Ad Hoc Networks最新文献_第3页

Explainable energy-efficient UAV-assisted cluster-based data collection in WSNs 可解释的节能无人机辅助的基于簇的无线传感器网络数据收集

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

Ad Hoc Networks

Pub Date : 2026-01-06 DOI: 10.1016/j.adhoc.2026.104137

Nadine Abbas

The use of unmanned aerial vehicles (UAVs) is becoming an integral element in modern wireless sensor networks (WSNs), due to their flexibility and cost-effectiveness, especially for data collection in challenging hard-to-reach environments. Cluster-based solutions further enhance data collection efficiency by allowing sensor nodes (SNs) to act as cluster heads (CHs) aggregating and relaying data to UAVs. Traditional approaches often rely on static clustering and lack transparency in decision-making regarding CH selection and UAV deployment. This work proposes an explainable energy-efficient UAV-assisted cluster-based data collection framework that integrates optimal and sub-optimal solutions as well as adopts machine learning-based CH prediction augmented with explainable AI techniques. First, we formulate a joint multi-objective optimization problem to minimize UAV usage, ensure energy-efficient CH selection, and guarantee data collection within deadline constraints. Second, we propose a sequential solving approach and then a scalable iterative cluster-based approach to provide real-time solutions for large-scale networks. Moreover, we develop machine learning (ML) models to predict CH selection using a customized dataset generated from extensive simulations of our proposed approach, capturing features like location, neighborhood density, data size, and deadlines. Furthermore, we use Explainable AI (XAI) techniques, particularly SHAP, to interpret the CH prediction model, providing insights into feature importance and decision rationale. This transparency enables network operators to validate CH assignments and strategically plan UAV deployment. Overall, the proposed framework achieves near-optimal trade-offs between UAV deployment, energy consumption, and execution time, leveraging flexible communication, emphasizing spatial and connectivity features and enhancing model interpretability for real-world applications.

由于其灵活性和成本效益，无人机的使用正成为现代无线传感器网络（wsn）的一个组成部分，特别是在具有挑战性的难以到达的环境中进行数据收集。基于集群的解决方案通过允许传感器节点（SNs）作为集群头（CHs）聚合和中继数据到无人机，进一步提高了数据收集效率。传统的方法通常依赖于静态聚类，在CH选择和无人机部署的决策中缺乏透明度。这项工作提出了一个可解释的节能无人机辅助基于集群的数据收集框架，该框架集成了最优和次最优解决方案，并采用了基于机器学习的CH预测和可解释的人工智能技术。首先，我们制定了一个联合多目标优化问题，以最大限度地减少无人机的使用，确保节能的CH选择，并保证在期限内收集数据。其次，我们提出了一种顺序求解方法，然后是基于可扩展迭代簇的方法，为大规模网络提供实时解决方案。此外，我们开发了机器学习（ML）模型，使用从我们提出的方法的广泛模拟生成的自定义数据集来预测CH选择，捕获位置，邻居密度，数据大小和截止日期等特征。此外，我们使用可解释的人工智能（XAI）技术，特别是SHAP，来解释CH预测模型，提供对特征重要性和决策原理的见解。这种透明度使网络运营商能够验证CH分配并战略性地规划无人机部署。总体而言，所提出的框架在无人机部署、能耗和执行时间之间实现了近乎最佳的权衡，利用了灵活的通信，强调了空间和连通性特征，并增强了模型对现实世界应用的可解释性。

