Ad Hoc Networks最新文献

A DEPMU-based network traffic anomaly detection scheme for IoT 基于depmu的物联网网络流量异常检测方案

IF 4.8 3区计算机科学 Q1 COMPUTER SCIENCE, INFORMATION SYSTEMS

Ad Hoc Networks

Pub Date : 2026-04-15 Epub Date: 2026-01-29 DOI: 10.1016/j.adhoc.2026.104160

Yueling Liu, Chunhai Li, Changsong Yang, Yong Ding

Thanks to the rapid development and widespread popularity of wireless network technology, Internet of Things (IoT) has been broadly used by the public in the daily life and work due to its convenience, low delay and high-efficiency. Despite plenty of tremendous advantages, IoT also suffers from some serious security problems and technology issues, for instance, dishonest user attack, malicious hacker intrusion, etc. For discovering malicious attacks, network traffic anomaly detection (NTAD) system has been deployed in IoT. However, in IoT, the network traffic data is characterized by massive, irregularity, temporal correlation, multiple feature and high dimensionality. These characteristics will greatly reduce the detection performance of NTAD. In this article, to solve the above issues, we aim to design a new NTAD scheme. Specifically, inspired by the traditional parsimonious memory unit (PMU), we design a new neural network model called deep encoder parsimonious memory unit (DEPMU), which consists of the encoding parsimonious memory unit (EPMU), the decoding parsimonious memory unit (DPMU), the loss compensation parsimonious memory unit (LEPMU), and two loss functions. Compared with the original PMU, DEPMU can better characterize and learn the time-series data, and can reduce the feature loss by adding a loss compensation mechanism. Subsequently, we adopt DEPMU to design a NTAD scheme for IoT, which can greatly improve the anomaly detection performance. Meanwhile, we prove the high efficiency of our scheme through computational complexity analysis. Finally, we also develop a prototype system and implement our scheme to test the overall performance. We can discover from the experimental results that our scheme can achieve better performance compared with some existing schemes.

随着无线网络技术的快速发展和广泛普及，物联网（Internet of Things, IoT）以其便捷、低时延、高效等特点被大众广泛应用于日常生活和工作中。尽管物联网有很多巨大的优势，但也存在一些严重的安全问题和技术问题，例如，不诚实的用户攻击，恶意黑客入侵等。为了发现恶意攻击，网络流量异常检测（nad）系统被部署在物联网中。而在物联网中，网络流量数据具有海量、不规则、时间相关、多特征、高维等特点。这些特性将大大降低nad的检测性能。为了解决上述问题，本文旨在设计一种新的NTAD方案。具体而言，我们在传统简约记忆单元（PMU）的启发下，设计了一种新的神经网络模型，称为深度编码器简约记忆单元（DEPMU），该模型由编码简约记忆单元（EPMU）、解码简约记忆单元（DPMU）、损失补偿简约记忆单元（LEPMU）和两个损失函数组成。与原有的PMU相比，DEPMU可以更好地表征和学习时间序列数据，并通过增加损失补偿机制减少特征损失。随后，我们采用DEPMU设计了一种针对物联网的NTAD方案，该方案可以大大提高异常检测性能。同时，通过计算复杂度分析证明了该方案的高效性。最后，我们还开发了一个原型系统，并实现了我们的方案来测试整体性能。实验结果表明，与现有的一些方案相比，我们的方案具有更好的性能。

{"title":"A DEPMU-based network traffic anomaly detection scheme for IoT","authors":"Yueling Liu, Chunhai Li, Changsong Yang, Yong Ding","doi":"10.1016/j.adhoc.2026.104160","DOIUrl":"10.1016/j.adhoc.2026.104160","url":null,"abstract":"<div><div>Thanks to the rapid development and widespread popularity of wireless network technology, Internet of Things (IoT) has been broadly used by the public in the daily life and work due to its convenience, low delay and high-efficiency. Despite plenty of tremendous advantages, IoT also suffers from some serious security problems and technology issues, for instance, dishonest user attack, malicious hacker intrusion, etc. For discovering malicious attacks, network traffic anomaly detection (NTAD) system has been deployed in IoT. However, in IoT, the network traffic data is characterized by massive, irregularity, temporal correlation, multiple feature and high dimensionality. These characteristics will greatly reduce the detection performance of NTAD. In this article, to solve the above issues, we aim to design a new NTAD scheme. Specifically, inspired by the traditional parsimonious memory unit (PMU), we design a new neural network model called deep encoder parsimonious memory unit (DEPMU), which consists of the encoding parsimonious memory unit (EPMU), the decoding parsimonious memory unit (DPMU), the loss compensation parsimonious memory unit (LEPMU), and two loss functions. Compared with the original PMU, DEPMU can better characterize and learn the time-series data, and can reduce the feature loss by adding a loss compensation mechanism. Subsequently, we adopt DEPMU to design a NTAD scheme for IoT, which can greatly improve the anomaly detection performance. Meanwhile, we prove the high efficiency of our scheme through computational complexity analysis. Finally, we also develop a prototype system and implement our scheme to test the overall performance. We can discover from the experimental results that our scheme can achieve better performance compared with some existing schemes.</div></div>","PeriodicalId":55555,"journal":{"name":"Ad Hoc Networks","volume":"185 ","pages":"Article 104160"},"PeriodicalIF":4.8,"publicationDate":"2026-04-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146175112","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}

引用次数: 0

Cross-layer joint optimization for semantic communication-driven MEC systems via deep reinforcement learning 基于深度强化学习的语义通信驱动MEC系统跨层联合优化

IF 4.8 3区计算机科学 Q1 COMPUTER SCIENCE, INFORMATION SYSTEMS

Ad Hoc Networks

Pub Date : 2026-04-15 Epub Date: 2026-01-29 DOI: 10.1016/j.adhoc.2026.104159

Meiyao Wen, Linyu Huang, Qian Ning

The integration of semantic communication (SemCom) with mobile edge computing (MEC) has opened new avenues to improve task execution efficiency in intelligent networks. This paper proposes a cross-layer joint optimization framework for SemCom-driven MEC systems, aiming to minimize the weighted sum of task completion time and user energy consumption. Specifically, the framework jointly optimizes the semantic extraction factor at the application layer, task offloading decisions at the control layer, and communication and computational resource allocation at the network and physical layers. To address the non-convex and mixed-integer nature of the problem, a Deep Deterministic Policy Gradient (DDPG)-based algorithm was employed to efficiently search for solutions. The simulation results validate the effectiveness of the proposed approach and demonstrate that the integration of SemCom into MEC significantly improves the system performance. The findings offer practical insights for system engineers to design efficient MEC systems, reducing transmission overhead and energy consumption, especially in latency-sensitive applications such as autonomous driving and industrial Internet of Things.

