In this paper, we investigate the power control and spectrum allocation challenge in vehicle-to-everything (V2X) networks with uncertain channels, where two types of services coexist, i.e., the large-capacity services supported by vehicle-to-infrastructure (V2I) links and the ultra-reliable services supported by vehicle-to-vehicle (V2V) links. Our goal is to maximize the transmission rate of V2I whilst ensuring the V2V outage probability constraint. To deal with the channel uncertainty, we first mine the channel correlation by mapping historical channel samples into high-dimensional space via a designed deep neural network (DNN). Then, a support vector clustering (SVC) based uncertainty set is derived as a union of convex subsets according to the output of piecewise activation DNN. Afterwards, a power control algorithm is proposed for the transformed non-convex problem. Specifically, a bisection-based method is developed to tackle the fractional signal-to-interference-plus-noise ratio (SINR) expression, and an exploration-comparison-approximation method is proposed to relax the union of convex subsets. Additionally, we design a set of virtual V2V links to enforce the spectrum allocation problem into a feasible bipartite graph matching problem. Simulation results demonstrate the proposed method overcomes the conservatism of previous uncertainty set construction methods, and outperforms other methods on V2I transmission rate.

The rapid development of 6G makes space-air-ground integrated networks (SAGIN) a promising solution to the coverage and capacity limitations of traditional cellular systems. However, time-varying topologies, stochastic channels, imbalanced user demands, and limited resources hinder on-demand service provisioning in wide-area environments. To address this challenge, this paper proposes an on-demand service framework that prioritizes users who contribute greater system utility once their demands are satisfied. The objective is to improve system utility through on-demand services without requiring prior knowledge of user demands or channel statistics. We first develop an on-demand utility model that captures diminishing returns in demand satisfaction while incorporating heterogeneous priority levels and latency constraints. Based on this model, we formulate a joint on-demand resource allocation and task offloading problem (ODRA-TO) to maximize system utility under long-term queue stability constraints. To efficiently solve ODRA-TO, we design an alternating direction method with three-stage iterations (ADMI) that decomposes the problem into learning-assisted task offloading, swap-stable matching based subchannel assignment, and gradient-based successive convex approximation for power control. Simulation results show that ADMI reduces the average queue length by 44.18%, improves on-demand utility by 34.59%, and achieves a 99.00% completion rate for high-priority services under dynamic SAGIN conditions.

Integrated Sensing, Communication, and Computation (ISCC) has the potential to meet diverse requirements of Internet of Vehicles (IoV), such as high reliability and low power consumption. However, existing works have not fully considered the problems of unreliable communication links and inefficient data processing under resource constraints in non-ideal environments. To address these issues, this paper proposes a predictive Integrated Sensing, Communication and Computation Over-the-Air (ISCCO) approach based on Orthogonal Time Frequency Space (OTFS) modulation. It takes high Doppler shifts, network dynamics, and resource constraints into account. In particular, the Road Side Unit (RSU) performs target tracking while communicating with the downlink users through Space Division Multiplexing (SDM), and receives the transmission results of the uplink. For the downlink, a predictive beamforming approach based on Extended Kalman Filtering (EKF) is employed, while Over-the-Air computation (AirComp) is utilized for the uplink. The transmit power and receive beamformer at the RSU, along with the transmit power of the uplink users, are jointly optimized through two formulated optimization problems: sensing performance maximization and power consumption minimization. To solve these problems, we adopt an Alternating Optimization (AO)-based algorithm for finding the local optimal solution. Simulation results validate the effectiveness of the AO-based algorithm, and the analysis of the trade-offs between multi-dimensional performance of ISCC and power consumption is conducted.