{"title":"Explainable energy-efficient UAV-assisted cluster-based data collection in WSNs","authors":"Nadine Abbas","doi":"10.1016/j.adhoc.2026.104137","DOIUrl":"10.1016/j.adhoc.2026.104137","url":null,"abstract":"<div><div>The use of unmanned aerial vehicles (UAVs) is becoming an integral element in modern wireless sensor networks (WSNs), due to their flexibility and cost-effectiveness, especially for data collection in challenging hard-to-reach environments. Cluster-based solutions further enhance data collection efficiency by allowing sensor nodes (SNs) to act as cluster heads (CHs) aggregating and relaying data to UAVs. Traditional approaches often rely on static clustering and lack transparency in decision-making regarding CH selection and UAV deployment. This work proposes an explainable energy-efficient UAV-assisted cluster-based data collection framework that integrates optimal and sub-optimal solutions as well as adopts machine learning-based CH prediction augmented with explainable AI techniques. First, we formulate a joint multi-objective optimization problem to minimize UAV usage, ensure energy-efficient CH selection, and guarantee data collection within deadline constraints. Second, we propose a sequential solving approach and then a scalable iterative cluster-based approach to provide real-time solutions for large-scale networks. Moreover, we develop machine learning (ML) models to predict CH selection using a customized dataset generated from extensive simulations of our proposed approach, capturing features like location, neighborhood density, data size, and deadlines. Furthermore, we use Explainable AI (XAI) techniques, particularly SHAP, to interpret the CH prediction model, providing insights into feature importance and decision rationale. This transparency enables network operators to validate CH assignments and strategically plan UAV deployment. Overall, the proposed framework achieves near-optimal trade-offs between UAV deployment, energy consumption, and execution time, leveraging flexible communication, emphasizing spatial and connectivity features and enhancing model interpretability for real-world applications.</div></div>","PeriodicalId":55555,"journal":{"name":"Ad Hoc Networks","volume":"184 ","pages":"Article 104137"},"PeriodicalIF":4.8,"publicationDate":"2026-01-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145908941","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

DRUID: Coordinating drone movements for compromised node identification 德鲁伊：协调无人机移动以识别受损节点

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

Ad Hoc Networks

Pub Date : 2026-01-05 DOI: 10.1016/j.adhoc.2026.104135

Mauro Farina, Erica Salvato, Martino Trevisan, Alberto Bartoli

In recent years, Unmanned Aerial Vehicles (UAVs) (also called drones) networks have become increasingly popular in scenarios where rapid deployment, flexible mobility, and real-time data acquisition are crucial, such as disaster relief, environmental monitoring, military operations, and smart city infrastructure. However, due to their dynamic nature and dependence on wireless communication, they are intrinsically vulnerable to a variety of cyberattacks. In this work, we present DRUID, a decentralized scheme for silently identifying a compromised drone that selectively alters the messages it forwards. The scheme uses a combination of secret sharing and multipath routing to allow a pair of communicating drones, namely

A

and

B

, to detect the presence of a compromised drone along any route between them, thereby categorizing each route as either safe or compromised. The scheme operates iteratively and consists of three key modules: (i) an Information Retrieval Procedure that allows

A

to learn more about the topology, (ii) a binary search-like Identification Procedure, and (iii) if the previous module fails to identify the compromised drone, a Node Repositioning Procedure that relocates nodes closer to the compromised path. We validate DRUID on a large and diverse set of 178 731 graphs representing realistic UAV networks with different communication ranges. Comparing our scheme to previous work, experiments show that DRUID achieves a 97 % identification rate—up from the 54 % of the most recent alternative approach. We analyze the cost associated with the node repositioning procedure in terms of computation time and drone movement, and show that it generally takes a few seconds.

近年来，在救灾、环境监测、军事行动和智慧城市基础设施等快速部署、灵活机动和实时数据采集至关重要的场景中，无人驾驶飞行器（uav）（也称为无人机）网络越来越受欢迎。然而，由于它们的动态性和对无线通信的依赖性，它们本质上容易受到各种网络攻击。在这项工作中，我们提出了DRUID，这是一种分散的方案，用于无声地识别受损的无人机，并选择性地更改其转发的消息。该方案使用秘密共享和多路径路由的组合，允许一对通信无人机，即a和B，在它们之间的任何路线上检测到受损无人机的存在，从而将每条路线分类为安全或受损。该方案迭代运行，由三个关键模块组成：(i)允许A了解更多拓扑信息的信息检索过程，（ii）类似二进制搜索的识别过程，以及（iii）如果前一个模块无法识别受损无人机，则节点重新定位过程将节点重新定位到更靠近受损路径的地方。我们在具有不同通信范围的实际无人机网络的178 731张图上验证了DRUID。将我们的方案与之前的工作进行比较，实验表明DRUID的识别率达到97%，而最近的替代方法的识别率为54%。我们从计算时间和无人机移动角度分析了节点重新定位过程的相关成本，并表明它通常需要几秒钟。