语义通信（SemCom）与移动边缘计算（MEC）的融合为提高智能网络中的任务执行效率开辟了新的途径。针对semcom驱动的MEC系统，提出了一种以任务完成时间和用户能耗加权和最小为目标的跨层联合优化框架。具体而言，该框架共同优化了应用层的语义提取因子、控制层的任务卸载决策以及网络层和物理层的通信和计算资源分配。为了解决该问题的非凸和混合整数性质，采用基于深度确定性策略梯度（Deep Deterministic Policy Gradient， DDPG）的算法高效地搜索解。仿真结果验证了该方法的有效性，并表明将SemCom集成到MEC中可以显著提高系统性能。这些发现为系统工程师设计高效的MEC系统提供了实用的见解，降低了传输开销和能耗，特别是在自动驾驶和工业物联网等对延迟敏感的应用中。

{"title":"Cross-layer joint optimization for semantic communication-driven MEC systems via deep reinforcement learning","authors":"Meiyao Wen, Linyu Huang, Qian Ning","doi":"10.1016/j.adhoc.2026.104159","DOIUrl":"10.1016/j.adhoc.2026.104159","url":null,"abstract":"<div><div>The integration of semantic communication (SemCom) with mobile edge computing (MEC) has opened new avenues to improve task execution efficiency in intelligent networks. This paper proposes a cross-layer joint optimization framework for SemCom-driven MEC systems, aiming to minimize the weighted sum of task completion time and user energy consumption. Specifically, the framework jointly optimizes the semantic extraction factor at the application layer, task offloading decisions at the control layer, and communication and computational resource allocation at the network and physical layers. To address the non-convex and mixed-integer nature of the problem, a Deep Deterministic Policy Gradient (DDPG)-based algorithm was employed to efficiently search for solutions. The simulation results validate the effectiveness of the proposed approach and demonstrate that the integration of SemCom into MEC significantly improves the system performance. The findings offer practical insights for system engineers to design efficient MEC systems, reducing transmission overhead and energy consumption, especially in latency-sensitive applications such as autonomous driving and industrial Internet of Things.</div></div>","PeriodicalId":55555,"journal":{"name":"Ad Hoc Networks","volume":"185 ","pages":"Article 104159"},"PeriodicalIF":4.8,"publicationDate":"2026-04-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146098529","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}

引用次数: 0

Improving object selection for Collective Perception Messages under congestion 在拥塞条件下改进集体感知信息的对象选择

IF 4.8 3区计算机科学 Q1 COMPUTER SCIENCE, INFORMATION SYSTEMS

Ad Hoc Networks

Pub Date : 2026-04-15 Epub Date: 2026-02-05 DOI: 10.1016/j.adhoc.2026.104175

Andreia Figueiredo , João Amaral , Pedro Rito , Miguel Luís , Susana Sargento

Collective Perception Messages (CPMs), defined by European Telecommunications Standards Institute (ETSI), enable vehicles and roadside infrastructure to exchange information about detected objects, enhancing situational awareness in cooperative environments. However, as the size of CPMs increases — particularly in dense traffic scenarios — the wireless channel can become saturated, leading to delays in transmission and reduced packet delivery ratios. This paper starts by assessing how the number of objects included per CPM impacts communication performance, highlighting the necessity for effective object selection strategies during periods of congestion. To address this issue, we propose a lightweight, real-time object prioritization algorithm based on deviations from the predicted path. Our method estimates each object’s expected state based on its last transmission, and prioritizes those whose current state deviates most from this prediction, as these are likely to be more informative. The evaluation uses a real-world dataset and demonstrates that our strategy significantly improves predictive accuracy by at least 7%. Moreover, the algorithm does not increase CPU or memory usage, demonstrating similar resource consumption compared to the method described in the Collective Perception Service (CPS) standard, making it well-suited for embedded platforms. These results confirm that Prediction–Deviation selection can enhance the efficiency and informativeness of CPMs, especially when the message size must be constrained due to network congestion.

集体感知信息（cpm）由欧洲电信标准协会（ETSI）定义，使车辆和路边基础设施能够交换有关检测到的物体的信息，增强协作环境中的态势感知。然而，随着cpm规模的增加——尤其是在流量密集的情况下——无线信道可能会变得饱和，从而导致传输延迟和数据包传送率降低。本文首先评估了每个CPM包含的对象数量如何影响通信性能，强调了在拥塞期间有效的对象选择策略的必要性。为了解决这个问题，我们提出了一种基于预测路径偏差的轻量级实时对象优先排序算法。我们的方法根据每个对象的最后一次传输估计其预期状态，并优先考虑当前状态与此预测偏差最大的对象，因为这些对象可能提供更多信息。评估使用了一个真实世界的数据集，并证明我们的策略显著提高了至少7%的预测准确性。此外，该算法不会增加CPU或内存使用，与集体感知服务（CPS）标准中描述的方法相比，显示出相似的资源消耗，使其非常适合嵌入式平台。这些结果证实了预测偏差选择可以提高cpm的效率和信息量，特别是当由于网络拥塞而必须限制消息大小时。

{"title":"Improving object selection for Collective Perception Messages under congestion","authors":"Andreia Figueiredo , João Amaral , Pedro Rito , Miguel Luís , Susana Sargento","doi":"10.1016/j.adhoc.2026.104175","DOIUrl":"10.1016/j.adhoc.2026.104175","url":null,"abstract":"<div><div>Collective Perception Messages (CPMs), defined by European Telecommunications Standards Institute (ETSI), enable vehicles and roadside infrastructure to exchange information about detected objects, enhancing situational awareness in cooperative environments. However, as the size of CPMs increases — particularly in dense traffic scenarios — the wireless channel can become saturated, leading to delays in transmission and reduced packet delivery ratios. This paper starts by assessing how the number of objects included per CPM impacts communication performance, highlighting the necessity for effective object selection strategies during periods of congestion. To address this issue, we propose a lightweight, real-time object prioritization algorithm based on deviations from the predicted path. Our method estimates each object’s expected state based on its last transmission, and prioritizes those whose current state deviates most from this prediction, as these are likely to be more informative. The evaluation uses a real-world dataset and demonstrates that our strategy significantly improves predictive accuracy by at least 7%. Moreover, the algorithm does not increase CPU or memory usage, demonstrating similar resource consumption compared to the method described in the Collective Perception Service (CPS) standard, making it well-suited for embedded platforms. These results confirm that Prediction–Deviation selection can enhance the efficiency and informativeness of CPMs, especially when the message size must be constrained due to network congestion.</div></div>","PeriodicalId":55555,"journal":{"name":"Ad Hoc Networks","volume":"185 ","pages":"Article 104175"},"PeriodicalIF":4.8,"publicationDate":"2026-04-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146175098","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}

