The vision of 6G networks aims to enable edge inference by leveraging ubiquitously deployed artificial intelligence (AI) models, facilitating intelligent environmental perception for a wide range of applications. A critical operation in edge inference is for an edge node (EN) to aggregate multi-view sensory features extracted by distributed agents, thereby boosting perception accuracy. Over-the-air computing (AirComp) emerges as a promising technique for rapid feature aggregation by exploiting the waveform superposition property of analog-modulated signals, which is, however, incompatible with existing digital communication systems. Meanwhile, hybrid reconfigurable intelligent surface (RIS), a novel RIS architecture capable of simultaneous signal amplification and reflection, exhibits potential for enhancing AirComp. Therefore, this paper proposes a Hybrid RIS-aided Digital AirComp (HRD-AirComp) scheme, which employs vector quantization to map high-dimensional features into discrete codewords that are digitally modulated into symbols for wireless transmission. By judiciously adjusting the AirComp transceivers and hybrid RIS reflection to control signal superposition across agents, the EN can estimate the aggregated features from the received signals. To endow HRD-AirComp with a task-oriented design principle, we derive a surrogate function for inference accuracy that characterizes the impact of feature quantization and over-the-air aggregation. Based on this surrogate, we formulate an optimization problem targeting inference accuracy maximization, and develop an efficient algorithm to jointly optimize the quantization bit allocation, agent transmission coefficients, EN receiving beamforming, and hybrid RIS reflection beamforming. Experimental results demonstrate that the proposed HRD-AirComp outperforms state-of-the-art digital AirComp baselines in terms of both inference accuracy and uncertainty, achieving performance close to the idealized case with perfect feature aggregation.

6G confronts a paradigm shift towards integrated sensing, caching, computation and communication (SC3) networks, designed to render comprehensive information services and support diversified applications, in which Age of information (AoI) serves as a pivotal metric for evaluating the data freshness during the end-to-end information service procedure. However, existing works mainly focus on single-level AoI modeling, which fails to maintain fresh information in heterogeneous network infrastructure including autonomous aerial vehicles (AAVs) and ground access points (APs). Therefore, in this paper, we propose a AAV-enabled integrated SC3network model with bi-level AoI concept, where a AAV exploits common signals to collect sensory information from targets whilst updating the cached items of APs. Thereafter, we formulate a long-term optimization problem to coordinate bi-level AoI by jointly scheduling target sensing and caching updates, together with AAV trajectory and beamforming design. To tackle this intractable problem, we develop a deep reinforcement learning-based solution named SC3multi-domain resource allocation (SC3-MDRA). This algorithm innovatively incorporates hindsight experience replay and sharpness-aware minimization to overcome sparse reward as well as enhance policy adaptivity, thereby making immediate decisions in response to dynamic AoI status. Additionally, SC3-MDRA allocates computation and bandwidth resources of APs for effectively delivering information to requesting users. Experimental results reveal that the proposed SC3-MDRA outperforms baseline methods in terms of both learning convergence and system overall performance. Besides, the tradeoff between information freshness and AAV energy consumption is delineated.

6G networks are envisioned to support on-demand AI model downloading to accommodate diverse inference requirements of end users. By proactively caching models at edge nodes, users can retrieve the requested models with low latency for on-device AI inference. However, the substantial size of contemporary AI models poses significant challenges for edge caching under limited storage capacity, as well as for the concurrent delivery of heterogeneous models over wireless channels. To address these challenges, we propose a fine-grained AI model caching and downloading system that exploits parameter reusability, stemming from the common practice of fine-tuning task-specific models from a shared pre-trained model with frozen parameters. This system selectively caches model parameter blocks (PBs) at edge nodes, eliminating redundant storage of reusable parameters across different cached models. Additionally, it incorporates coordinated multipoint (CoMP) broadcasting to simultaneously deliver reusable PBs to multiple users, thereby enhancing downlink spectrum utilization. Under this arrangement, we formulate a model downloading delay minimization problem to jointly optimize PB caching, migration (among edge nodes), and broadcasting beamforming. To tackle this intractable problem, we develop a distributed multi-agent learning framework that enables edge nodes to explicitly learn mutual influence among their actions, thereby facilitating cooperation. Furthermore, a data augmentation approach is proposed to adaptively generate synthetic training samples through a predictive model, boosting sample efficiency and accelerating policy learning. Both theoretical analysis and simulation experiments validate the superior convergence performance of the proposed learning framework. Moreover, experimental results demonstrate that our scheme significantly reduces model downloading delay compared to benchmark methods.

