With the emergence of 6G, autonomous systems, and massive Internet of Things, spectrum cartography and related radio map techniques are attracting increasing attention. These techniques are critical enablers for real-time, high-fidelity spatial, spectral, and temporal awareness in dense and dynamic radio frequency environments. However, research in this domain remains fragmented, with different methodologies studied in isolation. In this work, we present a comprehensive survey of the fundamental concepts and research taxonomy of spectrum cartography, together with a detailed technical tutorial covering its core methodologies. Specifically, we start by clarifying the relationships among spectrum cartography, radio map estimation, radio environment maps, and channel knowledge maps, and then establish a taxonomy of radio map types and available datasets. Afterwards, we present unified problem formulations covering key tasks such as map generation, interpolation, reconstruction, inpainting, estimation, distributed learning, and dynamic map updating. Furthermore, we categorize spectrum cartography methods and discuss their evolution from traditional model-based techniques, which rely on physical or statistical assumptions, to flexible data-driven paradigms and physics-enhanced learning, which combines the robustness of physical models with the flexibility of data-driven architectures. Finally, we discuss key applications unlocked by these techniques, such as autonomous unmanned aerial vehicle navigation and dynamic network management, together with technical challenges, deployment tradeoffs, standardization needs, and future research opportunities toward real-time, semantic-aware digital twins of the wireless environment.
The Decentralized Physical Infrastructure Network (DePIN) represents a transformative paradigm that redefines the construction, operation, and governance of Information and Communication Technology (ICT) infrastructure in the Web 3.0 era. DePIN integrates physical resources, such as networking equipment, storage, and computing power, with decentralized digital governance, forming a self-incentivized ecosystem that is collaboratively built, shared, and governed by the community. It provides a foundational framework for future communication networks, facilitating decentralized edge intelligence, efficient resource sharing, and trustworthy coordination among heterogeneous devices. Focusing on the feasibility of this emerging paradigm, this paper examines the technology landscape in the pre-DePIN era and gaps between existing methodologies and the forthcoming decentralized infrastructure for Web 3.0. It provides a systematic and comprehensive survey of the background, core characteristics, technical architecture, and applications of DePIN across various vertical domains. The paper analyzes the DePIN technology stack from six layers: physical infrastructure, blockchain, interaction, trust, incentive, and application, with special attention to their cross-layer feedback loops, implementation readiness, and deployment limitations. To further bridge conceptual analysis and practical deployment, we propose a DePIN feasibility assessment framework covering technical, governance, and economic dimensions. Moreover, we highlight promising research directions, providing insights and guidance for further exploration and deployment of DePIN.
The internet of vehicles (IoV) is rapidly evolving into an artificial intelligence (AI)-driven, multi-agent collaborative ecosystem, yet traditional communication paradigms struggle to address the challenges posed by dynamic network topologies, stringent resource constraints, and heterogeneous service demands. To bridge this gap, this survey systematically reviews the transformative potential of three emerging AI paradigms, including large language models (LLMs), Agentic AI, and Embodied AI, in reshaping future vehicular communications. We first introduce the concept of IoV and cellar vehicle-to-everything (C-V2X), while highlighting challenges IoV faces. Subsequently, we detail the fundamentals of these AI paradigms: architectural innovations, training methodologies, and prompt engineering for LLMs, and the core modules of Agentic AI and Embodied AI systems. The survey then analyzes domain-specific adaptation strategies, such as model compression techniques, specialized dataset construction, and cloud-edge-vehicle collaborative deployment for LLMs; distributed multi-agent collaboration frameworks for Agentic AI; as well as reconstruction of perception modules, world models, and executors in Embodied AI systems. Furthermore, we investigate practical applications across critical vehicular communication scenarios, covering beamforming, resource allocation, semantic communication, network optimization, and multi-agent collaboration. To validate these theoretical frameworks, we present a representative case study on Embodied AI-enhanced vehicular networks. Finally, we identify key challenges and outlines future research directions.