With the rapid development of Urban Air Mobility (UAM), high-precision positioning of Electric Vertical Take-Off and Landing (eVTOL) vehicles, core aerial nodes of Low-Altitude Wireless Networks (LAWN), has become key to LAWN's dynamic cooperative operation and UAM safety. Traditional satellite positioning (SP) and base station positioning are vulnerable to urban canyon blockage, causing interrupted position sensing among LAWN nodes. Though cooperative sensing technology based on on-board sensors and Vehicle-to-Everything (V2X) communications is a potential solution for LAWN positioning, existing LiDAR-based cooperative positioning schemes require sharing raw or processed point cloud data, which conflicts with LAWN's limited-bandwidth V2X communications and restricts multi-eVTOL cooperative efficiency. To address these eVTOL positioning challenges in LAWN, this paper proposes a coopera tive sensing positioning strategy integrating V2X communications and LiDAR: it acquires adjacent eVTOLs' relative positions via LiDAR, combines positioning information from base stations and satellites, and builds a lightweight data transmission framework to avoid point cloud sharing. Additionally, a proximity-based identification algorithm is designed to resolve multi-eVTOL identity discrimination in LAWN, ensuring reliable information matching among LAWN nodes. Simulation results show the method's minimum positioning error is 0.2 dm, meeting LAWN's high-precision positioning requirements for eVTOLs and provid ing technical support for LAWN to enable safe UAM operation.

This paper investigates the power and spectrum allocation problem in the vehicle-to-everything (V2X) network with uncertain channels. We consider the scenario where the high-capacity services supported by vehicle-to-infrastructure (V2I) links and the ultra-reliability services supported by vehicle-to-vehicle (V2V) links coexist. Our goal is to maximize the V2I transmission rate whilst satisfying the V2V outage probability constraint. To handle the challenge brought by the outage probability constraint with uncertain channels, we first apply conditional value at risk (CVaR) theory to transform the outage probability constraint into an expected expression related to the uncertain channel distribution. Then, a deep neural network (DNN) based method is proposed to approximate this expectation to a deterministic value. After that, we propose an actor-double-critic based deep-deterministic-policy-gradient (ADC-DDPG) network architecture to obtain the optimal power allocation decision. The architecture includes a power allocation actor module (PAA), a transmission rate maximization critic module (TRC), and an outage probability calculation critic module (OPC). Then, we propose a dual reward feedback deep reinforcement learning algorithm to train the proposed network architecture. Moreover, a virtual V2V link set is developed to transform the spectrum allocation problem into an easily-tackled standard binary matching form. Finally, the simulation results illustrate that the proposed algorithm ensures the ultra-reliability requirement of V2V links, and outperforms the traditional deep deterministic policy gradient (DDPG) algorithm in terms of V2I transmission rate.

As connected and automated vehicles (CAVs) are gradually introduced into connected-vehicle environments, CAVs and connected vehicles (CVs) may coexist in local vehicular networks, where CAVs can assist CVs by providing perception information, service-node support, and processing capability for vehicular tasks. Such cooperation requires task scheduling and resource allocation to be jointly coordinated under heterogeneous vehicle capabilities and time-varying service opportunities. However, existing joint optimization approaches for such vehicular task scheduling and resource-allocation problems either exhibit excessive coupling, resulting in high computational complexity and limited scalability, or adopt overly decoupled structures that may degrade decision quality. To address this dilemma, we propose a nested framework that bridges monolithic and hierarchical coupling by embedding task scheduling within high-quality resource-allocation solution sets. The framework first screens feasible service-node and resource combinations and then performs task scheduling over the selected solution space, thereby balancing decision coupling and computational efficiency. We further develop a Sequence-to-Sequence Proximal Policy Optimization (S2S-PPO) algorithm to solve the resulting sequential decision problems. Simulation results show that the proposed framework can reduce task latency, maintain solution quality close to the exact optimal baseline in small-scale cases, and exhibit stable scalability trends under expanded network scales and different road-topology settings. These results demonstrate the effectiveness of the proposed nested design under the considered mixed traffic scenarios.

