Ad Hoc Networks最新文献

{"title":"QPSO-driven deep hybrid network for robust CSI localization in disaster scenarios","authors":"Bidyarani Langpoklakpam,&nbsp;Lithungo K Murry","doi":"10.1016/j.adhoc.2026.104157","DOIUrl":"10.1016/j.adhoc.2026.104157","url":null,"abstract":"<div><div>Rapid and accurate victim localization is essential for efficient emergency response and rescue operations, as natural disasters continue to occur with increasing frequency. Channel State Information (CSI) plays a pivotal role in positioning in complex environments because it can provide detailed, real-time propagation characteristics. This paper proposes a hybrid deep learning framework for CSI localization, where residual learning (ResNet) and Swin Transformer are jointly combined to capture both local and global CSI characteristics. The extracted features are mapped to spatial coordinates using a Backpropagation Neural Network regression model, in which key hyperparameters are optimized using a Quantum-inspired Particle Swarm Optimization (QPSO) strategy to enhance convergence stability and localization accuracy. The proposed system is evaluated through cross-environment testing across three real-world scenarios exhibiting diverse multipath propagation, obstructions, and interference patterns, emulating realistic post-disaster wireless conditions. Along with RMSE and MAE, Node Localization Efficiency (NLE) is employed to assess effective node coverage, and comparative results against existing methods highlight the superiority of the proposed approach. Across three complex real-world scenarios, the proposed method achieves RMSE values of 0.5292 m, 0.6084 m, and 0.7231 m, while achieving up to 37.5 % improvement in NLE over state-of-the-art approaches.</div></div>","PeriodicalId":55555,"journal":{"name":"Ad Hoc Networks","volume":"184 ","pages":"Article 104157"},"PeriodicalIF":4.8,"publicationDate":"2026-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146079060","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"LSGRS: A geolocation and reputation-aware dynamic dual-layer sharding scheme for scalable vehicular blockchain networks","authors":"Zhiqiang Du ,&nbsp;Qingyu Jiao ,&nbsp;Xiaopeng Zhang ,&nbsp;Muhong Huang ,&nbsp;Siqi Zheng ,&nbsp;Wendong Zhang ,&nbsp;Yanfang Fu ,&nbsp;Alwyn Hoffman","doi":"10.1016/j.adhoc.2026.104150","DOIUrl":"10.1016/j.adhoc.2026.104150","url":null,"abstract":"<div><div>In vehicular networks, the massive influx of connected vehicles and high-frequency data transmissions expose scalability bottlenecks in existing architectures. Sharded blockchains enable parallel processing via network partitioning. However, conventional sharding schemes struggle to cope with the high mobility of vehicles and the dynamic nature of node states—particularly in terms of adapting shard strategies.</div><div>To address these limitations, this paper proposes a dynamic dual-layer sharding mechanism, termed LSGRS, specifically tailored for highly mobile vehicular networks. LSGRS incorporates a dynamic network sharding mechanism that optimizes node selection based on reputation scores, physical proximity, and predicted vehicle trajectories. This approach reduces intra-shard communication latency and mitigates the risk of shard compromise by malicious nodes. A dual-layer hierarchical architecture is designed to separate mining and packaging tasks across two distinct layers, effectively alleviating the performance bottlenecks. In addition, the proposed Dynamic Node Join and Exit Protocol (DNJEP) ensures real-time adaptation to node failures or adversarial behaviors without disrupting ongoing services. Finally, the proposed scheme is evaluated through experiments in a simulated urban environment built with Veins and SUMO. The results demonstrate that LSGRS outperforms baseline approaches in terms of communication overhead, transactions per second (TPS), and consensus latency.