This paper aims to survey the integration of Information-Centric Network (ICN) with the Internet of Things (IoT), considering various aspects, including the role of Named Data Networking (NDN) caching schemes in mitigating latency in IoT use cases. Current IP-based protocols do not suit IoT system requirements, as they are not sufficient for dynamic and latency-sensitive IoT applications, such as health-care. NDN provides a shift from location-based to content-based information retrieval, which improves data availability and network efficiency by bringing cached content closer to users and reducing the bandwidth required for data delivery. This paper also reviews and classifies current caching approaches for NDN-based IoT, including content delivery and placement, and discusses the advantages, drawbacks, and effects of these caching schemes on content retrieval latency. It further provides a comparative evaluation of representative caching strategies to highlight their latency-performance trade-offs. This survey outlines the present issues, suggests future work, and provides useful information for improving latency-conscious caching schemes for IoT networks.
The functioning of the Internet depends on the exchange of routing information across Autonomous Systems (ASes) using the Border Gateway Protocol (BGP). The mutual trust between ASes and the complexity of routing mechanisms expose BGP to significant cyber-threats and attacks, posing risks to the broader Internet-dependent economy. This survey presents a multi-stakeholder perspective on BGP routing security, integrating academic, industrial, and institutional points of view. It reviews policy frameworks developed by both public and private organizations, outlining synergies between industry and institutions and best practices for securing inter-domain routing. The paper further examines open challenges, identifying four research and innovation directions. For each direction, we survey potential solutions and categorize them based on stakeholder interest, thereby bridging the gap between academic advances and operational deployment. Finally, we analyze potential future directions in BGP security, highlighting differentiated priorities and strategic interests across stakeholders.
Mobile communication systems are undergoing a hardware-driven technological evolution, aiming to enhance the controllability and programmability of wireless propagation. Recently, pinching antennas (PAs) have emerged as an innovative flexible-antenna technology, capable of dynamically adjusting PA positions along waveguides to construct line-of-sight (LoS) links and mitigate large-scale path loss and LoS blockage. In parallel, integrating artificial intelligence (AI) into wireless system designs and enabling multifunctional wireless services have become an inevitable trend. PA systems particularly necessitate deep learning (DL)-enabled approaches due to their inherent optimization complexity, while PAs’ ability to customize wireless channel characteristics supports the deployment of multifunctional services. This article surveys the state-of-the-art studies to provide a comprehensive overview PAs. We first outline the fundamental principles of PAs, classify existing PA systems based on PA configurations, implementation schemes, and channel models, and summarize the key analysis results. Subsequently, we present the reviews of classic optimization techniques for PA-enabled information transmission, non-orthogonal multiple access (NOMA) schemes, and beyond fifth-generation (B5G) technologies. Next, we introduce DL-enabled PA designs, covering unsupervised learning, deep reinforcement learning (DRL), and convex-aided DL frameworks. We further summarize PA-enabled multifunctional service designs tailored to scenarios including information security, wireless powered networks, and integrated sensing and communication (ISAC). Finally, we highlight the critical challenges, open issues, and promising future research directions for PAs.
The rapid deployment of autonomous systems, including Connected Autonomous Vehicles (CAVs), Unmanned Aerial Vehicles (UAVs), and Autonomous Underwater Vehicles (AUVs), is enabling coordinated cross-domain operations across transportation, defense, environmental monitoring, and emergency response applications. However, the integration of such heterogeneous platforms introduces significant cybersecurity challenges arising from complex architectures, diverse communication technologies, dynamic operational conditions, and data-driven autonomous decision-making. Existing cybersecurity studies and defense mechanisms often focus on specific autonomous platforms or limited security layers, providing insufficient analysis of integrated cybersecurity interactions across heterogeneous UAV, CAV, and AUV environments. In this survey, we present a comprehensive analysis of cybersecurity threats, vulnerabilities, and defense mechanisms for integrated UAV, CAV, and AUV systems across single-domain, dual-domain, and fully integrated tri-domain operational scenarios. The survey systematically examines security challenges spanning physical, sensing, communication, network, control, and application layers, together with cross-domain attack propagation and collaborative security dependencies. In addition, we review machine learning-based intrusion detection techniques for integrated UAV, CAV, and AUV systems and present HybridEnsemble-ID as an example of a hybrid cybersecurity framework that integrates supervised, unsupervised, semi-supervised, and reinforcement learning approaches for adaptive threat detection in heterogeneous autonomous environments. Furthermore, we consolidate publicly available datasets, simulation platforms, emulators, and hardware-in-the-loop testbeds to support reproducible evaluation and benchmarking of cybersecurity mechanisms for integrated autonomous systems. Finally, we identify open challenges and future research directions, including unified cross-domain evaluation frameworks, deployment-oriented security validation, adaptive defense architectures, and standardized benchmarking methodologies for resilient cross-domain autonomous operations.