{"title":"DRUID: Coordinating drone movements for compromised node identification","authors":"Mauro Farina, Erica Salvato, Martino Trevisan, Alberto Bartoli","doi":"10.1016/j.adhoc.2026.104135","DOIUrl":"10.1016/j.adhoc.2026.104135","url":null,"abstract":"<div><div>In recent years, Unmanned Aerial Vehicles (UAVs) (also called drones) networks have become increasingly popular in scenarios where rapid deployment, flexible mobility, and real-time data acquisition are crucial, such as disaster relief, environmental monitoring, military operations, and smart city infrastructure. However, due to their dynamic nature and dependence on wireless communication, they are intrinsically vulnerable to a variety of cyberattacks. In this work, we present <span>DRUID</span>, a decentralized scheme for silently identifying a compromised drone that selectively alters the messages it forwards. The scheme uses a combination of secret sharing and multipath routing to allow a pair of communicating drones, namely <span><math><mi>A</mi></math></span> and <span><math><mi>B</mi></math></span>, to detect the presence of a compromised drone along any route between them, thereby categorizing each route as either safe or compromised. The scheme operates iteratively and consists of three key modules: (i) an Information Retrieval Procedure that allows <span><math><mi>A</mi></math></span> to learn more about the topology, (ii) a binary search-like Identification Procedure, and (iii) if the previous module fails to identify the compromised drone, a Node Repositioning Procedure that relocates nodes closer to the compromised path. We validate <span>DRUID</span> on a large and diverse set of 178731 graphs representing realistic UAV networks with different communication ranges. Comparing our scheme to previous work, experiments show that <span>DRUID</span> achieves a 97% identification rate—up from the 54% of the most recent alternative approach. We analyze the cost associated with the node repositioning procedure in terms of computation time and drone movement, and show that it generally takes a few seconds.</div></div>","PeriodicalId":55555,"journal":{"name":"Ad Hoc Networks","volume":"184 ","pages":"Article 104135"},"PeriodicalIF":4.8,"publicationDate":"2026-01-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145908940","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

Kalman filter scheduling for 6TiSCH network with traffic adaptation optimized for bursty traffic 基于突发通信量的6TiSCH网络自适应卡尔曼滤波调度

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

Ad Hoc Networks

Pub Date : 2026-01-05 DOI: 10.1016/j.adhoc.2025.104130

Yan Zhang, Shijie Xu, Qingqing Huang, Yan Han

The sensor nodes equipped with IEEE 802.15.4e (6TiSCH) wireless protocol stack and IPv6 time slot channel hopping mode have deterministic network characteristics after networking, providing low-latency and highly reliable communication for industrial scenarios with growing demand for low-power sensor networks. However, existing scheduling algorithms perform poorly under the bursty traffic commonly found in industrial environments. Due to the limitations of their design principles, they are unable to respond quickly to changes in traffic or differentiate between bursty traffic patterns to accurately sense traffic conditions, resulting in high latency, low reliability and additional power consumption. Therefore, we propose a scheduling method called the Kalman Filter Traffic Sensing Prediction Scheduling Function (KSF). KSF utilizes the filtered processing of node Cell usage and per-slot frame queue increment as the primary basis for scheduling decisions, coupled with adaptive filtering parameters, to achieve the ability to ignore transient fluctuation noise and respond quickly after the occurrence of bursts. In addition, we utilize filtering to predict the ratio of the number of received data packets to the number of sent data packets in the next slot frame to distinguish burst patterns and dynamically change KSF’s scheduling strategy. Experiments demonstrate that KSF exhibits more optimal scheduling performance under bursty traffic conditions, reducing latency by 14.82% compared to the well-known OTF while maintaining the lowest power consumption across all traffic rates.

采用IEEE 802.15.4e （6TiSCH）无线协议栈和IPv6时隙信道跳变模式的传感器节点组网后具有确定性的网络特性，为低功耗传感器网络需求日益增长的工业场景提供低时延、高可靠的通信。然而，现有的调度算法在工业环境中常见的突发流量下表现不佳。由于其设计原则的限制，它们无法快速响应流量变化或区分突发流量模式以准确感知交通状况，从而导致高延迟、低可靠性和额外的功耗。因此，我们提出一种调度方法，称为卡尔曼滤波交通感知预测调度函数（KSF）。KSF利用节点Cell使用率和每插槽帧队列增量的滤波处理作为调度决策的主要依据，再加上自适应滤波参数，实现了忽略瞬态波动噪声和在突发发生后快速响应的能力。此外，我们利用过滤来预测下一个槽帧中接收数据包数量与发送数据包数量的比例，以区分突发模式并动态改变KSF的调度策略。实验表明，KSF在突发流量条件下表现出更优的调度性能，与众所周知的OTF相比，延迟降低了14.82%，同时在所有流量速率下保持最低的功耗。

