the Internet of Underwater Things (IoUT) networks to address the challenge of ever-increasing demand of ocean exploration. However, the energy constraint of AUVs' movement makes it challengeable to obtain full-space movement and extravagant information exchange considering complex underwater environment and hostile acoustic channel characteristics. For the sake of enhancing the sustainability of the power supply, it is significant to design a green underwater information collection scheme for beneficially utilizing the maneuverability of AUVs. In this paper, we propose a heterogeneous multi-AUV aided underwater information collection scheme for optimizing the unit energy consumption under the constraint of the age of information (AoI). Moreover, the limited service M/G/1 vacation queueing system is used to model the process of information exchange, where the steady-state distribution and the waiting time of queue are derived. Finally, simulation results show the effectiveness of our proposed scheme and low-complexity solution, which outperform single-AUV scheme in terms of both energy efficiency and AoI.

At present, the development and exploration of the ocean are blossoming, but the maritime communication coverage still remains limited. By deploying unmanned aerial vehicle (UAV) mounted relay nodes between shore base stations and vessel users, the off-shore communication coverage and transmission efficiency can be substantially enhanced. Considering the specific transmission characteristics of air-sea and of air-shore channels and time-varying traffic of maritime information services, we formulate a minimum-maximization optimization problem of link capacity, where both the deployment of UAV-mounted relay node and the configuration of communication resources are optimized. To address this non-convex problem, we propose a particle swarm based algorithm, which is capable of three-dimensional position, antenna direction and time slot allocation scheme joint optimization. The simulation results demonstrate the high efficiency and reliability of our proposed algorithm in diverse offshore relay scenarios with different coastal environments, vessel distributions and network traffic.

Unmanned aerial vehicles (UAVs) have been widely used to provide enhanced information coverage as well as relay services for Internet of Things (IoT). Constrained by the limited computation and battery capabilities of IoT devices, computational-intensive tasks are difficult to be tackled locally. In this paper, a UAV-aided mobile edge computing (UAMEC) system is constructed to assist IoT devices to tackle computational-intensive tasks. Furthermore, we jointly optimize communications, computing and caching resources allocation strategies for the sake of maximizing the net revenue from the UAMEC under the condition of guaranteeing user's quality of experience (QoE). A multi-dimensional hybrid adaptive particle swarm (MHAPSO) algorithm is conceived to solve this joint optimization problem. Finally, the effectiveness and superiority of our proposed scheme are demonstrated.

Internet of Vehicles (IoV) has drawn great interest recent years. Various IoV applications have emerged for improving the safety, efficiency, and comfort on the road. Cloud computing constitutes a popular technique for supporting delay-tolerant entertainment applications. However, for advanced latency-sensitive applications (e.g., auto/assisted driving and emergency failure management), cloud computing may result in excessive delay. Edge computing, which extends computing and storage capabilities to the edge of the network, emerges as an attractive technology. Therefore, to support these computationally intensive and latency-sensitive applications in IoVs, in this article, we integrate mobile-edge computing nodes (i.e., mobile vehicles) and fixed edge computing nodes (i.e., fixed road infrastructures) to provide low-latency computing services cooperatively. For better exploiting these heterogeneous edge computing resources, the concept of software-defined networking (SDN) and edge-computing-aided IoV (EC-SDIoV) is conceived. Moreover, in a complex and dynamic IoV environment, the outage of both processing nodes and communication links becomes inevitable, which may have life-threatening consequences. In order to ensure the completion with high reliability of latency-sensitive IoV services, we introduce both partial computation offloading and reliable task allocation with the reprocessing mechanism to EC-SDIoV. Since the optimization problem is nonconvex and NP-hard, a heuristic algorithm, fault-tolerant particle swarm optimization algorithm is designed for maximizing the reliability (FPSO-MR) with latency constraints. Performance evaluation results validate that the proposed scheme is indeed capable of reducing the latency as well as improving the reliability of the EC-SDIoV.