Backscatter communication (BC) has emerged as a promising technology for low-cost and ultra-low-power wireless applications. However, the inherent openness of wireless channels, the passive reflection characteristics of BC systems, and stringent resource constraints introduce significant security vulnerabilities. Conventional cryptographic solutions are often impractical due to their high computational complexity, while physical layer security (PLS) offers lightweight and cost-effective security guarantees. Despite the growing interest, a comprehensive survey on PLS techniques for BC systems remains lacking. This paper presents the first comprehensive survey on PLS techniques for BC systems, encompassing Physical Layer Authentication (PLA), Physical Layer Key Generation (PLKG), Physical Layer Secure Transmission (PLST), and Physical Layer Anti-Jamming Transmission (PLAJT), along with the integration of emerging technologies such as Reconfigurable Intelligent Surfaces (RIS) and autonomous aerial vehicles (AAVs). Structured around the unique constraints and security implications of BC systems, this survey contributes three distinctive perspectives: a BC-specific threat model grounded in intrinsic security characteristics, a practical evaluation framework, and a cross-paradigm analysis revealing complementarities and performance differences of different PLS paradigms. Specifically, this paper first introduces the fundamental principles of BC systems. It identifies the impact of system characteristics on security performance and summarizes existing potential security threats. It also elaborates on the core principles and basic concepts of PLS, RIS and AAV technologies. This paper further establishes a classification framework of PLS systems suitable for BC scenarios and constructs a complete evaluation criterion system to support systematic qualitative and quantitative analysis of existing research works. With the established classification framework and evaluation criterion system, this paper conducts a systematic review on state of the art PLS solutions for BC systems published in recent years. This review clarifies the mainstream prerequisite assumptions and essential performance trade-offs widely adopted in the design of PLS for BC systems. This paper also outlines the design principles of PLS for BC systems. It provides system designers with a comprehensive analytical perspective to further promote research on BC system oriented PLS technologies. This paper finally summarizes a set of important open challenges and presents targeted future research directions.
The terahertz (THz) band and the optical frequency window are emerging as powerful enablers of next generation biomedical technologies, supporting high resolution sensing of biological processes, targeted intra-body actuation, and intra-body communication for coordinated functional operation. Although each modality has demonstrated significant promise individually, research efforts have largely progressed in isolation, limiting the exploration of their potential for integrated deployment. This comprehensive survey bridges these domains by presenting a unified analysis of THz and optical modalities for intra-body sensing, actuation, and communication. Building on this foundation, we articulate a vision for intra-body systems in which multiple functions are seamlessly coordinated at cellular and subcellular scales. Applications such as continuous health monitoring and closed loop therapeutic interventions illustrate the transformative potential of these integrated approaches. We outline a three-stage roadmap covering foundational modeling, in-vitro integration, and in-vivo translation, identifying critical open challenges including compact source generation, power delivery, biosensing miniaturization, and biocompatibility.
