Transactions on Emerging Telecommunications Technologies最新文献

The convergence of Internet of Things (IoT) technologies with Vehicular Ad-hoc Networks (VANETs) serves as a critical infrastructure for intelligent transportation systems. It enables vehicle-to-vehicle and vehicle-to-infrastructure communications. However, the dynamic and heterogeneous nature of VANETs creates significant security challenges. Particularly, in detecting novel intrusion patterns when labeled data is limited. Traditional intrusion detection systems struggle with rapid adaptation to emerging attack vectors, often requiring extensive retraining and substantial labeled datasets. We present MAFSID (Multi-Agent Few-Shot Intrusion Detection), a collaborative learning framework utilizing multiple specialized detection agents through few-shot learning paradigms. Our system employs five distinct agents, i.e., Network Traffic Analyzer, Anomaly Detector, Behavior Analyzer, Protocol Analyzer, and Threat Classifier. These agents communicate and share knowledge to collectively identify intrusions with minimal training samples. Comprehensive evaluations on public datasets such as NSL-KDD, CICIDS2017, and In-Vehicle datasets demonstrate MAFSID's superior performance. Furthermore, robustness analysis reveals high resilience against adversarial attacks, especially communication jamming, showing minimal impact on performance. Notably, the collaborative agent framework enables rapid adaptation to new attack patterns while maintaining robust performance across diverse network conditions.

Federated learning (FL), as an advanced distributed learning paradigm, faces significant challenges, particularly in terms of slow convergence and instability, which are exacerbated by heterogeneous data distributions. A critical issue in this context is data heterogeneity, which increases gradient estimation variance and drives the model toward local minima that are distant from the global optimum. Previous studies have primarily focused on using Control Variates (CVs) to reduce gradient estimate variance without introducing bias. In this work, we propose a novel distributed gradient optimization framework, FedNCV, aimed at effectively reducing gradient variance. Central to this approach is the use of the REINFORCE Leave-One-Out (RLOO), a CV-based technology, which serves as the core gradient estimator for FedNCV at both the client and server levels. We have developed an algorithm based on FedNCV and provided three theoretical results. Experimental evaluations demonstrate that the proposed method enhances performance. The dual structure of FedNCV equips it to address the challenges of data heterogeneity and scalability in federated networks, offering a promising solution for applications in heterogeneous FL environments. Additionally, the efficacy of FedNCV was validated across four diverse datasets under a Dirichlet distribution with $$, setting new performance benchmarks when compared to six leading methods.

In the evolving landscape of the Space Air Ground-Integrated Network (SAGIN), addressing the challenges of limited labeled data and the need for adaptable models is crucial for effective data processing across diverse and heterogeneous sources. Weakly supervised few-shot semantic segmentation (WSFSS), aims to segment the unseen targets by solely relying on support images with class-level annotations, offering a robust solution to the complexities inherent in SAGIN environments. Existing works usually generate pseudo masks for training images, and then convert WSFSS to the plain few-shot semantic segmentation task. However, these rough pseudo masks, almost always, contain background noise due to imprecise object localization during mask generation, which thus leads to undesirable segmentation results. To mitigate the above challenge, we inject implicit text supervision into WSFSS, and propose an efficient text-guided hierarchical attention (ETHA) framework to explicitly alleviate the mask noise issue. Specifically, we first propose a cross-modal interaction attention module to capture comprehensive object guidance from text embeddings and refocus the model's attention toward the area of the focused object. In addition, we propose a lightweight dual visual cross-attention module to efficiently aggregate the contextual information among each branch and common object clues from both branches, which provides enhanced visual features to facilitate the cross-modal information interaction. Based on single-scale features, ETHA has established new state-of-the-art results on the golden WSFSS datasets, that is, PASCAL-$$ and COCO-$$. These results highlight ETHA's potential for improving accessibility and efficiency for robust applications in SAGIN environment.