引用次数: 0

HetTraffic: Multi-link traffic prediction and allocation for 6G heterogeneous networks HetTraffic：针对6G异构网络的多链路流量预测与分配

IF 4.8 3区计算机科学 Q1 COMPUTER SCIENCE, INFORMATION SYSTEMS

Ad Hoc Networks

Pub Date : 2026-04-15 Epub Date: 2026-01-31 DOI: 10.1016/j.adhoc.2026.104153

Yali Lv , Jian Huang , Jingpu Duan , Yaping Sun , Xiong Li

The rapid evolution of wireless communication necessitates advanced solutions beyond current 5G capabilities to realize the ambitious vision of 6G. The forthcoming 6G era will witness an unprecedented scale of device connectivity, challenging conventional resource allocation paradigms with its inherent heterogeneity and dynamic nature. A key issue involves intelligently and dynamically assigning diverse user traffic to highly heterogeneous links, while still satisfying Quality of Service (QoS) requirements. Moreover, resource management strategies that rely solely on reactive real-time measurements often lead to suboptimal performance. To overcome these limitations, this paper proposes HetTraffic, a novel comprehensive framework for joint traffic prediction and allocation in 6G heterogeneous networks. HetTraffic first introduces a novel link-level traffic prediction method leveraging a hybrid Graph Attention Network (GAT) and Long Short-Term Memory (LSTM) architecture. This approach effectively captures both the complex spatial dependencies from user mobility and the temporal fluctuations within traffic data. Building upon these predictions, we develop a multi-agent reinforcement learning-based allocation strategy utilizing the Multi-Agent Proximal Policy Optimization (MAPPO) algorithm. This is designed for efficient, decentralized resource optimization across heterogeneous links, proactively accounting for real-time conditions, QoS demands, and predicted traffic. Comprehensive experiments conducted on a dedicated 6G heterogeneous network testbed, utilizing a curated link-level traffic dataset, demonstrate the significant advantages and superior performance of our proposed traffic prediction and allocation methods compared to existing state-of-the-art approaches.

无线通信的快速发展需要超越当前5G功能的先进解决方案来实现6G的宏伟愿景。即将到来的6G时代将见证前所未有的设备连接规模，以其固有的异质性和动态性挑战传统的资源分配模式。一个关键问题涉及智能和动态地将不同的用户流量分配到高度异构的链路，同时仍然满足服务质量（QoS）要求。此外，仅依赖于响应式实时度量的资源管理策略通常会导致次优性能。为了克服这些限制，本文提出了一种新的综合框架HetTraffic，用于6G异构网络的联合流量预测和分配。HetTraffic首先介绍了一种利用混合图注意网络（GAT）和长短期记忆（LSTM）架构的链路级流量预测方法。这种方法有效地捕获了用户移动的复杂空间依赖性和交通数据中的时间波动。在这些预测的基础上，我们利用多智能体近端策略优化（MAPPO）算法开发了一种基于多智能体强化学习的分配策略。它旨在跨异构链路进行高效、分散的资源优化，主动考虑实时条件、QoS需求和预测流量。在专用的6G异构网络测试平台上进行的综合实验，利用精心策划的链路级流量数据集，与现有的最先进的方法相比，我们提出的流量预测和分配方法具有显著的优势和卓越的性能。

{"title":"HetTraffic: Multi-link traffic prediction and allocation for 6G heterogeneous networks","authors":"Yali Lv , Jian Huang , Jingpu Duan , Yaping Sun , Xiong Li","doi":"10.1016/j.adhoc.2026.104153","DOIUrl":"10.1016/j.adhoc.2026.104153","url":null,"abstract":"<div><div>The rapid evolution of wireless communication necessitates advanced solutions beyond current 5G capabilities to realize the ambitious vision of 6G. The forthcoming 6G era will witness an unprecedented scale of device connectivity, challenging conventional resource allocation paradigms with its inherent heterogeneity and dynamic nature. A key issue involves intelligently and dynamically assigning diverse user traffic to highly heterogeneous links, while still satisfying Quality of Service (QoS) requirements. Moreover, resource management strategies that rely solely on reactive real-time measurements often lead to suboptimal performance. To overcome these limitations, this paper proposes <em>HetTraffic</em>, a novel comprehensive framework for joint traffic prediction and allocation in 6G heterogeneous networks. <em>HetTraffic</em> first introduces a novel link-level traffic prediction method leveraging a hybrid Graph Attention Network (GAT) and Long Short-Term Memory (LSTM) architecture. This approach effectively captures both the complex spatial dependencies from user mobility and the temporal fluctuations within traffic data. Building upon these predictions, we develop a multi-agent reinforcement learning-based allocation strategy utilizing the Multi-Agent Proximal Policy Optimization (MAPPO) algorithm. This is designed for efficient, decentralized resource optimization across heterogeneous links, proactively accounting for real-time conditions, QoS demands, and predicted traffic. Comprehensive experiments conducted on a dedicated 6G heterogeneous network testbed, utilizing a curated link-level traffic dataset, demonstrate the significant advantages and superior performance of our proposed traffic prediction and allocation methods compared to existing state-of-the-art approaches.</div></div>","PeriodicalId":55555,"journal":{"name":"Ad Hoc Networks","volume":"185 ","pages":"Article 104153"},"PeriodicalIF":4.8,"publicationDate":"2026-04-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146098530","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}

引用次数: 0

Target localization in UAV swarm under multi-error coupling: A cooperative utility of information optimization approach 多误差耦合下无人机群目标定位：一种信息优化协同效用方法