The exponential growth of the Internet of Things (IoT) demands robust yet efficient authentication mechanisms to secure heterogeneous communication networks including resource-constrained devices and multi-tier infrastructures. This paper presents a systematic and comprehensive survey of lightweight authentication protocols in IoT communication networks, supported by a structured systematic literature review methodology. The survey outlines foundational cryptographic primitives and security requirements, and introduces a multi-dimensional taxonomy that maps authentication schemes across cryptographic primitives, authentication types, application domains, security goals, and implementation environments. Building on this taxonomy, the paper reviews cryptography-based, hardware-based, and hybrid authentication approaches, followed by quantitative comparative analysis of computational cost, communication and storage overhead, energy consumption, and formal security validation models. The results show that hash/XOR-based schemes offer very low computation but remain vulnerable to advanced attacks, PUF-based designs provide hardware-rooted trust yet suffer from environmental instability and reconstruction overhead, ECC-based schemes achieve strong security with compact keys but incur non-trivial resource cost, while post-quantum mechanisms enhance quantum resilience but currently impose significant communication and latency burdens for constrained IoT devices. Blockchain-assisted authentication improves decentralised trust at the expense of protocol and infrastructure overhead. Finally, the paper consolidates critical insights and open challenges including scalable key management, lightweight formal security assurance, reliable hardware trust, and post-quantum readiness, providing directions for future secure IoT authentication research.
The rapid proliferation of edge computing resources, together with transformative breakthroughs in artificial intelligence (AI), has catalyzed the emergence of edge AI. As a representative paradigm of AI-communication integration in 6G, edge AI extends AI capabilities to edge nodes and mobile devices, enabling ubiquitous, real-time, and secure intelligent services. Edge AI inference refers to the routine invocation phase of trained models, where forward propagation is executed to produce task-specific outputs under stringent latency and reliability requirements. However, it encounters critical challenges stemming from the escalating computational demands of AI models and the intrinsic resource limitations of edge environments. To address these challenges, this paper presents a comprehensive survey on edge AI inference in 6G mobile networks, identifying key principles and innovative solutions for optimizing inference performance under edge resource constraints. We begin by outlining communication-efficient techniques and adaptive model deployment strategies designed to mitigate transmission, computation, and storage overhead in edge AI inference. We then review collaborative inference mechanisms that facilitate effective workload distribution across cloud-edge-device computing nodes. This is followed by an in-depth exploration of task-oriented optimization, which coordinates network resources throughout the end-to-end inference workflow to boost the performance of both discriminative and generative edge AI inference. Finally, we examine practical platforms and promising applications, discuss future research directions and open issues, with the hope of inspiring continued advancements in this evolving field.
The rapid proliferation of artificial intelligence in the industrial sector is catalyzing a smart manufacturing paradigm driven by industrial AI agents. In contrast to IT counterparts, future industrial AI agents operate under unique constraints that demand both adaptive intelligence and strictly deterministic communication with ultra-low latency and ultra-high reliability. To address these stringent requirements, the convergence of reinforcement learning (RL) and Time-Sensitive Networking (TSN) has emerged as a critical enabler. This paper presents the fundamentals of RL, TSN, and industrial AI agent communication, and establishes a mapping between AI agent cognitive behaviors, communication requirements, and deterministic networking mechanisms. We propose a four-dimensional framework encompassing intelligent scheduling, dynamic resource management, network convergence and mobility, and application-aware networking to systematically analyze existing RL-for-TSN research and elucidate challenges and requirements for future industrial AI agent communication. Finally, we outline a research roadmap spanning constrained RL algorithms and agent-network co-design to pave the way for deterministic industrial edge intelligence.
With the continuous evolution of the 6G sea-land-air integrated communication paradigm, underwater intelligent communication systems have emerged as a prominent research focus in recent years. Due to harsh environmental conditions, limited communication bandwidth, high propagation delays, and severe energy constraints, the modelling and operation of underwater intelligent communication systems face significant challenges in achieving intelligence, coordination, and sustainability. To meet the above challenges, multi-agent reinforcement learning (MARL), characterized by its decentralized and autonomous decision-making capabilities, has been regarded as a promising framework for intelligent, distributed, and adaptive coordination in such environments. Nevertheless, there is still a lack of comprehensive surveys on using MARL to optimize underwater communication networks. Therefore, this survey provides a comprehensive overview of recent advancements in applying MARL techniques to optimize underwater communication networks and bridges this gap. Specifically, we review the fundamental components of MARL and explain why it is particularly well suited for optimizing underwater networks. Then, we review MARL-based underwater applications in both static communication systems and dynamic mobile networks, including routing strategies, network security, resource allocation, MAC layer optimization, cooperative multi-robot formation, and target tracking. Furthermore, we summarize algorithmic innovations tailored to underwater communication optimization, focusing on improvements in training efficiency, robustness, and scalability. Building on these analyses, we further analyze open challenges and outline future research directions for advancing MARL-enabled underwater communication networks.