Enhancing safety and efficiency in connected and autonomous vehicles (CAVs) is a paramount concern, necessitating advanced sensing, communication and computing solutions. Despite the potential of utilizing the sensors deployed on roadside infrastructure to improve driving reliability, conflicts between heterogeneous needs of CAVs and real-time available resources present significant challenges. To address these, this paper introduces an innovative intelligence-guided reinforcement learning (IRL) framework designed to improve the the reliability and effectiveness of sensing information processing and transmission. Firstly, we propose a comprehensive road sensor networks (RSNs) framework that considers the heterogeneous sensing requirements of CAVs and available communication and computing resources, ensuring the reliability of sensing assistance provided to CAVs across varied and complex driving environments. Furthermore, we formulate the sensing information processing and transmission issue as an active inference to construct higher-level cognition about the present environment without relying on rewards. To address this, an IRL-guided method is implemented. “Intelligence” as a metric is leveraged for guiding the learning process and improving adaptability in diverse driving environments. Through extensive simulations, our proposal demonstrates superior performance over existing approaches, showing marked improvements in detection accuracy, latency reduction, and efficient data handling across various traffic scenarios.

In the high-mobility vehicle-to-everything (V2X) network, guaranteeing the reliability of data transmission under uncertain channels faces great challenges. In addition, how to allocate the limited wireless resource to achieve the coexistence of multiple types of services in V2X networks needs to be investigated, i.e., the large-capacity services supported by vehicle-to-infrastructure (V2I) links and the ultra-reliability services supported by vehicle-to-vehicle (V2V) links. From the age of information (AoI) perspective of V2V links, we design a robust resource allocation method to maximize the V2I transmission rate. First, a closed-form expression of the time-average AoI is derived. Then, we further transform the AoI requirement into an outage probability constraint. To describe the distribution of uncertain channels, on the one hand, an ambiguous set constructed method is proposed based on the first and second-order statistical knowledge from historical uncertain channel samples. On the other hand, a feasible region tightening method is proposed according to the ellipsoidal expression. Afterwards, a midpoint-search-driven method is designed to solve the transformed power allocation problem. Then, a virtual V2V link set is proposed for the spectrum allocation problem. In the end, we confirm that the proposed distributionally robust algorithm outperforms others through simulation.

The sixth generation (6G) wireless networks are envisioned to feature wide-area coverage, diversified full-scenario services, massive connections and dynamic heterogeneity, resulting in large-scale and complex network optimization problems. Traditional model-based methods, while effective in simple scenarios with precise mathematical models, struggle with high computational intensity and long processing times in the realistic and intricate applications of 6G. Pure data-driven deep learning (DL) methods offer powerful approximation capabilities and fast online inference but are hindered by insufficient datasets and poor interpretability. To address these issues, knowledge-driven DL integrates domain knowledge into neural networks, combining the strengths of both model-based and data-driven approaches. This survey systematically reviews knowledge-driven DL in wireless networks from a novel perspective of the knowledge integration approach. It provides a comprehensive definition of domain knowledge in wireless networks and clarifies the types of knowledge and their representations that can be integrated into neural networks. Furthermore, a leading taxonomy of knowledge integration approaches in wireless networks is proposed, encompassing the integration of domain knowledge into neural network model selection, neural network model customization, knowledge and data fusion architecture construction, loss function design, and hyperparameter configuration. Based on this taxonomy, literature on knowledge-driven resource allocation and signal processing is thoroughly reviewed. This survey aims to provide an insightful guideline for effectively incorporating domain knowledge into neural networks in the field of wireless communications, ultimately advancing efficient and reliable intelligent 6G networks.

Urban Air Mobility (UAM) has emerged as a transformative solution to alleviate urban congestion by utilizing low-altitude airspace, thereby reducing pressure on ground transportation networks. To enable truly efficient and seamless door-to-door travel experiences, UAM requires close integration with existing ground transportation infrastructure. However, current research on optimal integrated routing strategies for passengers in air-ground mobility systems remains limited, with a lack of systematic exploration.To address this gap, we first propose a unified optimization model that integrates strategy selection for both air and ground transportation. This model captures the dynamic characteristics of multimodal transport networks and incorporates real-time traffic conditions alongside passenger decision-making behavior. Building on this model, we propose a Unified Air-Ground Mobility Coordination (UAGMC) framework, which leverages deep reinforcement learning (RL) and Vehicle-to-Everything (V2X) communication to optimize vertiport selection and dynamically plan air taxi routes. Experimental results demonstrate that UAGMC achieves a 34% reduction in average travel time compared to conventional proportional allocation methods, enhancing overall travel efficiency and providing novel insights into the integration and optimization of multimodal transportation systems. This work lays a solid foundation for advancing intelligent urban mobility solutions through the coordination of air and ground transportation modes. The related code can be found at https://github.com/Traffic-Alpha/UAGMC.