Mobile Mdge Computing (MEC) has emerged as a crucial technology for supporting computation-intensive and latency-sensitive Internet of things (IoT) applications. Meanwhile, UAVs can serve as MEC servers, providing cost-effective computation offloading services to IoT terminals with their flexible deployment capabilities, especially in areas lacking ground infrastructure. Nevertheless, the computation offloading process suffers from potential security risks, while the randomness and uncertainty of terminals’ data sensing may exacerbate the queue backlogs. To tackle these challenges, a UAV-enabled secure aerial computing network that integrates MEC and blockchain is proposed, with the aim of jointly designing data sensing, offloading, and computing, together with UAV trajectory planning to maximize the long-term average data sensing rate under queuing delay and block creation delay constraints. To address the coupling between long-term constraints and short-term decisions, we apply Lyapunov optimization to decompose the original problem into three deterministic subproblems for each time slot. We then develop a multi-agent learning-based approach to collaboratively train terminal transmission power and UAV flight trajectories. Moreover, sensing rate and edge resource allocation are adaptively optimized in response to real-time data arrivals and queue backlogs. Simulation results demonstrate the superior performance of our solution, achieving over a 13.16% improvement in data sensing rate and more than a 29.47% reduction in queue delay compared to benchmark methods.

The exploitation of low-altitude (LA) airspace is advancing globally, boosting the sensing demands for heterogenous flying aircraft. To fulfill an accurate and intelligent sensing, the future 6G base station (BS) requires to aggregate features of multiple sensors’ views, then perform edge inference via loading the artificial intelligence (AI) model. However, this process confronts communication and computation bottlenecks owing to high-dimensional feature uploading as well as frequent memory access. To overcome these bottlenecks, we propose a multi-task over-the-air edge inference system for LA airspace, where feature aggregation is efficiently achieved employing over-the-air computation, and multiple inference tasks arriving at the BS are processed in batches to reduce memory access. Under this arrangement, we formulate a joint batching and beamforming design problem to maximize the number of completed tasks, constrained by completion latency and inference accuracy requirements. To address this intractable problem, we first examine the case with synchronous task arrivals and single batch. A spatial correlation-aware beamforming design approach is proposed to effectively suppress feature aggregation error and ensure inference accuracy. Next, we delve into the general asynchronous task arrival case. An AI-generated online policy is developed, which innovatively utilizes diffusion model to output batching decisions, thereby adapting to the dynamic and uncertain nature of task arrivals. Simulation results obtained on real-world dataset corroborate the importance of capturing the spatial correlation among sensors. In addition, the proposed approach realizes outstanding performance compared to benchmark batching, beamforming, and learning methods, and the completed task amount is close to offline optimization with prior task information.