The demand for low-latency video streaming has grown rapidly with the emergence of mission-critical and highly interactive applications such as extended reality (XR), remote surgery, cloud gaming, teleoperation, and autonomous systems. Despite continuing advances in networking and computing technologies, achieving consistently low end-to-end delay while maintaining service quality remains a fundamental challenge. Existing surveys often focus on individual components of the streaming pipeline, whereas a unified and up-to-date view of low-latency Internet video streaming is still lacking. This survey addresses this gap by providing a structured review of low-latency Internet video streaming from the perspectives of applications, core challenges, and future trends. We first summarize representative application domains and clarify how their latency requirements motivate different technical bottlenecks. We then organize the literature around three recurring challenges: Large Data, Network Bandwidth, and Prediction, covering representative solutions in video compression and delivery, transport and congestion control, adaptive streaming, and QoS/QoE prediction and evaluation. In addition, we review how recent AI techniques are reshaping the low-latency streaming pipeline, including learning-based optimization, generative/foundation-model-enabled streaming, and intelligent network control. By synthesizing representative techniques, distilling lessons learned, and highlighting open challenges and research directions, this survey provides a unified and forward-looking reference for researchers and practitioners working on next-generation low-latency video streaming systems.
The integration of reconfigurable intelligent surfaces (RIS) with terahertz (THz) communication is emerging as a transformative enabler for sixth-generation (6G) wireless networks, offering unprecedented capacity, spectrum utilization, and propagation control. This paper presents a comprehensive survey of RIS-assisted THz systems, with a focus on their design, performance, and integration challenges. A unified taxonomy is introduced, organizing the discussion into key domains: system architectures, channel modeling, hardware implementation, signal processing, and experimental validation. The paper consolidates diverse technological perspectives by examining metasurface hardware, wideband and near-field propagation modeling, and signal processing tailored to THz-specific impairments. It further reviews advanced methods for beamforming, phase error mitigation, and channel estimation under practical hardware constraints. Beyond performance optimization, the work explores deployment paradigms including indoor coverage, autonomous aerial vehicles (AAV)-based relays, and RIS integration in joint communication and sensing frameworks. A major contribution of this survey is the detailed mapping between RIS design strategies and 6G use cases, offering deployment guidelines and highlighting trade-offs across different tuning mechanisms, fabrication techniques, and control architectures. The paper also addresses open research challenges such as hardware scalability and real-time reconfigurability, ultimately serving as a foundational reference for researchers and engineers developing next-generation RIS-enabled THz communication systems.
Synthetic network traffic generation has emerged as a promising alternative for various data-driven applications in the networking domain. It enables the creation of synthetic data that preserves real-world characteristics while addressing key challenges such as data scarcity, privacy concerns, and purity constraints associated with real data. In this survey, we provide a comprehensive review of synthetic network traffic generation approaches, covering essential aspects such as data types and generation models. With the rapid advancements in Artificial Intelligence (AI) and Machine Learning (ML), we focus particularly on deep learning (DL)-based techniques while also providing a detailed discussion of statistical methods and their extensions, including commercially available tools. We present a comprehensive comparison of generation approaches and provide an AI tool to apply this comparison for any network traffic generation papers. Furthermore, we highlight open challenges in this domain and discuss potential future directions for further research and development. This survey serves as a foundational resource for researchers and practitioners, offering a structured analysis of existing methods, challenges, and opportunities in synthetic network traffic generation.