IF 4.8 3区计算机科学 Q1 COMPUTER SCIENCE, INFORMATION SYSTEMS

Ad Hoc Networks

Pub Date : 2026-04-15 Epub Date: 2026-01-22 DOI: 10.1016/j.adhoc.2026.104154

Zou Zhou, Zuozhun Qin, Jie Peng, Hongbing Qiu, Junyi Wang

In complex electromagnetic environments, the scale effect of Unmanned Aerial Vehicle (UAV) swarm presents significant potential for enhancing cooperative effectiveness. However, the accuracy of Time Difference of Arrival (TDOA)-based localization for non-cooperative emitters using UAV swarm is significantly affected by the coupling of multi-source errors, which mainly include UAV position error (UPE), clock synchronization error (CSE), and TDOA measurement error (TME). To address the challenges of evaluating cooperative effectiveness under multi-source errors coupling and balancing localization accuracy with computational efficiency, a cooperative utility of information (CUoI) optimization approach is proposed. First, a TDOA observation uncertainty model is constructed by integrating multi-source errors. Then, the information gain of target position estimation is derived to build the CUoI evaluation model. Next, the characteristic of Dueling Deep Q-Network (Dueling DQN) that decouples state value from action advantage is leveraged, enabling precise evaluation of the potential benefits of different hyperparameter adjustment strategies. This characteristic facilitates adaptive tuning of key hyperparameters in Particle Swarm Optimization (PSO). Finally, a dynamic PSO framework based on Dueling DQN is proposed to effectively balance localization accuracy and computational efficiency. Numerical experiments demonstrate that the proposed algorithm achieves reductions in average localization RMSE of 19.1%, 6.0%, and 1.4%, respectively, compared to Semidefinite Relaxation-TDOA (SDR-TDOA), Grey Wolf Optimizer (GWO), and Multi-swarm Discrete Quantum-inspired Particle Swarm Optimization with Adaptive Simulated Annealing (MDQPSO-ASA).

在复杂电磁环境下，无人机群的规模效应对提高协同效能具有重要的潜力。然而，利用无人机群对非合作发射体进行TDOA定位的精度受到多源误差耦合的显著影响，多源误差主要包括无人机位置误差（UPE）、时钟同步误差（CSE）和TDOA测量误差（TME）。为解决多源误差耦合下的协同效果评估问题，以及平衡定位精度和计算效率的问题，提出了一种信息协同效用优化方法。首先，综合多源误差，建立了TDOA观测不确定性模型；然后，导出目标位置估计的信息增益，建立cui评价模型；接下来，利用Dueling Deep Q-Network （Dueling DQN）将状态值与动作优势解耦的特性，能够精确评估不同超参数调整策略的潜在收益。这一特性为粒子群优化（PSO）中关键超参数的自适应调整提供了便利。最后，提出了一种基于Dueling DQN的动态粒子群算法框架，有效地平衡了定位精度和计算效率。数值实验表明，与半确定松弛- tdoa （SDR-TDOA）、灰狼优化（GWO）和多群离散量子启发粒子群自适应模拟退火优化（MDQPSO-ASA）相比，该算法的平均定位RMSE分别降低了19.1%、6.0%和1.4%。

{"title":"Target localization in UAV swarm under multi-error coupling: A cooperative utility of information optimization approach","authors":"Zou Zhou, Zuozhun Qin, Jie Peng, Hongbing Qiu, Junyi Wang","doi":"10.1016/j.adhoc.2026.104154","DOIUrl":"10.1016/j.adhoc.2026.104154","url":null,"abstract":"<div><div>In complex electromagnetic environments, the scale effect of Unmanned Aerial Vehicle (UAV) swarm presents significant potential for enhancing cooperative effectiveness. However, the accuracy of Time Difference of Arrival (TDOA)-based localization for non-cooperative emitters using UAV swarm is significantly affected by the coupling of multi-source errors, which mainly include UAV position error (UPE), clock synchronization error (CSE), and TDOA measurement error (TME). To address the challenges of evaluating cooperative effectiveness under multi-source errors coupling and balancing localization accuracy with computational efficiency, a cooperative utility of information (CUoI) optimization approach is proposed. First, a TDOA observation uncertainty model is constructed by integrating multi-source errors. Then, the information gain of target position estimation is derived to build the CUoI evaluation model. Next, the characteristic of Dueling Deep Q-Network (Dueling DQN) that decouples state value from action advantage is leveraged, enabling precise evaluation of the potential benefits of different hyperparameter adjustment strategies. This characteristic facilitates adaptive tuning of key hyperparameters in Particle Swarm Optimization (PSO). Finally, a dynamic PSO framework based on Dueling DQN is proposed to effectively balance localization accuracy and computational efficiency. Numerical experiments demonstrate that the proposed algorithm achieves reductions in average localization RMSE of 19.1%, 6.0%, and 1.4%, respectively, compared to Semidefinite Relaxation-TDOA (SDR-TDOA), Grey Wolf Optimizer (GWO), and Multi-swarm Discrete Quantum-inspired Particle Swarm Optimization with Adaptive Simulated Annealing (MDQPSO-ASA).</div></div>","PeriodicalId":55555,"journal":{"name":"Ad Hoc Networks","volume":"185 ","pages":"Article 104154"},"PeriodicalIF":4.8,"publicationDate":"2026-04-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146175101","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}

引用次数: 0

Multi-agent DRL-based task offloading and trajectory optimization for low altitude UAV IoT systems 基于多智能体drl的低空无人机物联网系统任务卸载与轨迹优化

IF 4.8 3区计算机科学 Q1 COMPUTER SCIENCE, INFORMATION SYSTEMS

Ad Hoc Networks

Pub Date : 2026-04-15 Epub Date: 2026-02-03 DOI: 10.1016/j.adhoc.2026.104164

Shanchen Pang , Miaomiao Fan , Xiao He , Wenhao Ji , Sibo Qiao , Chenhao Zhang

In low-altitude Internet of Things (IoT) networks, the Unmanned Aerial Vehicle (UAV) is employed as a mobile edge node to provide computational services for task processing. However, the spatio-temporal dynamics of User Devices (UDs) and the heterogeneity of task prioritization exacerbate the multidimensional resource competition encountered during the processing of tasks. This significantly affects energy consumption, service delay, and task completion rate, degrading user Quality of Service (QoS). To address these challenges, we propose a collaborative Multi-Agent Deep Reinforcement Learning (MADRL) algorithm to improve user QoS through the joint optimization of UAV three-Dimensional (3D) trajectories, resource allocation, and task offloading strategies. Specifically, we design a Graph Convolutional Network (GCN)-based UAV actor network to optimize the dynamic trajectory by modeling user distribution in a topology-aware manner. In addition, we construct a centralized critic network based on a multi-head attention mechanism, wherein attention scaling is utilized to quantify differences in task demands and guide resource decision-making. These two components are jointly optimized through a ”topology association–demand difference” cooperative evaluation mechanism, enabling a multi-dimensional coupling of spatio-temporal characteristics and task demand decision-making. Experimental results demonstrate that the proposed algorithm reduces system energy consumption and delay by approximately 18.5% and 22.7%, respectively, while improving the task completion rate by about 16.2%.