Air-ground integrated network is able to make up for the limitations of small coverage as well as fixed resource deployment of ground 5G network and provide flexible services via edge computing. However, duplicated data may be offloaded to edge server, resulting in waste of network resources. In the meantime, due to the limited on-board energy storage of unmanned aerial vehicle (UAV), the endurance and trajectory optimization of UAV need to be focused on. Moreover, the high dynamics of network nodes will lead to the dilemma of information uncertainty and curse of dimensionality. In order to tackle the above challenges, we design a UAV-assisted heterogeneous cache-computing network model, among which, devices are divided into two categories: the request ground device (RGD) which requires task offloading and the free ground device (FGD) which can process offloaded tasks from the RGD through the device-to-device (D2D) link. Then we formulate a problem of jointly optimizing cache strategy, task segmentation, computing resource allocation and UAV trajectory planning to minimize the system latency constrained by UAV energy recharging. Due to the coexistence of discrete and continuous variables as well as coupling between long-term energy constraint and short-term decision making, we decompose it into two sub-problems. Low complexity matching algorithm is used to select the optimal UAV task caching strategy, while Lyapunov optimization is leveraged to decompose the second sub-problem, for which CVX is utilized to solve the task segmentation and local computing resource allocation, and soft actor-critic (SAC) approach is used to realize the computing resource allocation of FGD and UAV plus trajectory planning. Extensive simulation results showcase the advantages of our algorithm in reducing the system latency compared to benchmarks.

The global aging process accelerates, making geriatric disease prevention and management urgent. Combining IoMT with wearable devices for health monitoring is effective, but IoMT task offloading has challenges like weak terminal processing, poor battery life, low - latency needs, and privacy protection difficulties. To address these challenges, this paper proposes a Cloud-edge Collaborative Air-ground Integrated IoMT Network (C2AI2N) that combines ground Base Stations, air-based Autonomous Aerial Vehicles (AAVs) and cloud servers to adapt to future IoMT scenarios and enhance the system capabilities. Edge servers are able to use Wireless Power Transfer (WPT) to provide energy to wearable devices. We then consider the Ultra-Reliable Low Latency Communication (URLLC) task constraints as well as power limitations, and formulate the problem of minimizing the overall system energy consumption. Next, we employ the Lyapunov optimization method to break down the problem into three sub-problems: 1) device-side task splitting; 2) task offloading as well as charging strategy; and 3) server-side resource allocation, which are addressed via Lagrange multiplier method, Clustered Asynchronous Federated Learning (CAFL) based rainbow Deep Q-Network (DQN) approach, and the coati optimization algorithm, respectively. Extensive simulations demonstrate that the proposed method outperforms other baselines. Specifically, the energy consumption of the proposed method can be reduced by over 30.37% compared to the baselines. In addition, the backlog of the URLLC virtual queue can be decreased by more than 59.27%, achieving superior URLLC satisfaction rates.

The deep integration of sensing, communication, caching, and computation (SC3) is emerging as a key feature of 6G network, designed to support ubiquitous intelligent applications and enhance human quality of life. Simultaneously, unmanned aerial vehicles (UAVs) have been identified as promising edge nodes to bolster terrestrial wireless networks. To harness the coordination benefits of SC3 and address potential conflicts, we propose a UAV-empowered joint SC3 6G network, in which UAVs are outfitted with edge servers to cache and process sensing data from integrated sensing and communication devices before delivering the results to users. To maintain the freshness of sensing data in such network, we formulate an average information retrieving delay minimization problem, coordinating cross-domain resources while considering performance constraints in sensing, communication, and energy. We then develop a cross-domain resource optimization algorithm to jointly design UAV 3D deployment, subcarrier assignment, power control, caching update, and computational resource allocation. This approach combines block coordinate descent, matching theory, and successive convex approximation to iteratively solve the optimization problem. Experimental evaluations demonstrate that the proposed scheme converges rapidly and outperforms benchmark methods in reducing average information retrieving delay through the coordination of SC3 cross-domain resources.

Accompanied by the construction of new power system with renewable energy, like the offshore wind power and desert photovoltaic power, terrestrial ground 5G is no longer able to fulfill the communication requirement of industrial power IoT (IPIoT) with tremendous equipment in remote areas. Under this circumstances, we put forward the NOMA-enabled space–air–ground integrated IPIoT network (SAGIN-IPIoT) model, with a satellite to achieve wide coverage, and multiple UAVs to strengthen hot spot communication. NOMA is leveraged to improve system throughput by allowing the common frequency resource shared among multiple users. An energy-efficient (EE) maximization problem is formulated to jointly optimize subchannel and terminal power. However, since the objective involves both continuous and binary variables, it is a mixed integer nonlinear programming (MINLP) issue. Thus, we decompose it into two subproblems, which are respectively solved by matching game and Lagrange dual method with low complexity. According to the theoretical analysis and simulations, we can conclude that the method has better performance than the benchmark method.