Accurate Channel State Information is prerequisite for intelligent sensing and ubiquitous connectivity. However, the diversity of channel conditions-from sparse to dense and static to fast-varying-fundamentally challenges traditional single and fixed estimation algorithms. To address this issue, this paper proposes a Physics-Prioritized Mixture of Experts (PP-MoE) scheme, leveraging the MoE paradigm's ability to allocate resources to specialized experts tailored for distinct physical environments. The proposed scheme features an innovative heterogeneous expert library, where the architecture of each expert is customized with embedded physical priors to match its specific propagation environment. To enable intelligent scheduling, we design a hybrid decision gating network that collaboratively leverages physical formula computation and data-driven deep learning to achieve accurate channel environment identification and expert routing. Furthermore, to overcome the expert collapse problem, we propose a three-stage training strategy-pretraining, freezing, and fine-tuning to ensure training stability specialization. Extensive simulations demonstrate that PP-MoE significantly outperforms traditional and deep learning baselines. Notably, in the low-SNR region (0-15 dB), it achieves an NMSE nearly an order of magnitude lower than LMMSE. Additionally, PP-MoE maintains high efficiency with only 0.0256 GFLOPs. This work provides an effective paradigm for designing adaptive and physically reliable wireless physical layers.

The Internet of Agents environment imposes high computational demands and requires dynamic Quality of Service (QoS) for AIGC tasks. To address these challenges, this paper proposes an intelligent computational offloading strategy based on PPO. First, a two-dimensional joint action space encompassing offloading location decisions and generation quality level selection was constructed to accommodate the scalable characteristics of AIGC tasks. Second, a Quality of Experience reward function integrating user preference modes is designed. Furthermore, to address training instability caused by dimensional discrepancies in the multidimensional state space, a dual online normalization mechanism for states and rewards is introduced. This is combined with a dynamic learning rate scheduling strategy to enhance the algorithm's feature extraction efficiency in complex environments. Simulation results demonstrate that the proposed method significantly outperforms baseline algorithms in terms of convergence speed and stability.

In vehicular cooperative perception (CP), numerous resource allocation strategies have been proposed to enhance urban autonomous driving. However, existing studies often overlook the competition between self-perception and cooperative perception, where degrading a ground vehicle's (GV's) self-perception may introduce safety risks and reduce passenger comfort. Moreover, air-ground cooperation-which can improve sensing precision, reduce task execution delay, and enhance CP service availability-has received limited attention. It is worth noting that autonomous aerial vehicles are unavailable for cooperative perception during the recharging process.To address these issues, this paper investigates on-demand scheduling strategy in air-ground cooperative perception. At the millisecond timescale, resource competition is considered in real-time sensing, communication, and computation (SC$<^>{2}$2) resource allocation. At the minute timescale, the idle flying period between consecutive tasks is utilized for autonomous aerial vehicles' recharging through attachment to GVs along the route. Specifically, we first develop a model that captures the mutual influence between autonomous aerial vehicles and GVs on perception performance under resource constraints. Then, a mixed-timescale solution is proposed: at the small timescale, a multi-agent deep reinforcement learning algorithm with gradient-free projection and auxiliary supervision is designed to schedule SC2 resources; at the large timescale, a Hungarian-based algorithm is employed to control autonomous aerial vehicles' recharging. Simulation results show that the proposed approach outperforms benchmark schemes by reducing task execution delay and energy consumption, and enhancing CP service availability, while satisfying sensing precision, GV safety, and passenger comfort requirements.