在低空物联网（IoT）网络中，无人机（UAV）作为移动边缘节点，为任务处理提供计算服务。然而，用户设备的时空动态和任务优先级的异质性加剧了任务处理过程中遇到的多维资源竞争。这会严重影响能耗、业务延迟和任务完成率，降低用户的QoS （Quality of service）。为了解决这些挑战，我们提出了一种协作式多智能体深度强化学习（MADRL）算法，通过联合优化无人机三维（3D）轨迹、资源分配和任务卸载策略来提高用户QoS。具体来说，我们设计了一个基于图卷积网络（GCN）的无人机行动者网络，通过拓扑感知的方式对用户分布进行建模，优化动态轨迹。此外，我们构建了一个基于多头注意机制的集中式批评网络，其中使用注意尺度来量化任务需求差异并指导资源决策。通过“拓扑关联-需求差异”协同评价机制共同优化这两个分量，实现了时空特征与任务需求决策的多维耦合。实验结果表明，该算法将系统能耗和延迟分别降低约18.5%和22.7%，将任务完成率提高约16.2%。

{"title":"Multi-agent DRL-based task offloading and trajectory optimization for low altitude UAV IoT systems","authors":"Shanchen Pang , Miaomiao Fan , Xiao He , Wenhao Ji , Sibo Qiao , Chenhao Zhang","doi":"10.1016/j.adhoc.2026.104164","DOIUrl":"10.1016/j.adhoc.2026.104164","url":null,"abstract":"<div><div>In low-altitude Internet of Things (IoT) networks, the Unmanned Aerial Vehicle (UAV) is employed as a mobile edge node to provide computational services for task processing. However, the spatio-temporal dynamics of User Devices (UDs) and the heterogeneity of task prioritization exacerbate the multidimensional resource competition encountered during the processing of tasks. This significantly affects energy consumption, service delay, and task completion rate, degrading user Quality of Service (QoS). To address these challenges, we propose a collaborative Multi-Agent Deep Reinforcement Learning (MADRL) algorithm to improve user QoS through the joint optimization of UAV three-Dimensional (3D) trajectories, resource allocation, and task offloading strategies. Specifically, we design a Graph Convolutional Network (GCN)-based UAV actor network to optimize the dynamic trajectory by modeling user distribution in a topology-aware manner. In addition, we construct a centralized critic network based on a multi-head attention mechanism, wherein attention scaling is utilized to quantify differences in task demands and guide resource decision-making. These two components are jointly optimized through a ”topology association–demand difference” cooperative evaluation mechanism, enabling a multi-dimensional coupling of spatio-temporal characteristics and task demand decision-making. Experimental results demonstrate that the proposed algorithm reduces system energy consumption and delay by approximately 18.5% and 22.7%, respectively, while improving the task completion rate by about 16.2%.</div></div>","PeriodicalId":55555,"journal":{"name":"Ad Hoc Networks","volume":"185 ","pages":"Article 104164"},"PeriodicalIF":4.8,"publicationDate":"2026-04-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146175099","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}

引用次数: 0

DDPG-based data collection for AoI in multi-UAV-assisted IoT networks 基于ddpg的多无人机辅助物联网AoI数据采集

IF 4.8 3区计算机科学 Q1 COMPUTER SCIENCE, INFORMATION SYSTEMS

Ad Hoc Networks

Pub Date : 2026-04-15 Epub Date: 2026-02-01 DOI: 10.1016/j.adhoc.2026.104162

Jianbin Xue , Xiao Li , Zhenqin Wang , Chang Li

Unmanned aerial vehicles (UAVs) are increasingly employed to facilitate effective information acquisition within Internet of Things (IoT) systems due to their superior mobility and operational flexibility. Maintaining the freshness of information in UAV-assisted IoT systems is crucial for real-time monitoring applications, particularly when dealing with stochastic generation patterns of sensory data. Orchestrating multiple energy-constrained UAVs to ensure the temporal validity of collected information poses significant technical challenges due to dynamic mission constraints and limited onboard power supplies. To address this issue, we investigate the problem of information freshness optimization in a multi-UAV collaborative data collection environment, proposes an attention-based deep deterministic policy gradient (A-DDPG) algorithm, constructs a multi-UAV-assisted IoT data collection system model that considers data freshness, communication quality, and energy efficiency, and models the UAV trajectory planning, hovering point selection, and task assignment problems as a Markov decision process. Due to the limitations of the standard DDPG algorithm in handling high-dimensional state spaces and multiple constraints, we introduce an attention layer in the A-DDPG algorithm to enhance the perception of key state features, designs a prioritized experience replay mechanism to enhance data sampling efficiency in reinforcement learning processes., implements normalization strategies based on the characteristics of different state components, and develops action constraint handling methods to ensure that UAV behaviors meet physical constraints. Through comprehensive simulation tests, the proposed algorithm is compared with existing technologies, demonstrating its high effectiveness in terms of average age of information, energy efficiency, and task completion rate.

由于其优越的机动性和操作灵活性，无人机（uav）越来越多地被用于促进物联网（IoT）系统内有效的信息获取。在无人机辅助物联网系统中，保持信息的新鲜度对于实时监控应用至关重要，特别是在处理随机生成模式的感官数据时。由于动态任务约束和机载电源有限，协调多个能量受限的无人机以确保收集信息的时间有效性面临重大技术挑战。针对这一问题，研究了多无人机协同数据采集环境下的信息新鲜度优化问题，提出了一种基于注意力的深度确定性策略梯度（a - ddpg）算法，构建了考虑数据新鲜度、通信质量和能源效率的多无人机辅助物联网数据采集系统模型，并将无人机的轨迹规划、悬停点选择和任务分配问题建模为马尔可夫决策过程。针对标准DDPG算法在处理高维状态空间和多约束条件方面的局限性，本文在a -DDPG算法中引入注意层来增强对关键状态特征的感知，设计优先体验重播机制来提高强化学习过程中的数据采样效率。，根据不同状态分量的特征实现归一化策略，开发动作约束处理方法，保证无人机行为满足物理约束。通过综合仿真测试，将该算法与现有技术进行了比较，证明了该算法在平均信息年龄、能源效率和任务完成率方面具有较高的有效性。