</div></div>","PeriodicalId":55555,"journal":{"name":"Ad Hoc Networks","volume":"184 ","pages":"Article 104150"},"PeriodicalIF":4.8,"publicationDate":"2026-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146079055","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"A transformer-based method for radio-frequency fingerprinting of IoT devices","authors":"Carlos Herrera-Loera ,&nbsp;Carolina Del-Valle-Soto ,&nbsp;Leonardo J. Valdivia ,&nbsp;Miguel Bazdresch ,&nbsp;Carlos Mex-Perera","doi":"10.1016/j.adhoc.2026.104155","DOIUrl":"10.1016/j.adhoc.2026.104155","url":null,"abstract":"<div><div>Radio Frequency Fingerprint Identification (RFFI) is a technique used to classify wireless devices by examining the unique signal distortions that arise from hardware imperfections inherent to each device. This method allows a wireless receiver to identify one or more transmitters accurately. Previous studies have presented RFFI results with wireless modulations such as LoRa, ZigBee, and Bluetooth Low Energy (BLE). This paper presents a method for RFFI in Internet of Things (IoT) devices using Gaussian Frequency Shift Keying (GFSK) modulation. The proposed RFFI method is based on an encoder module of a Vision Transformer (ViT); the results are compared with those obtained using a convolutional neural network (CNN). We present an analysis of the method’s accuracy in several propagation scenarios. We also analyze the effect of various parameters on the method’s accuracy, including training dataset size, training vector length, the portion of the transmitted packet to train on, and the number of epochs for training.</div><div>Experimental findings indicate that the transformer-based classifier performs slightly better than a CNN (up to 5% better accuracy) in non-line-of-sight (NLOS) propagation conditions. More significantly, the transformer-based classifier requires only half the training epochs compared to the CNN.</div></div>","PeriodicalId":55555,"journal":{"name":"Ad Hoc Networks","volume":"184 ","pages":"Article 104155"},"PeriodicalIF":4.8,"publicationDate":"2026-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146079057","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Distributed video analytics for IoT intelligent systems","authors":"Rodolfo W.L. Coutinho ,&nbsp;Azzedine Boukerche","doi":"10.1016/j.adhoc.2026.104149","DOIUrl":"10.1016/j.adhoc.2026.104149","url":null,"abstract":"<div><div>Computer vision embedded in Internet of Things (IoT) systems will enable a new era of smart applications where video inference provides contextual awareness for the system. The limited resource capabilities of IoT devices and edge computing servers, often used to support computation-intensive IoT tasks, might not be enough to process video content produced by IoT devices in a computer vision-based smart system. In contrast to state-of-the-art where IoT video inference is performed locally at IoT devices or at edge and cloud servers, we propose a novel collaborative IoT paradigm where IoT devices share their idle resources for the processing of video frames in video analytics systems. We proposed a novel stochastic framework for modeling scenarios of collaborative IoT and edge/cloud continuum for video analytics systems. The proposed mathematical framework considers the unique characteristics of video analytics systems, IoT devices, and edge and cloud servers used to process video flows from IoT cameras in a collaborative manner. The obtained results show that the collaborative processing at neighboring IoT devices, i.e., IoT helpers, contributes to reduce the overall latency for video inference. However, high offloading costs might becoming a limiting factor which would request the design of more efficient offloading strategies.</div></div>","PeriodicalId":55555,"journal":{"name":"Ad Hoc Networks","volume":"184 ","pages":"Article 104149"},"PeriodicalIF":4.8,"publicationDate":"2026-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146023875","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"FS-MCS: A reinforcement learning-based data inference scheme for sparse mobile crowd sensing","authors":"Yuan Tao ,&nbsp;Taochun Wang ,&nbsp;Fulong Chen ,&nbsp;Biao Jie ,&nbsp;Junmei Cai ,&nbsp;Dong Xie","doi":"10.1016/j.adhoc.2025.104109","DOIUrl":"10.1016/j.adhoc.2025.104109","url":null,"abstract":"<div><div>With the development of mobile computing and the Internet of Things (IoT), Sparse mobile crowd sensing (SMCS) has emerged as a new data collection paradigm, demonstrating significant application potential in fields. However, due to budget constraints and area inaccessibility, maximizing the quality of inferred data with limited sensing users and resources has become an important research challenge. This paper proposes a fix-based data inference method (FS-MCS) based on deep reinforcement learning, which aims to reduce