{"title":"Kalman filter scheduling for 6TiSCH network with traffic adaptation optimized for bursty traffic","authors":"Yan Zhang, Shijie Xu, Qingqing Huang, Yan Han","doi":"10.1016/j.adhoc.2025.104130","DOIUrl":"10.1016/j.adhoc.2025.104130","url":null,"abstract":"<div><div>The sensor nodes equipped with IEEE 802.15.4e (6TiSCH) wireless protocol stack and IPv6 time slot channel hopping mode have deterministic network characteristics after networking, providing low-latency and highly reliable communication for industrial scenarios with growing demand for low-power sensor networks. However, existing scheduling algorithms perform poorly under the bursty traffic commonly found in industrial environments. Due to the limitations of their design principles, they are unable to respond quickly to changes in traffic or differentiate between bursty traffic patterns to accurately sense traffic conditions, resulting in high latency, low reliability and additional power consumption. Therefore, we propose a scheduling method called the Kalman Filter Traffic Sensing Prediction Scheduling Function (KSF). KSF utilizes the filtered processing of node Cell usage and per-slot frame queue increment as the primary basis for scheduling decisions, coupled with adaptive filtering parameters, to achieve the ability to ignore transient fluctuation noise and respond quickly after the occurrence of bursts. In addition, we utilize filtering to predict the ratio of the number of received data packets to the number of sent data packets in the next slot frame to distinguish burst patterns and dynamically change KSF’s scheduling strategy. Experiments demonstrate that KSF exhibits more optimal scheduling performance under bursty traffic conditions, reducing latency by 14.82% compared to the well-known OTF while maintaining the lowest power consumption across all traffic rates.</div></div>","PeriodicalId":55555,"journal":{"name":"Ad Hoc Networks","volume":"184 ","pages":"Article 104130"},"PeriodicalIF":4.8,"publicationDate":"2026-01-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145928709","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

UpsFed-IDS: U-shaped split federated intrusion detection system for securing UAV communication in dynamic networks UpsFed-IDS：用于动态网络中无人机通信安全的u形分离联邦入侵检测系统

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

Ad Hoc Networks

Pub Date : 2025-12-31 DOI: 10.1016/j.adhoc.2025.104133

Zongpu Wei, Jinsong Wang, Zening Zhao, Zhao Zhao, Kai Shi

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.

在无人机中集成入侵检测系统（IDS）对于保障无人机的运行可靠性和整体安全性至关重要。部署在数据中心的集中式IDS已经变得不切实际，这主要是由于对数据隐私和计算限制的担忧。基于联邦学习（FL）的入侵检测缓解了传统入侵检测固有的数据泄漏问题。然而，它与无人机系统的融合仍然面临着不可避免的挑战。首先，对无人机进行局部模型训练的要求带来了大量的计算开销。其次，无人机数据的非独立和同分布（non-IID）特性直接影响了IDS模型的性能。第三，无人机网络连通性的不断动态变化削弱了联邦入侵检测系统的鲁棒性。为了解决这些挑战，本文提出了一种u形分裂联邦入侵检测系统（UpsFed-IDS），用于保护无人机通信。受FL和分裂学习（SL）的启发，我们卸载了一部分IDS模型训练到地面控制站（GCS）。这种方法确保原始数据和标签保留在无人机上，从而增强了数据隐私保护并减少了无人机方面的计算开销。在该系统中，我们提出了一种针对分裂的头部个性化方法，将全局特征学习与局部模型个性化解耦，从而增强了IDS模型在异构数据场景下的性能。此外，设计了一种客户端故障转移机制来解决动态无人机网络在训练过程中出现的断开问题，有效地提高了系统的整体鲁棒性。对UAVCAN攻击和WSN-DS数据集进行了广泛的实验评估。结果表明，UpsFed-IDS在攻击识别性能和局部计算开销方面都优于现有的FL框架。

{"title":"UpsFed-IDS: U-shaped split federated intrusion detection system for securing UAV communication in dynamic networks","authors":"Zongpu Wei, Jinsong Wang, Zening Zhao, Zhao Zhao, 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":"2025-12-31","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}

引用次数: 0

A deep learning-based approach for heterogeneous hotspot-coverage in UAV deployment 基于深度学习的无人机部署异构热点覆盖方法