{"title":"DDPG-based data collection for AoI in multi-UAV-assisted IoT networks","authors":"Jianbin Xue , Xiao Li , Zhenqin Wang , Chang Li","doi":"10.1016/j.adhoc.2026.104162","DOIUrl":"10.1016/j.adhoc.2026.104162","url":null,"abstract":"<div><div>Unmanned aerial vehicles (UAVs) are increasingly employed to facilitate effective information acquisition within Internet of Things (IoT) systems due to their superior mobility and operational flexibility. Maintaining the freshness of information in UAV-assisted IoT systems is crucial for real-time monitoring applications, particularly when dealing with stochastic generation patterns of sensory data. Orchestrating multiple energy-constrained UAVs to ensure the temporal validity of collected information poses significant technical challenges due to dynamic mission constraints and limited onboard power supplies. To address this issue, we investigate the problem of information freshness optimization in a multi-UAV collaborative data collection environment, proposes an attention-based deep deterministic policy gradient (A-DDPG) algorithm, constructs a multi-UAV-assisted IoT data collection system model that considers data freshness, communication quality, and energy efficiency, and models the UAV trajectory planning, hovering point selection, and task assignment problems as a Markov decision process. Due to the limitations of the standard DDPG algorithm in handling high-dimensional state spaces and multiple constraints, we introduce an attention layer in the A-DDPG algorithm to enhance the perception of key state features, designs a prioritized experience replay mechanism to enhance data sampling efficiency in reinforcement learning processes., implements normalization strategies based on the characteristics of different state components, and develops action constraint handling methods to ensure that UAV behaviors meet physical constraints. Through comprehensive simulation tests, the proposed algorithm is compared with existing technologies, demonstrating its high effectiveness in terms of average age of information, energy efficiency, and task completion rate.</div></div>","PeriodicalId":55555,"journal":{"name":"Ad Hoc Networks","volume":"185 ","pages":"Article 104162"},"PeriodicalIF":4.8,"publicationDate":"2026-04-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146175113","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}

引用次数: 0

qIoV: A quantum-driven approach for environmental monitoring and rapid response systems using internet of vehicles qIoV：一种量子驱动的方法，用于环境监测和使用车联网的快速反应系统

IF 4.8 3区计算机科学 Q1 COMPUTER SCIENCE, INFORMATION SYSTEMS

Ad Hoc Networks

Pub Date : 2026-04-15 Epub Date: 2026-02-02 DOI: 10.1016/j.adhoc.2026.104158

Ankur Nahar , Koustav Kumar Mondal , Debasis Das , Rajkumar Buyya

This paper addresses the critical demand for advanced rapid response mechanisms in managing a wide array of environmental hazards, including urban pipeline leaks, industrial gas discharges, methane emissions from landfills, chlorine leaks from water treatment plants, and residential carbon monoxide releases. Conventional sensing and alert systems often struggle with the timely analysis of high-dimensional sensor data and suffer delays as data volume increases. We propose a novel framework, qIoV, which integrates quantum computing with the Internet of Vehicles (IoVs) to leverage the computational efficiency, parallelism, and entanglement properties inherent in quantum mechanics. The qIoV framework utilizes vehicular-mounted environmental sensors for highly accurate air quality assessments, where quantum principles enhance both sensitivity and precision. A core innovation is the Quantum Mesh Network Fabric (QMF), which dynamically adapts the quantum network topology to vehicular movement, maintaining quantum state integrity among environmental and vehicular disruptions, thereby ensuring robust data transmission. Furthermore, we implement a variational quantum classifier (VQC) with advanced entanglement techniques, significantly reducing latency in hazard alerts and facilitating rapid communication with emergency response teams and the public. Our experimental evaluations using the IBM OpenQASM 3 platform with a 127-qubit system achieved over 90% precision, recall, and F1-score in pair plot analysis, alongside an 83% increase in toxic gas detection speed compared to conventional methods. Theoretical analysis further substantiates the efficiency of quantum rotation, teleportation protocols, and the fidelity of quantum entanglement, highlighting the potential of quantum computing in environmental hazard management.

本文解决了在管理各种环境危害方面对先进快速反应机制的关键需求，包括城市管道泄漏、工业气体排放、垃圾填埋场甲烷排放、水处理厂氯泄漏和住宅一氧化碳排放。传统的传感和警报系统往往难以及时分析高维传感器数据，并且随着数据量的增加而遭受延迟。我们提出了一个新的框架，qIoV，它将量子计算与车联网（IoVs）集成在一起，以利用量子力学固有的计算效率、并行性和纠缠特性。qIoV框架利用车载环境传感器进行高精度的空气质量评估，其中量子原理提高了灵敏度和精度。核心创新是量子网状网络结构（QMF），它动态调整量子网络拓扑以适应车辆运动，在环境和车辆中断中保持量子态完整性，从而确保稳健的数据传输。此外，我们采用先进的纠缠技术实现了变分量子分类器（VQC），大大减少了危险警报的延迟，并促进了与应急响应团队和公众的快速沟通。我们使用IBM OpenQASM 3平台和127量子位系统进行实验评估，在对图分析中实现了超过90%的精度，召回率和f1得分，同时与传统方法相比，有毒气体检测速度提高了83%。理论分析进一步证实了量子旋转、隐形传态协议和量子纠缠的保真度的效率，突出了量子计算在环境危害管理中的潜力。