Next-generation (xG) wireless systems, including sixth-generation (6G) networks and beyond, are expected to deliver data rates on the order of terabits per second and sub-millisecond latency. Meeting these requirements increasingly relies on artificial intelligence (AI)-enabled radio access and physical-layer (PHY) processing. However, realizing such AI-driven PHY functionality with deep learning (DL) is challenging as deep neural networks (DNNs) are computationally intensive and memory hungry, often exceeding the capabilities of resource-constrained user equipment (UE) and edge hardware. This paper surveys model-compression techniques for efficient wireless intelligence, focusing on pruning, quantization, and knowledge distillation (KD), together with architectural and algorithmic optimizations. For each technique, we summarize theoretical foundations, practical implementation strategies, and wireless-specific considerations, and discuss how design choices translate into latency, energy, and memory outcomes on deployment hardware. We review applications across core PHY wireless tasks, including automatic modulation classification (AMC), channel state information (CSI) processing and feedback, beamforming (BF), recognition and identification, channel estimation and detection, and localization. Drawing on comparative analysis of more than 50 studies, we highlight trade-offs among model size, computational complexity, energy consumption, and task-level performance under wireless evaluation protocols. We further discuss hardware-software compatibility considerations for compressed model deployment and outline open challenges and future research directions for compression-aware deployment in xG wireless systems.
The rapid advancement toward sixth-generation (6G) wireless networks has significantly intensified the complexity and scale of optimization problems, including resource allocation and trajectory design, often formulated as combinatorial problems in large discrete decision spaces. However, traditional optimization methods, such as heuristics and deep reinforcement learning (DRL), face practical challenges in meeting stringent latency and scalability requirements, especially in large-scale, highly dynamic, and reconfiguration-sensitive deployments in increasingly heterogeneous and resource-constrained network environments. Large language models (LLMs) present a transformative paradigm by enabling natural language-driven problem formulation, context-aware reasoning, and adaptive solution refinement through advanced semantic understanding and structured reasoning capabilities. This paper provides a systematic and comprehensive survey of LLM-enabled optimization frameworks tailored for wireless networks. We first introduce foundational design concepts and distinguish LLM-enabled methods from conventional optimization paradigms. Subsequently, we critically analyze key enabling methodologies, including natural language modeling, solver collaboration, and solution verification processes. Moreover, we explore representative case studies to demonstrate LLMs’ transformative potential in practical scenarios such as optimization formulation, low-altitude economy networking, and intent networking. Finally, we discuss current research challenges, examine prominent open-source frameworks and datasets, and identify promising future directions to facilitate robust, scalable, and trustworthy LLM-enabled optimization solutions for next-generation wireless networks.
The growing complexity, scale, and heterogeneity of 6G wireless systems call for a shift toward AI-native architectures that are not only data-driven but also topology-aware, adaptive, and distributed. Graph Neural Networks (GNNs), with their native support for graph-structured data, are well-suited for modeling the irregular and dynamic relationships inherent in wireless communication systems. However, standalone GNNs may be insufficient to address key 6G challenges such as continual learning, data scarcity, and dynamic adaptation. This survey, therefore, explores the emerging synergy between GNNs and complementary AI paradigms, including deep reinforcement learning (DRL), federated learning (FL), meta-learning, generative models, mixture-of-experts (MoE), and world models, enabling hybrid AI-GNN architectures for intelligent control, predictive adaptation, and scalable optimization across the wireless stack. We systematically review how these GNN-centric AI models can support intelligent functionality in next-generation network architectures such as O-RAN, as well as core 6G domains, including edge computing for wireless systems, advanced MIMO, traffic prediction, and digital twins. The survey also highlights key challenges in hybrid AI-GNN adoption, particularly scalability, generalization across dynamic topologies, interpretability, and symbolic reasoning, and discusses emerging strategies such as graph causality learning, dynamic GNNs, and neurosymbolic integration to address them. By consolidating recent advances and outlining open research directions, this work positions hybrid AI-GNN architectures as a promising approach toward enabling intelligent, energy-efficient, and context-aware wireless systems for 6G and beyond.