With the rapid development of 5G and Internet of Vehicles (IoV) technologies, vehicles have evolved from mere transportation devices to mobile living spaces for humans. Thus, the complexity of data processing is growing along with the increasing of vehicle applications, which makes relying solely on on-board processing capabilities insufficient. Traditional Roadside Units (RSUs)-based edge computing has limitations such as high deployment cost and finite coverage. To address those issues, we construct an Air-Ground Vehicular Cooperative Computing Network (AVC$^{\textbf{2}}$N) that introduces Cooperative Vehicles (CVs) to reduce deployment cost while incorporating Unmanned Aerial Vehicles (UAVs) to expand communication coverage. We aim to balance offloading efficiency and environmental sustainability by proposing a system cost minimization problem with weighted sum of delay and energy consumption. However, it faces new challenges such as the lack of Global State Information (GSI), and the Ultra-Reliable Low-Latency Communications (URLLC) queue delay constraints, rendering traditional methods inadequate. To address the coupling between immediate decision and long-term queuing constraints, we employ Lyapunov optimization to partition the initial problem into two distinct sub-problems. The first focuses on optimizing the system transmission cost, which is tackled using a collaborative Deep Reinforcement Learning (DRL) framework. Specifically, we design an algorithm based on Multi Agent Deep Deterministic Policy Gradient (MADDPG), which effectively addresses GSI uncertainty and ensures URLLC awareness. The second sub-problem addresses server-side computing resource optimization, and we propose a greedy algorithm to tackle it. Experimental results showcase the effectiveness of our approach, demonstrating notable achievement in terms of learning convergence speed, overall system cost, queue delay, and queue backlog.

For the two new functions that will be deployed at 6G wireless network edge, namely radio sensing and artificial intelligence (AI) inference, it is challenging to simultaneously boost the dual-functional performance due to their intrinsic conflict and resource competition. Meanwhile, to promote ubiquitous connectivity in 6G, unmanned aerial vehicles (UAVs) serve as representative aerial platforms for service enhancement. This paper investigates a UAV-enabled joint AI inference and target tracking system, where multiple terminals transmit local features to the UAV for edge inference, while the UAV is able to track mobile target by emitting sensing signals. Although these two functions are performed in a time division manner, we propose a contribution-based feature selection and transmission control strategy, which enables the terminals to transmit fewer features to achieve the desired inference accuracy, thereby releasing more resources for target tracking. Additionally, a tracking Cramér-Rao bound (CRB) minimization problem is established to jointly optimize the UAV trajectory and beamforming, under the constraints of AI inference uncertainty requirements. To tackle this intractable issue, we develop a hybrid offline-online approach which optimizes an offline trajectory to ensure statistically favorable feature transmission, followed by an online adaptation that adjusts the UAV velocity and beamforming exploiting the real-time estimated target state. Simulation results corroborate the outstanding performance of our approach compared to benchmark methods. Besides, interesting tradeoffs resulted from the hybrid design are elucidated.

Through deploying satellites and unmanned aerial vehicles (UAVs) with onboard processing capability, the space-air-ground edge computing network (SAGECN) is poised to support ubiquitous access and computation offloading for Internet of Things (IoT) terminals deployed in remote areas. However, the current SAGECN faces several challenges in realizing its full potential, such as scarce spectrum resources, diverse computational demands, and dynamic network circumstances. To meet these challenges, we propose a cluster-non-orthogonal multiple access (C-NOMA)-enabled SAGECN model, where a satellite and multiple UAVs act as collaborative edge servers to execute tasks from IoT terminals. Since each offloaded task should be processed via a specific program, the edge servers carry out program caching, whilst transfer the tasks that do not match the cached programs to another server in a multi-hop manner. Considering the delay-sensitive requirements of computation tasks, we formulate a joint task offloading, communication-computation-cache resource assignment, and routing plan problem, aimed at minimizing the average system latency. To cope with this challenging issue, we partition it into three subproblems. First, a multi-agent learning-based approach is developed to collaboratively train the task offloading, flight trajectory, and program caching. As a step further, two optimization subroutines are embedded to perform routing plan, subchannel allocation, and power control, thereby rendering the overall solution. Experimental results reveal that our approach achieves outstanding performance in terms of system delay and spectrum efficiency.