{"title":"qIoV: A quantum-driven approach for environmental monitoring and rapid response systems using internet of vehicles","authors":"Ankur Nahar , Koustav Kumar Mondal , Debasis Das , Rajkumar Buyya","doi":"10.1016/j.adhoc.2026.104158","DOIUrl":"10.1016/j.adhoc.2026.104158","url":null,"abstract":"<div><div>This paper addresses the critical demand for advanced rapid response mechanisms in managing a wide array of environmental hazards, including urban pipeline leaks, industrial gas discharges, methane emissions from landfills, chlorine leaks from water treatment plants, and residential carbon monoxide releases. Conventional sensing and alert systems often struggle with the timely analysis of high-dimensional sensor data and suffer delays as data volume increases. We propose a novel framework, <em>qIoV</em>, which integrates quantum computing with the Internet of Vehicles (IoVs) to leverage the computational efficiency, parallelism, and entanglement properties inherent in quantum mechanics. The qIoV framework utilizes vehicular-mounted environmental sensors for highly accurate air quality assessments, where quantum principles enhance both sensitivity and precision. A core innovation is the Quantum Mesh Network Fabric (QMF), which dynamically adapts the quantum network topology to vehicular movement, maintaining quantum state integrity among environmental and vehicular disruptions, thereby ensuring robust data transmission. Furthermore, we implement a variational quantum classifier (VQC) with advanced entanglement techniques, significantly reducing latency in hazard alerts and facilitating rapid communication with emergency response teams and the public. Our experimental evaluations using the IBM OpenQASM 3 platform with a 127-qubit system achieved over 90% precision, recall, and F1-score in pair plot analysis, alongside an 83% increase in toxic gas detection speed compared to conventional methods. Theoretical analysis further substantiates the efficiency of quantum rotation, teleportation protocols, and the fidelity of quantum entanglement, highlighting the potential of quantum computing in environmental hazard management.</div></div>","PeriodicalId":55555,"journal":{"name":"Ad Hoc Networks","volume":"185 ","pages":"Article 104158"},"PeriodicalIF":4.8,"publicationDate":"2026-04-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146098583","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}

引用次数: 0

DeepSpect: An RF spectrogram-based deep learning approach for near-real-time attack detection in FANETs DeepSpect：一种基于射频频谱图的深度学习方法，用于近实时攻击检测

IF 4.8 3区计算机科学 Q1 COMPUTER SCIENCE, INFORMATION SYSTEMS

Ad Hoc Networks

Pub Date : 2026-04-15 Epub Date: 2026-02-12 DOI: 10.1016/j.adhoc.2026.104178

Cengizhan Yapıcıoğlu , Sedef Demirci , Mehmet Demirci

Flying ad-hoc networks (FANETs) facilitate autonomous communication and collaboration among unmanned aerial vehicles (UAVs) and are increasingly utilized in defense, disaster response, agriculture, and environmental monitoring. However, their limited computational resources and critical operational roles make them susceptible to cyber–physical threats such as jamming, deauthentication, and physical attacks. Existing solutions often target individual attacks and rely on complex, resource-intensive methods that are impractical for lightweight drones. In this study, we propose a novel deep learning-based approach for near-real-time multi-class attack detection in FANETs using RF spectrogram images. RF spectrograms provide a robust, environment-independent representation of drone communications, enabling accurate attack detection without high computational overhead. We introduce DroneAttackRF, the first publicly available real-world dataset of RF spectrograms collected from DJI Ryze Tello and Piranha F-55 drones under various attack scenarios. We develop and evaluate seven deep learning classifiers, including two customized models based on CNN and Autoencoder, as well as five transfer learning models based on VGG-16, ResNet50, InceptionV3, MobileNet, and Xception. The developed models achieved competitive or higher performance compared to prior studies, with the CNN-based model attaining 98.9% accuracy in multi-class detection of different attack types, though dataset and methodology differences limit the feasibility of direct comparison. Additionally, our approach demonstrated fast detection capability, with RF spectrogram acquisition taking only 0.52 s and CNN-based attack classification completing in 0.55 s. The proposed approach demonstrates significant improvements in detection accuracy and efficiency, offering a practical and scalable solution for enhancing UAV network security.

飞行自组织网络（fanet）促进了无人驾驶飞行器（uav）之间的自主通信和协作，并越来越多地用于国防，灾害响应，农业和环境监测。然而，它们有限的计算资源和关键的操作角色使它们容易受到网络物理威胁，如干扰、去认证和物理攻击。现有的解决方案通常针对个人攻击，并且依赖于复杂的、资源密集型的方法，这些方法对于轻型无人机来说是不切实际的。在这项研究中，我们提出了一种新的基于深度学习的方法，用于使用射频频谱图图像进行近实时的多类攻击检测。射频频谱图提供了一个强大的、与环境无关的无人机通信表示，可以在没有高计算开销的情况下实现准确的攻击检测。我们介绍DroneAttackRF，这是第一个公开可用的真实世界射频频谱数据集，收集自大疆Ryze Tello和食人鱼F-55无人机在各种攻击场景下的射频频谱图。我们开发并评估了七个深度学习分类器，包括两个基于CNN和Autoencoder的定制模型，以及五个基于VGG-16、ResNet50、InceptionV3、MobileNet和Xception的迁移学习模型。尽管数据集和方法的差异限制了直接比较的可行性，但与之前的研究相比，所开发的模型取得了相当或更高的性能，基于cnn的模型在不同攻击类型的多类检测中达到了98.9%的准确率。此外，我们的方法证明了快速检测能力，射频频谱图采集仅需0.52秒，基于cnn的攻击分类仅需0.55秒。该方法显著提高了检测精度和效率，为增强无人机网络安全提供了一种实用且可扩展的解决方案。

{"title":"DeepSpect: An RF spectrogram-based deep learning approach for near-real-time attack detection in FANETs","authors":"Cengizhan Yapıcıoğlu , Sedef Demirci , Mehmet Demirci","doi":"10.1016/j.adhoc.2026.104178","DOIUrl":"10.1016/j.adhoc.2026.104178","url":null,"abstract":"<div><div>Flying ad-hoc networks (FANETs) facilitate autonomous communication and collaboration among unmanned aerial vehicles (UAVs) and are increasingly utilized in defense, disaster response, agriculture, and environmental monitoring. However, their limited computational resources and critical operational roles make them susceptible to cyber–physical threats such as jamming, deauthentication, and physical attacks. Existing solutions often target individual attacks and rely on complex, resource-intensive methods that are impractical for lightweight drones. In this study, we propose a novel deep learning-based approach for near-real-time multi-class attack detection in FANETs using RF spectrogram images. RF spectrograms provide a robust, environment-independent representation of drone communications, enabling accurate attack detection without high computational overhead. We introduce DroneAttackRF, the first publicly available real-world dataset of RF spectrograms collected from DJI Ryze Tello and Piranha F-55 drones under various attack scenarios. We develop and evaluate seven deep learning classifiers, including two customized models based on CNN and Autoencoder, as well as five transfer learning models based on VGG-16, ResNet50, InceptionV3, MobileNet, and Xception. The developed models achieved competitive or higher performance compared to prior studies, with the CNN-based model attaining 98.9% accuracy in multi-class detection of different attack types, though dataset and methodology differences limit the feasibility of direct comparison. Additionally, our approach demonstrated fast detection capability, with RF spectrogram acquisition taking only 0.52 s and CNN-based attack classification completing in 0.55 s. The proposed approach demonstrates significant improvements in detection accuracy and efficiency, offering a practical and scalable solution for enhancing UAV network security.</div></div>","PeriodicalId":55555,"journal":{"name":"Ad Hoc Networks","volume":"185 ","pages":"Article 104178"},"PeriodicalIF":4.8,"publicationDate":"2026-04-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146175097","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}