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

Ad Hoc Networks

Pub Date : 2025-12-30 DOI: 10.1016/j.adhoc.2025.104127

Kolichala Rajashekar , Vamsi Krishna Sunkara , Subhajit Sidhanta

The deployment of unmanned aerial vehicles (UAVs) for wireless coverage in dynamic environments, such as public gatherings, road junctions, and urban intersections, presents numerous challenges owing to variations in the size of the hotspots, the mobility patterns of the users, and quality of service (QoS) requirements. Although the iterative or heuristic algorithms used in previous papers can potentially adapt to these changes, they would either incur significant runtime overhead on computationally constrained UAV hardware or require uninterrupted backhaul communication. In this paper, we formalize the above Dynamic UAV Deployment (DUDE) problem, show that it is NP-hard, and propose a hybrid Convolutional Neural Network-based (CNN-based) approach to predict the optimal 3D placement of a single UAV. Our CNN-based model is trained on a custom synthetic dataset that encompasses diverse user distributions and hotspot sizes, allowing it to perform extensive offline training and then infer UAV positions online in real-time, thereby eliminating the need for repeated online iterations. Experimental results demonstrate that our model achieves a mean absolute error of 3.5 and an average

R^{2}

score exceeding 96% in predicting the UAV’s 3D position across heterogeneous hotspot areas and different statistical distributions of position of users. We also provide extensive comparisons with greedy user-assignment schemes and demonstrate improved connectivity under QoS constraints.

由于热点的大小、用户的移动性模式和服务质量（QoS）要求的变化，在公共集会、道路路口和城市十字路口等动态环境中部署无人驾驶飞行器（uav）进行无线覆盖带来了许多挑战。虽然以前论文中使用的迭代或启发式算法可以潜在地适应这些变化，但它们要么会在计算受限的无人机硬件上产生显著的运行时开销，要么需要不间断的回程通信。在本文中，我们形式化了上述动态无人机部署（DUDE）问题，证明了它是np困难的，并提出了一种基于混合卷积神经网络（cnn）的方法来预测单个无人机的最佳3D布局。我们基于cnn的模型是在包含不同用户分布和热点大小的自定义合成数据集上训练的，允许它进行广泛的离线训练，然后实时在线推断无人机位置，从而消除了重复在线迭代的需要。实验结果表明，该模型在异质热点区域和不同用户位置统计分布下预测无人机三维位置的平均绝对误差为3.5，平均R2评分超过96%。我们还提供了与贪婪用户分配方案的广泛比较，并展示了在QoS约束下改进的连通性。

{"title":"A deep learning-based approach for heterogeneous hotspot-coverage in UAV deployment","authors":"Kolichala Rajashekar , Vamsi Krishna Sunkara , Subhajit Sidhanta","doi":"10.1016/j.adhoc.2025.104127","DOIUrl":"10.1016/j.adhoc.2025.104127","url":null,"abstract":"<div><div>The deployment of unmanned aerial vehicles (UAVs) for wireless coverage in dynamic environments, such as public gatherings, road junctions, and urban intersections, presents numerous challenges owing to variations in the size of the hotspots, the mobility patterns of the users, and quality of service (QoS) requirements. Although the iterative or heuristic algorithms used in previous papers can potentially adapt to these changes, they would either incur significant runtime overhead on computationally constrained UAV hardware or require uninterrupted backhaul communication. In this paper, we formalize the above Dynamic UAV Deployment (DUDE) problem, show that it is NP-hard, and propose a hybrid Convolutional Neural Network-based (CNN-based) approach to predict the optimal 3D placement of a single UAV. Our CNN-based model is trained on a custom synthetic dataset that encompasses diverse user distributions and hotspot sizes, allowing it to perform extensive offline training and then infer UAV positions online in real-time, thereby eliminating the need for repeated online iterations. Experimental results demonstrate that our model achieves a mean absolute error of 3.5 and an average <span><math><msup><mrow><mi>R</mi></mrow><mrow><mn>2</mn></mrow></msup></math></span> score exceeding 96% in predicting the UAV’s 3D position across heterogeneous hotspot areas and different statistical distributions of position of users. We also provide extensive comparisons with greedy user-assignment schemes and demonstrate improved connectivity under QoS constraints.</div></div>","PeriodicalId":55555,"journal":{"name":"Ad Hoc Networks","volume":"183 ","pages":"Article 104127"},"PeriodicalIF":4.8,"publicationDate":"2025-12-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145884185","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

Reliability analysis of multi-state wireless sensor networks with functional dependency based on dynamic Bayesian networks 基于动态贝叶斯网络的功能依赖多状态无线传感器网络可靠性分析