error accumulation and improve the accuracy of data inference by optimizing the cell selection strategy. Specifically, FS-MCS uses Kriging interpolation to compute the weights of spatiotemporal data and combines Deep Q Learning to dynamically select sensing cells. This allows for the collection of the most informative data in each time series, fixing the inference model and inferring the remaining data. By considering the data correlation over different time periods and budget constraints, FS-MCS can maximize data quality and minimize the impact of outliers on inference results, while ensuring the budget is adhered to. The results show that the proposed method performs well in terms of data quality, convergence speed, and budget utilization, especially when dealing with complex spatiotemporal data and dynamic environmental changes, where it demonstrates significant advantages.</div></div>","PeriodicalId":55555,"journal":{"name":"Ad Hoc Networks","volume":"183 ","pages":"Article 104109"},"PeriodicalIF":4.8,"publicationDate":"2026-03-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145694384","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"PriV2I: Privacy-preserving V2I authentication protocol with fine-grained access control","authors":"Zhengze Liu ,&nbsp;Nianmin Yao ,&nbsp;Shengyuan Bai ,&nbsp;Tengyi Mai","doi":"10.1016/j.adhoc.2025.104122","DOIUrl":"10.1016/j.adhoc.2025.104122","url":null,"abstract":"<div><div>As vehicular ad hoc networks (VANETs) increase in size and complexity, ensuring secure, flexible, and privacy-preserving vehicle-to-infrastructure (V2I) authentication remains a major challenge. Existing protocols often focus solely on identity verification, overlooking the need for access control based on vehicle attributes. Furthermore, vehicles must obtain authentication credentials from various trusted entities, including automakers, regulators, and government agencies. However, the absence of a unified credential issuance mechanism introduces fragmentation and inconsistencies during the registration process. To address these issues, we propose a V2I authentication protocol, called PriV2I, that integrates distributed credential issuance, attribute-based access control, and strong anonymity guarantees. During vehicle registration, our approach uses Shamir’s Secret Sharing with a threshold <span><math><mi>t</mi></math></span> of <span><math><mi>n</mi></math></span> across multiple certification authorities (CAs) to consolidate credentials. A vehicle credential can only be issued by a predefined threshold number of CAs, enhancing security and flexibility. Within the authentication protocol, Pointcheval-Sanders (PS) signatures enable fine-grained access control based on vehicle attributes such as type and role. Meanwhile, noninteractive zero-knowledge proofs protect identity privacy by allowing vehicles to prove credential possession and policy compliance without revealing sensitive information. The proposed scheme also supports batch authentication at Roadside Units (RSUs) to efficiently handle high-density environments and includes a comprehensive revocation mechanism to trace and revoke malicious vehicles promptly and securely. In our implementation, the computation cost during the authentication phase is 75.58 ms. The communication overhead per authentication exchange is 992 bytes across two messages. Overall, the protocol provides a secure, scalable, and privacy-preserving solution tailored to modern VANET environments.</div></div>","PeriodicalId":55555,"journal":{"name":"Ad Hoc Networks","volume":"183 ","pages":"Article 104122"},"PeriodicalIF":4.8,"publicationDate":"2026-03-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145790980","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Collaborative DRL-driven task offloading for maritime edge computing","authors":"Huogen Yang ,&nbsp;Yiwen Hu ,&nbsp;Zhongming Yang ,&nbsp;Xiaohui Yang ,&nbsp;Guangxue Yue","doi":"10.1016/j.adhoc.2025.104124","DOIUrl":"10.1016/j.adhoc.2025.104124","url":null,"abstract":"<div><div>The introduction of Mobile edge computing enables resource-constrained maritime terminal users to access low-latency computing services; however, the dynamic nature of the marine environment and scarce resources render traditional computation offloading strategies inadequate for meeting actual demands, making task offloading a critical issue for