Since the existing terrestrial fifth generation (5G) network has limited coverage, it is difficult to meet the growing demand for seamless network connection. Meanwhile, current network resource allocation methods mainly research on how to improve system performance only from the perspective of resource utilization, but rarely take users’ specific needs for network content into consideration. This brings severe challenges to efficient network service and flexible resource allocation. Therefore, we construct the content service-oriented resource allocation model for space–air–ground integrated sixth generation networks (SAGIN 6G), and formulate the three-sided matching issue among the space–air–ground integrated network equipment (SAGINE), content sources, and users. In this model, users request to establish connection with SAGIN which forwards users’ request to content service provider (CSP). CSP manages the creation of content data, and finally returns the requested content to users through SAGIN. For the content service-oriented resource allocation in SAGIN, finding the optimal stable three-sided matching with the largest cardinality is an NP-complete problem. Therefore, to efficiently solve the above issue, we design some reasonable restrictions and convert it to a restricted three-sided matching problem with size and cycle preferences. We further develop the content-oriented resource allocation algorithm (COR2A) and the user-oriented resource allocation algorithm (UOR2A) in a distributed manner. Extensive simulations verify our approach outperforms traditional benchmark resource allocation schemes in terms of system throughput, CSP revenue, and user experience.

UAV-based air-ground integrated computing networks (AGIN) have gained significant traction in remote areas for the Power Internet of Things (PIoT). This paper considers an AGIN-PIoT, where computing tasks generated by ground PIoT devices are offloaded to aerial UAVs that perform edge computing. Jointly optimizing task offloading and UAV trajectory poses challenges such as many decision variables, information uncertainty, and long-term queue delay constraints. Due to the limited battery capacity of PIoT devices and UAVs, our objective is to minimize system energy consumption under long-term queue delay constraints by jointly optimizing task offloading, trajectory planning, and computing resource assignment. In light of Lyapunov optimization, we decompose the original challenging optimization problem into two sub-problems: (1) task offloading and UAV trajectory planning and (2) aerial edge resource allocation. Accordingly, we develop a multi-agent deep reinforcement learning-based algorithm called AGIN-MADDPG for the former to achieve the maximum accumulative reward and propose a greedy solution for the latter. Extensive experiments and numerical results demonstrate that our approach can avoid the problem of gradient vanishing and outperforms other benchmark methods in terms of power consumption, task backlog, queue delay, and system throughput.

As the fast development of vehicular network, the layout of the roadside unit (RSU) is indispensable. Due to the shackles of factors, such as coverage and cost, there is an urgent need for effective solution to solve the contradiction that RSU cannot be deployed on large scale. Parked cars provide a feasible solution for replacing RSUs and effectively reducing the arrangement of edge nodes. Inspired by this, parked cars as RSUs (P-RSUs) are leveraged to support cities’ vehicular network in this article. We first construct the P-RSU-based cellular-V2V heterogeneous networks (C-V2V HetNets) system model, and then formulate an optimization problem to maximize the energy efficiency (EE) of C-V2V HetNets with parked cars. Since the proposed issue is an NP-hard mixed-integer nonlinear programming (MINLP) problem coupled with P-RSU incentive, we reformulate it into two subproblems, which are the P-RSU recruitment and the joint resource allocation. For the first subproblem, an effective reverse auction-based mechanism is given to encourage parked cars participate and become P-RSUs. For the second subproblem, nonlinear fractional programming is used to optimize transmission power, and many-to-one matching is utilized to effectively obtain channel reusing scheme constrained by QoS. Moreover, a multihop-based transmission strategy is given to further expand vehicular network coverage. Algorithms are evaluated based on real-world scenarios using SUMO. Numerical results demonstrate that the proposed approach can both effectively recruit P-RSUs with low cost and achieve excellent system performance in terms of EE, spectrum efficiency, and network coverage compared to other benchmark algorithms.