The operation of Unmanned Aerial Vehicles (UAVs) in low-altitude airspace is a key component in the Space-Air-Ground Integrated Network (SAGIN). Efficient and rational path planning is essential for UAV operations. Existing path-planning algorithms typically rely on uniform-grid models based on Cartesian coordinate systems, which seldom account for the unique characteristics of UAV terminal airspaces. UAV terminal airspaces are often defined as cylindrical volumes where multiple UAVs converge at vertiports, allowing for higher operational densities compared to en-route airspaces. While a fine-grained grid model is essential for UAV terminal airspace, it is inefficient for en-route airspace due to excessive computational costs. This paper presents a path-planning approach based on a variable grid model to find the optimal path while effectively utilizing airspace resources and minimizing computational overhead. Specifically, for the UAV terminal airspace, we propose the Grid-Optimized A* Path-Planning (GO-APP) algorithm, which establishes a sector-grid model based on a cylindrical coordinate system to find the optimal path. Extending to the en-route airspace, the Variable-Grid A* Path-Planning (VG-APP) algorithm integrates the GO-APP and A* algorithms to search the optimal path by stages. Simulations indicate that as the obstacle density in terminal airspace increases from 10% to 40%, GO-APP demonstrates a path length improvement ranging from 0.78% to 24.46% relative to A*. In generating a path from en-route airspace to terminal airspace, VG-APP significantly outperforms A*, reducing path length by up to 15.39% along with improving computational efficiency by 30.54%. Additionally, experiments in the real-world city of Hangzhou validate the effectiveness of the proposed approach.

Despite presenting significant safety concerns, the proliferation of Cyber-Physical Systems (CPS) in industries such as electricity distribution and medicine has spurred innovative applications. The integration of 5G technology, Software-Defined Networking (SDN), and Vehicular Ad Hoc Networks (VANET) within Industrial CPS (ICPS) enables dynamic, real-time interactions but also exposes these systems to vulnerabilities and potential malicious activities. Addressing these risks requires robust security strategies that safeguard both the physical and digital components of CPS. In this context, the Strengthening the Safety of 5G-Enabled CPS using Key Agreement Method (SSKAM) framework is proposed. It incorporates a triple-element user identification technique leveraging user passwords, mobile devices, and unique biometrics. This approach ensures two-way authentication and facilitates the establishment of secure session keys for encrypted communication between registered users, CPS smart devices, and VANETs. The study also explores advanced key agreement methods tailored for 5G-enabled CPS, focusing on enhancing identification protocols while maintaining a balance between computational efficiency, communication overhead, and resilience to emerging threats. By integrating SDN, the framework enforces dynamic security measures at the network level, ensuring real-time adaptability to potential threats. Comprehensive evaluations demonstrate the efficacy of the proposed SSKAM framework in mitigating risks such as replay attacks, impersonation, and man-in-the-middle assaults. The results highlight its viability in safeguarding the integrity and confidentiality of CPS, offering a scalable, efficient, and practical solution to address the evolving security challenges in 5G-enabled CPS integrated with IoT ecosystems and VANETs.

The traditional Dynamic Window Approach (DWA) for local path planning of unmanned aerial vehicles (UAVs) exhibits limitations in flexibility and robustness. Specifically, its fixed evaluation function weights, velocity sampling resolution, and dynamic window ranges fail to adapt to changing environmental conditions, resulting in reduced adaptability of the velocity search space. To address this issue, this study proposes an improved DWA based on the Sparrow Search Algorithm (SSA). First, the proposed algorithm adaptively adjusts dynamic window parameters according to the complexity of the obstacle environment, thereby optimizing the UAV's velocity search space. Second, a velocity sampling resolution strategy is introduced to achieve an effective trade-off between the quality and quantity of predicted trajectories based on the density of dynamic obstacles. Third, by leveraging the strong global search capability and rapid convergence properties of the SSA, the weights of the evaluation function are adaptively optimized to enhance global optimality. Experimental results show that, compared with DWA in a dense multiple dynamic-static obstacles scenario, the proposed algorithm achieves improvements of 8.8%, 66.7%, and 18% in path length, safety distance, and number of iterations, respectively. These enhancements contribute to improved planning efficiency, safety, and overall optimality in UAV operations.

The rapid evolution of the Internet of Things (IoT) has led to the growth of the Internet of Medical Things (IoMT), encompassing interconnected medical devices, wearable sensors, and healthcare systems. IoMT extends the capabilities of IoT into the healthcare sector, representing a promising technology for the future of healthcare. The underlying technology has transformed the healthcare landscape by continuously collecting patient data in real time and providing a remote monitoring and diagnostic system. However, IoMT introduces significant challenges in data security, privacy, and system reliability due to centralized data storage models, which can create a single point of failure and raise concerns about privacy and security. To address these challenges, this research proposes a blockchain-based framework to enhance data security and privacy by decentralizing data storage and managing device authentication using smart contracts. Given the sensitive nature of medical data and the potential repercussions of security breaches, each medical device has a unique digital identity represented by a blockchain-based smart contract, supporting the multi-device mapping required for managing multiple diseases in healthcare diagnosis and treatment. This approach enhances healthcare security and efficiency. A proof-of-work mechanism ensures secure transaction validation and experimental results demonstrate that the proposed framework significantly improves data integrity and security while optimizing system performance, as measured by gas consumption and latency. The assessment demonstrates the feasibility of employing blockchain technology to enhance the security and privacy of the IoMT healthcare system, providing a robust solution to existing security challenges and protecting patient data.