achieving prompt and efficient service with optimal resource utilization. In particular, fine-tuning the offloading decision process is crucial for enhancing network stability and extending system endurance. To address these challenges, this paper proposes a deep reinforcement learning-based task offloading method for maritime edge computing. The method derives the optimal transmission power for task offloading and incorporates the power allocation problem into the offloading decision framework, ensuring that offloading decisions are efficiently executed within a specific power range. We model the task offloading problem as a Markov decision process, and based on this formulation, we design an improved Double Deep Q-Network (Double DQN) Energy-Delay Tradeoff Optimization algorithm (ID-EDTO), which enables the system to dynamically obtain state feedback from task requests and adapt its offloading strategies accordingly. Experimental results demonstrate that the proposed method outperforms both traditional baseline methods, such as random selection, Lyapunov optimization, and joint resource allocation, as well as DRL based algorithms including PPO, SAC, and A3C, in terms of reducing latency and energy consumption.</div></div>","PeriodicalId":55555,"journal":{"name":"Ad Hoc Networks","volume":"183 ","pages":"Article 104124"},"PeriodicalIF":4.8,"publicationDate":"2026-03-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145791036","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Joint channel connectivity and interference management in DT-assisted cognitive vehicular networks","authors":"Xuan Li ,&nbsp;Wen Jiang ,&nbsp;Wanting Wang ,&nbsp;Tianqing Zhou ,&nbsp;Kan Wang ,&nbsp;Shuai Liu","doi":"10.1016/j.adhoc.2025.104094","DOIUrl":"10.1016/j.adhoc.2025.104094","url":null,"abstract":"<div><div>In cognitive vehicular networks (CVNs), cognitive vehicles are permitted to opportunistically utilize idle spectrum bands. However, the reclaiming of channels by licensed users may result in significant interference or even network disconnection, failing to meet the reliable data transmission requirements in CVNs. Vehicles need to frequently exchange a large amount of information to cope with unpredictable topology changes and channel reuse scenarios, resulting in significant communication and computational overhead, which conflicts with the low-latency requirements of vehicular networks. To address this, we introduce digital twin (DT) technology into CVNs, enabling cognitive vehicles to effectively avoid transmission interruption caused by primary user channel occupancy. First, we propose a DT-assisted connectivity algorithm (DT-CA) that maps real-world vehicular networks to their digital replicas, enabling interaction in the virtual world. DT-CA assists vehicles in forming specific clusters to ensure channel connectivity. Subsequently, we propose a vehicle-to-vehicle (V2V) connectivity algorithm that quantifies vehicle mobility using communication probabilities and dynamically optimizes cluster structures. Finally, we conduct extensive simulation studies in different traffic scenarios, such as T-junctions and crossroads, which demonstrate that the DT-assisted algorithms have significant advantages in enhancing the connectivity and cluster stability of CVNs, while also exhibiting dynamic adaptability and low complexity.</div></div>","PeriodicalId":55555,"journal":{"name":"Ad Hoc Networks","volume":"183 ","pages":"Article 104094"},"PeriodicalIF":4.8,"publicationDate":"2026-03-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145790981","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"A dual layer LSTM-CNN framework for real time and precise per-message intrusion detection in In-vehicle networks","authors":"Yu Fu,&nbsp;Junhui She,&nbsp;Yinan Xu,&nbsp;Yihu Xu,&nbsp;Ziyi Wang,&nbsp;Yujing Wu","doi":"10.1016/j.adhoc.2025.104119","DOIUrl":"10.1016/j.adhoc.2025.104119","url":null,"abstract":"<div><div>The rapid proliferation of Intelligent Connected Vehicles (ICVs) presents escalating cybersecurity challenges, particularly within Controller Area Networks (CAN), where traditional Intrusion Detection Systems (IDS) often fail to meet the stringent requirements for real-time and fine-grained anomaly detection under limited computational resources. This study introduces a novel lightweight dual-tier anomaly detection framework, termed Enhanced