In recent years, Unmanned Aerial Vehicles (UAVs) have become an indispensable component of modern emergency response systems in urban areas. Leveraging their flexibility, rapid response capabilities, and immunity to ground traffic constraints, UAVs exhibit significant advantages in complex urban environments. In particular, during natural disasters or emergencies, UAVs can swiftly reach the scene, providing real-time data support, which greatly enhances rescue efficiency and decision-making accuracy. This paper focuses on the role of UAVs within Wireless Rechargeable Sensor Networks (WRSNs) in urban settings. We establish a city scenario where UAVs employ Non-Line-of-Sight (NLOS) and Line-of-Sight (LOS) propagation models to perform three-dimensional wireless charging for two-dimensionally distributed sensor nodes, with charging occurring only during hover states. The objective is to design a flight path that minimizes UAV energy consumption while ensuring all sensor nodes' charging requirements are met. Addressing this challenge involves solving complex trade-offs, especially balancing UAV charging efficiency with mobility energy expenditure, which constitutes an NP-hard problem. To tackle this issue, we propose a Radius Expansion Limitation Algorithm to calculate optimal charging positions and solve the Traveling Salesman Problem (TSP) to optimize the UAV's trajectory. Extensive evaluations demonstrate that our proposed path planning scheme significantly improves UAV energy utilization efficiency, offering an efficient and reliable solution for urban emergency response.

The dynamic nature of vehicular ad-hoc networks (VANETs) presents significant challenges, including computational constraints, decision conflicts, and performance degradation under environmental dynamics. Traditional offloading strategies often struggle to balance real-time responsiveness, energy efficiency, and knowledge retention in heterogeneous vehicle clusters. To address these issues, this paper proposes a federated continual reinforcement learning framework with asynchronous updates and spatiotemporal knowledge fusion. An optimized multi-agent deep deterministic policy gradient algorithm with lazy agent updates (OMADDPG-LA) is introduced to reduce policy conflicts while improving communication efficiency. Additionally, a spatiotemporally decoupled knowledge retention mechanism adjusts parameter importance weights based on vehicle-specific behaviors and incorporates generative replay to alleviate storage constraints. A federated aggregation paradigm with credibility-aware weighting further enhances adaptability to dynamic tasks. Experimental results show that the proposed framework significantly improves information freshness, energy efficiency, and communication overhead compared to existing methods. It also achieves better convergence stability under dynamic conditions, providing a privacy-preserving and scalable solution for collaborative intelligence in large-scale VANETs.

Low Earth Orbit (LEO) satellite networks are integral to next-generation communication systems, providing global coverage, low latency, and minimal signal loss. However, their unique characteristics, such as constrained onboard resources, Line-of-Sight (LoS) propagation, and vulnerability to eavesdropping over wide coverage areas, present significant challenges to physical layer security. To address these challenges, this paper focuses on the design of anti-intercept waveforms for satellite-ground links within Orthogonal Frequency Division Multiplexing (OFDM) systems, aiming to enhance security against eavesdropping threats. We formulate a secrecy rate maximization problem that aims to balance secrecy performance and communication reliability under eavesdropping constraints and sub-carrier power limitations. To solve this non-convex optimization problem, we propose a bisection search-activated neural network (BSA-Net) that integrates unsupervised learning for secure coding optimization and bisection search for dynamic power allocation. The proposed method is structured in two stages: the first optimizes secure coding under power constraints, while the second allocates power across sub-carriers under eavesdropping constraints. Extensive simulation results demonstrate the efficacy of our approach, showcasing significant improvements in secrecy rate performance.

The trend of coordinated development of intelligent and connected vehicles is driving the continuous expansion of application scenarios and service functions for the Internet of Vehicles (IoV). The IoV has evolved from basic in-vehicle information services to intelligent networking, enhancing perception and decision-making. It is now developing towards coordinated vehicle-infrastructure control with moderate sensing and decision-making. However, the traditional separation of sensing, communication, and computation functions—and even their pairwise integrations, such as Integrated Sensing and Communication (ISAC), Integrated Communication and Computation (ICC), and Integrated Sensing and Computation (ISC)—create systemic inefficiencies like data redundancy and resource imbalance, which pose a critical bottleneck to the advancement of IoV. The unique characteristics of the IoV, particularly its mission objectives centered on safety and sensing, and the highly dynamic, resource-constrained network environment—compound these systemic inefficiencies, necessitating a dedicated investigation into an IoV-specific ISCC paradigm. Motivated by this, this paper provides the first in-depth, IoV-centric survey of the ISCC paradigm. First, we comprehensively review the development trajectory of the IoV, and based on this evolutionary path, define and categorize the development of ISCC into three distinct phases—Collaborative, Fusion, and Integrated—to clarify its integration process. Second, we analyze the limitations of partial integration paradigms (ISAC, ICC, ISC) and then establish a comprehensive taxonomy of ISCC implementation pathways, encompassing both physical-layer signal integration and network-level task-oriented resource management. Third, we survey the preliminary applications and research on the ISCC in the IoV and related fields. Finally, we outline the challenges and potential solutions to facilitate the realization of the ISCC in the IoV.