To achieve the goal of “Industrial 4.0,” cellular network with wide coverage has gradually become an intensely important carrier for industrial Internet of Things (IIoT). The fifth generation cellular network is expected to be a unifying network that may connect billions of IIoT devices for the sake of supporting advanced IIoT business. In order to realize wide and seamless information coverage, heterogeneous network architecture becomes a beneficial method, which can also improve the near-ceiling network capacity. In order to guarantee the quality of service (QoS) as well as the fairness of different IIoT devices with limited network resources, the network association in IIoT should be performed in a more intelligent manner. In this article, we propose a distributed Q-learning aided power allocation algorithm for two-layer heterogeneous IIoT networks. Moreover, we discuss the spirit of designing reward functions, followed by four delicately defined reward functions considering both the QoS of femtocell IoT user equipments and macrocell IoT user equipments and their fairness. Also, both fixed and dynamic learning rates and different kinds of multiagent cooperation modes are investigated. Finally, simulation results show the effectiveness and superiority of our proposed Q-learning based power allocation algorithm.

Due to the limited computing resources of swarm of drones, it is difficult to handle computation-intensive tasks locally, hence the cloud based computation offloading is widely adopted. However, for the business which requires low latency and high reliability, the cloud-based solution is not suitable, because of the slow response time caused by long distance data transmission. Therefore, to solve the problem mentioned above, in this paper, we introduce fog computing into swarm of drones (FCSD). Focusing on the latency and reliability sensitive business scenarios, the latency and reliability is constructed as the constraints of the optimization problem. And in order to enhance the practicality of the FCSD system, we formulate the energy consumption of FCSD as the optimization target function, to decrease the energy consumption as far as possible, under the premise of satisfying the latency and reliability requirements of the task. Furthermore, a heuristic algorithm based on genetic algorithm is designed to perform optimal task allocation in FCSD system. The simulation results validate that the proposed fog based computation offloading with the heuristic algorithm can complete the computing task effectively with the minimal energy consumption under the requirements of latency and reliability.

Industrial Internet of Things (IIoT) is a revolution which is changing the visage of industry in a profound manner. However, it brings many opportunities as well as many puzzles and challenges. Facing with billions of programmable IIoT devices, the traditional IIoT architecture based on cloud computing is no longer suitable, therefore, Mobile Edge Computing (MEC) has been seen as the promising technology to support IIoT business in 5G era. However, the existing mainstream MEC framework exposes numerous problems when supporting IIoT, such as complex development, low development reuse rate, poor software maintainability and mobility, poor flexibility, etc. Therefore, in order to solve the problems mentioned above, in this paper, we propose IIoT-MEC, a novel MEC framework specially for IIoT. We use Docker container to slice computing and storage resources of MEC server into numerous resource blocks (RBs). Based on the concept of virtualization, some RBs for “Device Function Virtualization (DFV)” are used to map physical devices into virtual devices and present in a set of normalized APIs, which shield the hardware development of diverse IIoT devices, so as to simplify the IIoT development into software development only. Some RBs are used to support the operation of IIoT services, in the form of distributed computing. On these basis, a flexible object-oriented IIoT development architecture is constructed. IIoT-MEC can overcome the drawbacks of the existing MEC framework in supporting IIoT. And the implementation procedure of IIoT-MEC is demonstrated with an application example. We also discuss how the IIoT-MEC would be used and what we need to do in future research.

Recently, the industrial Internet of Things (IIoT) has drawn high attention in academia and industry in the context of industry 4.0. In the IIoT, smart IoT devices are adopted to improve production efficiency. But, these devices will generate huge amounts of production data, which need to be processed effectively. To support IIoT services efficiently, cloud computing is usually considered as one of the possible solutions. However, the IIoT services still suffer from the high-latency and unreliable links problem between cloud and IIoT terminals. To combat these issues, fog computing is a promising solution which extends computing and storage to the network edge. In this paper, we are motivated to integrate the fog computing to the cloud-based IIoT to build a cloud-fog integrated IIoT (CF-IIoT) network. To achieve the ultra-low service response latency, we introduce the distributed computing to the CF-IIoT network and propose leveraging the real-coded genetic algorithm for constrained optimization problem(RCGA-CO) algorithm to optimize the load balancing problem of the distributed cloud-fog network. Most importantly, considering the unreliable situation in the CF-IIoT (e.g., fog nodes damage, wireless links outage), we propose a task reallocation and retransmission mechanism to reduce the average service latency of the CF-IIoT network architecture. The performance evaluation results validate that the RCGA-CO-based CF-IIoT and our proposed mechanism can provide ultra-low latency service in IIoT scenario.