In this article, with the aim to enhance the secure transmission and improve the utilization of spectrum resources in Internet of Things (IoT), a multicarrier nonorthogonal multiple access (MC-NOMA)-assisted full duplex (FD) network is investigated, in which nonorthogonal multiple access (NOMA) is implemented in both uplink and downlink transmissions. The lightweight and low-power physical layer security (PLS) technology is employed to protect the information from eavesdropping. Taking the imperfect channel state information (CSI) into account, we formulate a problem to optimize the beamforming vector, artificial noise (AN), transmit power, and subcarrier assignment policy aiming to maximize the worst case sum secrecy rate under the Quality of Service (QoS) and power consumption constraints. Since the formulated problem is nonconvex and difficult to be solved, we decompose it into two joint optimization subproblems. The first is resource allocation with given subcarrier assignment, which is solved by using the block coordinate descent (BCD) approach. The second is subcarrier assignment solved by the matching theory. Our simulation shows that the proposed scheme is robust against the CSI imperfectness of the eavesdropping and self-interference channels, while providing significant sum secrecy rate improvement compared with the orthogonal multiple access (OMA), half duplex (HD) systems, and other benchmark schemes.

By combining information communication technology with power grid, the smart grid-oriented Power Internet of Things (PIoT) has become a critical technology to guarantee the safe and reliable power grid operation and improve system energy efficiency. Nevertheless, PIoT devices have only limited communication and computing resources since they are mostly deployed in remote areas that may be out of service coverage of existing terrestrial 5G networks. To overcome the resource limitation, we leverage Air-Ground Integrated C-NOMA Heterogeneous PIoT Networks (PAGIC HetNets), and study the core challenges in PAGIC HetNets. As PIoT devices are normally powered by battery, we aim at minimizing the energy consumption of PIoT devices and thoroughly investigate the problem of task offloading and resource allocation with minimal energy consumption. This problem belongs to a mixed integer nonlinear programming (MINLP) with extra difficulty that the long-term queuing delay and short-term constraints are coupled. To tackle the difficulty, we use Lyapunov optimization to transform this hard problem into three subproblems. The first subproblem is task splitting and local computing resource assignment at the PAGIC user side, which we solve with the Lagrangian multiplier method. The second subproblem is queue-aware channel reusing, and matching theory is adopted to solve it. The third subproblem is optimizing the aerial server resource allocation, for which we propose a greedy-based solution. Numerical simulations demonstrate that our approach can obtain excellent performance in terms of energy consumption, spectrum efficiency, task backlog, and queuing delay with lower complexity compared with several benchmark methods.

Abstract In this work, an artificial neural network (ANN) based time‐varying channel modeling framework is proposed, including a playback model and a prediction model. The purpose of the ANN‐based modeling framework is to playback 5G measured radio channels at certain measurement positions, and further predict large scale channel parameters (LSCPs) at unmeasured positions with limited amount of measurement data. 28 GHz channel measurements were also conducted at a high‐voltage substation for the first time worldwide to meet with 5G radio system deployment for China Energy Internet. Meanwhile, the performance of the playback channels is evaluated by comparison with the measurements and traditional geometry based stochastic modeling (GBSM) simulated channels. An optimized radial basis function (ORBF) ANN is applied in the prediction model, and the predicted LSCPs are compared with the other approaches, which shows that the ORBF has the best performance. This work offers a solution to predict radio channels and parameters in case of big measured or simulated channel datasets.