Edge computing significantly enhances the data processing efficiency and response speed of Connected and Autonomous vehicles (CAVs) by offloading computing tasks to network edge nodes. However, critical challenges arise, including limited edge resources, the complexity of resource allocation strategies and signal obstruction caused by buildings. Existing approaches often address communication and computing resources allocation separately, which hinders their ability to fulfill the stringent latency and reliability demands of CAVs. To tackle these challenges, this paper first employs Unmanned Aerial Vehicles (UAVs) as both communication relay nodes and mobile edge computing nodes, establishing a unique CAVs communication architecture. Based on the architecture, this paper proposes the DM-JCR algorithm grounded in the diffusion model, which is designed to optimize the joint allocation of communication and computing resources in edge computing nodes. Specifically, the algorithm uses environmental information category as a prompt to guide the diffusion model to generate an optimized joint allocation strategy for communication and computing resources. The denoising process of the diffusion model is improved based on the practical requirements of the joint communication and computing resources allocation problem. Finally, the DM-JCR algorithm is deployed in the proposed architecture to validate its effectiveness. Experimental results show that the proposed algorithm achieves superior performance compared to baseline methods in both latency and energy consumption.

Traffic congestion has become a critical bottleneck in urban development as cities rapidly expand. Multiintersection coordinated signal control with refined data support provides a promising solution. However, in practical deployment, the processing pipeline from Connected Autonomous Vehicles (CAVs) to roadside units and then to edge servers inherently involves delays, causing control decisions based on delayed observations to misalign with actual traffic demand and degrade control performance. Thus, this paper proposes a Delay-Aware State Compensation Multi-Agent Traffic Signal Control (DASCMATSC) approach. The method first establishes a delay model for homogeneous CAV environments, quantifying the impact of communication and computational delays on system performance. Second, a Long Short-Term Memory network-based state compensator fuses historical observations, actions, and delay labels to reconstruct current traffic states. Finally, under a centralized training with decentralized execution framework, compensated states are embedded into a multi-agent proximal policy optimization algorithm to achieve coordinated control. Simulation experiments in a $3 \times 3$ grid network with full CAV penetration demonstrate that the proposed method improves average speed, cumulative waiting time, and number of stops by $2.9 \%, 9.4 \%$, and $6.2 \%$ respectively compared to baseline methods, effectively mitigating the negative impact of observation delays on control performance.

Connected and Autonomous Vehicles (CAVs) are emerging as an inevitable trend in the future of the automotive industry, drawing collaborative attention from academia, industry, and government sectors. However, the inherent limitations in the sensing capabilities of CAVs restrict their safety and reliability while navigating complex road environment, confining them to address only a narrow scope of potential issues. Enhancing vehicle sensing capabilities by strategically deploying sensors along the road for cooperative sensing can effectively extend the range of vehicle sensing, achieving beyond-line-of-sight, high-precision sensing. It's foreseeable that an increased number and accuracy of roadside sensors would provide more advanced assistance to vehicles. Yet, due to deployment cost constraints, a high-density sensor deployment remains impractical. The challenge of optimizing sensor coordination within a limited budget still prevails. In this paper, contrasting with prior simplistic sensor setups and ideal environmental factors from previous works, we first remodel sensors and networks in light of the intricate real-world environment. By considering the spatial model of actual conditions, the 3D sensor network deployment issue is redefined. Furthermore, we introduce an optimization approach for 3D sensor deployment using the decision transformer, circumventing issues related to numerical reward modeling and disparities in policy learning and deployment data. Extensive simulation results attest that our proposed method effectively and efficiently enhances vehicle sensing capabilities within a limited cost.