Utilizing blockchain in vehicular ad-hoc networks (VANETs) can proficiently resolve concerns pertaining to data security and privacy. The limited throughput of blockchain obstructs its extensive implementation in VANETs. Current studies on enhancing blockchain throughput frequently encounter the issue of action space proliferation, leading to inadequate scalability. This research presents a strategy for optimizing blockchain performance on VANETs using deep reinforcement learning (DRL). The suggested method enhances blockchain throughput by selecting block producers and consensus methods, and by modifying block size and intervals, while maintaining decentralization, low latency, and security in VANET-based blockchain systems. Furthermore, to augment network security, an anomaly detection mechanism is incorporated, utilizing machine learning techniques to identify and mitigate potential attacks aimed at VANETs. The proposed system enhances throughput and fortifies resilience against malicious operations by identifying anomalous patterns in network behavior. The method uses the BDQ framework to meticulously partition the action space, tackling the action space explosion issue that occurs with conventional DRL techniques in blockchain throughput optimization. Simulation results indicate that the suggested solution significantly improves the throughput and security of the VANET-based blockchain system.

Smart connected, and autonomous vehicles represent a revolutionary advancement in transportation by incorporating cutting-edge technologies like Internet of Things (IoT), Artificial Intelligence (AI), and 5G/6G wireless communication. These technologies enhance efficiency, reliability, and sustainability in modern vehicular networks. However, the rapid expansion of intelligent vehicles has also led to rising cyber risks, introducing new forms of attacks that threaten security, privacy, and safety. Even minor anomalies in vehicular units may cause severe consequences, including fatalities. To resolve these issues, this study proposes an effective Intrusion Detection System (IDS) using Siamese Gated Memory Networks. The model learns traffic behaviors and generates proximity scores, which are processed through dense feedforward layers for classifying multiple vehicular threats. To further optimize detection, a Modified Beetle Optimization (MBO) technique is integrated into the feedforward layers. The approach is trained and examined on benchmark datasets, comprising NSL-KDD 2019, UNSW-NB-15, and VeReMi, using key metrics like precision, specificity, accuracy, F1 score, and recall. Recommended experimental analysis demonstrates superior performance examined with conventional techniques, achieving 0.993 accuracy, 0.991 precision, 0.99 recall, and 0.992 F1 score. The findings validate the robustness of hybrid Siamese networks with ensemble meta-heuristic optimization in securing Intelligent Transportation Systems.

Due to the shortage of explainable, resource-efficient solutions and the lack of unified multi-attack detection abilities, existing vehicular ad hoc networks (VANET) security frameworks fail to meet the critical requirements of real-time vehicular environments. Most traditional models rely heavily on centralized processing, making them unsuitable for dynamic, latency-sensitive distributed VANET architectures. These limitations create a serious threat to the safety and reliability of vehicular communication systems. To address these challenges, this study proposes an explainable machine learning framework for real-time multi-attack threat detection (EXMAT) in edge-enabled VANET environments. The framework is designed specifically for edge-enabled VANET platforms. EXMAT combines the novel XGBoost classifier with custom-engineered behavioral features and post hoc explainability to provide accurate decisions directly at the vehicular edge. The novelty of the model lies in its combined feature space, which fuses behavioral dynamics, communication patterns, and lightweight Boolean anomaly flags. The simulation of the model is performed under the VeReMi dataset. To strengthen the dataset for precise analysis of the threat, we synthetically extended it with complex attack patterns. Experimental results show that the proposed model achieves an overall classification accuracy of 95.78% with an almost perfect F1-score for standard behavior samples of 99.98% and 94.36% for replay attacks. These results highlight EXMAT's ability to be applied in real-time vehicular networks, enhancing traffic safety and security against unknown cyberattacks.