引用次数: 0

A spatio-temporal graph learning framework with attention mechanism for secure RPL in mobile IoT 基于注意机制的移动物联网安全RPL时空图学习框架

IF 4.8 3区计算机科学 Q1 COMPUTER SCIENCE, INFORMATION SYSTEMS

Ad Hoc Networks

Pub Date : 2026-04-15 Epub Date: 2026-01-27 DOI: 10.1016/j.adhoc.2026.104156

Zohre Shoaei, Rasool Esmaeilyfard, Reza Javidan

Mobility‑aware IoT networks operate under rapidly shifting topologies, where even authorized nodes can perform stealthy routing attacks that bypass standard cryptographic defenses. These threats are compounded by dynamic connectivity patterns, fluctuating link qualities, and heterogeneous node behaviors, creating a high‑dimensional, non‑stationary security landscape. We introduce a temporal–spatial trust framework that represents the network as a continuously evolving dynamic graph, embedding per‑node behavioral states together with aggregated neighborhood patterns across structural, mobility, and traffic domains. These high‑context sequences feed into a multi‑layer GRU‑based Sequence‑to‑Sequence architecture equipped with multi‑head attention, enabling concurrent modeling of local temporal fluctuations and long‑range spatial dependencies. A composite trust scoring mechanism integrates model‑inferred anomalies with deterministic protocol checks and peer‑reported reputation, regulated by hyper‑parameter‑optimized fusion weights. Trust scores are embedded into RPL’s rank metric and filtered through a hysteresis‑governed parent selection policy to ensure both rapid threat isolation and topological stability. Extensive simulations in Contiki-NG, leveraging real-world urban mobility traces from the Microsoft GeoLife dataset and the RADAR benchmark, demonstrate robustness against five specific threats (Rank, Blackhole, Sybil, Sinkhole, and Selective Forwarding). Results indicate up to 96 % detection accuracy, a 38 % reduction in detection latency, and 20–40 % lower control overhead, all while maintaining a runtime memory footprint under 10 KB. By combining dynamic graph‑based context encoding, attention‑driven sequence learning, and multi‑source trust fusion, the proposed approach offers a deployable, high‑fidelity, and scalable security enhancement for RPL in next‑generation IoT environments.

移动感知物联网网络在快速变化的拓扑下运行，即使是授权节点也可以执行绕过标准加密防御的隐形路由攻击。这些威胁与动态连接模式、波动的链接质量和异构节点行为相结合，形成了高维、非固定的安全环境。我们引入了一个时空信任框架，该框架将网络表示为一个不断发展的动态图，将每个节点的行为状态与跨结构、移动性和交通领域的聚合邻居模式嵌入在一起。这些高上下文序列输入到一个多层基于GRU的序列对序列架构中，该架构配备了多头关注，从而能够对局部时间波动和远程空间依赖性进行并发建模。复合信任评分机制将模型推断的异常与确定性协议检查和同行报告的声誉集成在一起，由超参数优化的融合权重调节。信任分数被嵌入到RPL的等级度量中，并通过滞后控制的亲本选择策略进行过滤，以确保快速隔离威胁和拓扑稳定性。在Contiki-NG中进行了大量模拟，利用来自微软GeoLife数据集和RADAR基准的真实城市交通轨迹，证明了对五种特定威胁（Rank, Blackhole, Sybil， Sinkhole和选择性转发）的鲁棒性。结果表明，检测准确率高达96%，检测延迟降低38%，控制开销降低20 - 40%，同时运行时内存占用保持在10 KB以下。通过结合基于动态图的上下文编码、注意力驱动的序列学习和多源信任融合，所提出的方法为下一代物联网环境中的RPL提供了可部署、高保真和可扩展的安全性增强。

{"title":"A spatio-temporal graph learning framework with attention mechanism for secure RPL in mobile IoT","authors":"Zohre Shoaei, Rasool Esmaeilyfard, Reza Javidan","doi":"10.1016/j.adhoc.2026.104156","DOIUrl":"10.1016/j.adhoc.2026.104156","url":null,"abstract":"<div><div>Mobility‑aware IoT networks operate under rapidly shifting topologies, where even authorized nodes can perform stealthy routing attacks that bypass standard cryptographic defenses. These threats are compounded by dynamic connectivity patterns, fluctuating link qualities, and heterogeneous node behaviors, creating a high‑dimensional, non‑stationary security landscape. We introduce a temporal–spatial trust framework that represents the network as a continuously evolving dynamic graph, embedding per‑node behavioral states together with aggregated neighborhood patterns across structural, mobility, and traffic domains. These high‑context sequences feed into a multi‑layer GRU‑based Sequence‑to‑Sequence architecture equipped with multi‑head attention, enabling concurrent modeling of local temporal fluctuations and long‑range spatial dependencies. A composite trust scoring mechanism integrates model‑inferred anomalies with deterministic protocol checks and peer‑reported reputation, regulated by hyper‑parameter‑optimized fusion weights. Trust scores are embedded into RPL’s rank metric and filtered through a hysteresis‑governed parent selection policy to ensure both rapid threat isolation and topological stability. Extensive simulations in Contiki-NG, leveraging real-world urban mobility traces from the Microsoft GeoLife dataset and the RADAR benchmark, demonstrate robustness against five specific threats (Rank, Blackhole, Sybil, Sinkhole, and Selective Forwarding). Results indicate up to 96 % detection accuracy, a 38 % reduction in detection latency, and 20–40 % lower control overhead, all while maintaining a runtime memory footprint under 10 KB. By combining dynamic graph‑based context encoding, attention‑driven sequence learning, and multi‑source trust fusion, the proposed approach offers a deployable, high‑fidelity, and scalable security enhancement for RPL in next‑generation IoT environments.</div></div>","PeriodicalId":55555,"journal":{"name":"Ad Hoc Networks","volume":"185 ","pages":"Article 104156"},"PeriodicalIF":4.8,"publicationDate":"2026-04-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146175100","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}

引用次数: 0