LSTM-CNN (ELC), which synergistically integrates temporal and spatial deep learning paradigms to address these limitations. The first tier employs an optimized hybrid architecture combining Long Short-Term Memory (LSTM) networks for temporal dependency modeling and Convolutional Neural Networks (CNNs) for spatial feature extraction, enabling rapid and accurate preliminary anomaly screening. The second tier utilizes an enhanced CNN classifier to perform refined multi-class identification of four prevalent attack types, including Denial of Service (DoS), Fuzzy, and RPM/GEAR spoofing, achieving an F1-score of 99.8 %. Comprehensive evaluations on real-world vehicular CAN datasets demonstrate that ELC attains an average per-message detection of 0.153 ms and sustains a processing throughput of 7000 messages per second, all within a power envelope of 7.3 W making it well suited for deployment in resource-constrained Electronic Control Units (ECUs). In addition, we validate ELC on the public 4TU CAN Bus Intrusion Dataset v2 and Survival Analysis Dataset maintaining comparable performance under cross-dataset settings and underscoring generalization and reproducibility. Unlike conventional batch-based approaches, ELC provides message-level granularity and sub-millisecond responsiveness, thereby ensuring timely threat mitigation within the 10 ms message interval constraints of CAN systems. These results indicate that the proposed framework holds strong potential as a practical and effective solution for real-time, embedded intrusion detection in resource-constrained vehicular environments.</div></div>","PeriodicalId":55555,"journal":{"name":"Ad Hoc Networks","volume":"183 ","pages":"Article 104119"},"PeriodicalIF":4.8,"publicationDate":"2026-03-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145738228","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"UpsFed-IDS: U-shaped split federated intrusion detection system for securing UAV communication in dynamic networks","authors":"Zongpu Wei,&nbsp;Jinsong Wang,&nbsp;Zening Zhao,&nbsp;Zhao Zhao,&nbsp;Kai Shi","doi":"10.1016/j.adhoc.2025.104133","DOIUrl":"10.1016/j.adhoc.2025.104133","url":null,"abstract":"<div><div>Integrating an intrusion detection system (IDS) into UAVs is critical for safeguarding their operational reliability and overall security. Centralized IDS deployed in data centers has become impractical, primarily due to concerns over data privacy and computational constraints. Federated learning (FL)-based IDS alleviates the data leakage issue inherent in traditional IDS. Nevertheless, its integration with UAV systems still encounters unavoidable challenges. Firstly, the requirement for local model training on UAVs imposes substantial computational overhead. Secondly, the non-independent and identically distributed (non-IID) data characteristics of UAVs directly impair the performance of the IDS model. Thirdly, the constant dynamic changes in UAV network connectivity undermine the robustness of the federated IDS. To address these challenges, this paper presents a U-shaped split federated intrusion detection system (UpsFed-IDS) for securing UAV communication. Inspired by FL and Split Learning (SL), we offload a portion of the IDS model training to the Ground Control Station (GCS). This approach ensures that raw data and labels remain on the UAVs, which enhances data privacy protection and reduces the computational overhead on the UAV side. Within this system, we propose a split-specific head personalization method to decouple global feature learning from local model personalization under the SL scheme, which strengthens the IDS model performance in heterogeneous data scenarios. Furthermore, a client failover mechanism is designed to tackle disconnections occurring during training in dynamic UAV networks, which effectively improves the overall robustness of the system. Extensive experimental evaluations are conducted on the UAVCAN attack and WSN-DS datasets. The results demonstrate that UpsFed-IDS outperforms existing FL frameworks in both attack recognition performance and local computation overhead.</div></div>","PeriodicalId":55555,"journal":{"name":"Ad Hoc Networks","volume":"183 ","pages":"Article 104133"},"PeriodicalIF":4.8,"publicationDate":"2026-03-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145884183","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}