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

Ad Hoc Networks

Pub Date : 2025-12-30 DOI: 10.1016/j.adhoc.2025.104125

Haozhe Liu , Jinfang Zhao , Qun Zhao , Hongliang Sun

Wireless sensor networks (WSNs) are extensively employed in contemporary practical applications. Consequently, analyzing the reliability of WSNs is a significant research area. Recent research has focused on the impact of multiple operational states and functional dependencies on system reliability. However, current reliability modeling approaches rarely address both the effects of data transmission blocking and component dependency failures. Furthermore, studies on system states frequently neglect the diverse operational modes of WSNs, potentially leading to an inaccurate characterization of system behavior over time. To address these shortcomings, this study conceptualizes multi-state WSNs as modular k-out-of-n systems with FDEP, where each module comprising a cluster head (CH) node and its corresponding sensor nodes. Dynamic Bayesian network (DBN) models are employed to construct the structure function of the multi-state WSN. The parameters encoded in the DBN graphical structure of the multi-state WSN are generated automatically by a customized algorithm. Furthermore, an inferencing α-factor method is introduced in DBN model to integrate prior knowledge with observations for updating system reliability while accounting for common cause failures (CCFs). Finally, taking a multi-state meteorological surveillance system as an example, its traffic model is multi-hop transmission, consisting of 8 modules and 46 sensor nodes. The dynamic reliability was evaluated comprehensively when considering FDEP and CCF to illustrate applicability of the proposed framework.

无线传感器网络在当今的实际应用中得到了广泛的应用。因此，分析无线传感器网络的可靠性是一个重要的研究领域。最近的研究主要集中在多种运行状态和功能依赖对系统可靠性的影响。然而，目前的可靠性建模方法很少同时考虑数据传输阻塞和组件依赖故障的影响。此外，对系统状态的研究往往忽略了WSNs的各种工作模式，这可能导致对系统行为随时间变化的不准确描述。为了解决这些缺点，本研究将多状态wsn概念化为具有FDEP的模块化k-out- n系统，其中每个模块包括一个簇头（CH）节点及其相应的传感器节点。采用动态贝叶斯网络（DBN）模型构建多状态无线传感器网络的结构函数。多状态WSN的DBN图形结构中编码的参数通过自定义算法自动生成。此外，在DBN模型中引入推理α-因子方法，将先验知识与观测结果相结合，在考虑共因故障（CCFs）的同时更新系统可靠性。最后，以某多状态气象监测系统为例，其业务模型为多跳传输，由8个模块和46个传感器节点组成。在考虑FDEP和CCF的情况下，对框架的动态可靠性进行了综合评估，以说明该框架的适用性。

{"title":"Reliability analysis of multi-state wireless sensor networks with functional dependency based on dynamic Bayesian networks","authors":"Haozhe Liu , Jinfang Zhao , Qun Zhao , Hongliang Sun","doi":"10.1016/j.adhoc.2025.104125","DOIUrl":"10.1016/j.adhoc.2025.104125","url":null,"abstract":"<div><div>Wireless sensor networks (WSNs) are extensively employed in contemporary practical applications. Consequently, analyzing the reliability of WSNs is a significant research area. Recent research has focused on the impact of multiple operational states and functional dependencies on system reliability. However, current reliability modeling approaches rarely address both the effects of data transmission blocking and component dependency failures. Furthermore, studies on system states frequently neglect the diverse operational modes of WSNs, potentially leading to an inaccurate characterization of system behavior over time. To address these shortcomings, this study conceptualizes multi-state WSNs as modular <em>k</em>-out-of-<em>n</em> systems with FDEP, where each module comprising a cluster head (CH) node and its corresponding sensor nodes. Dynamic Bayesian network (DBN) models are employed to construct the structure function of the multi-state WSN. The parameters encoded in the DBN graphical structure of the multi-state WSN are generated automatically by a customized algorithm. Furthermore, an inferencing <em>α</em>-factor method is introduced in DBN model to integrate prior knowledge with observations for updating system reliability while accounting for common cause failures (CCFs). Finally, taking a multi-state meteorological surveillance system as an example, its traffic model is multi-hop transmission, consisting of 8 modules and 46 sensor nodes. The dynamic reliability was evaluated comprehensively when considering FDEP and CCF to illustrate applicability of the proposed framework.</div></div>","PeriodicalId":55555,"journal":{"name":"Ad Hoc Networks","volume":"183 ","pages":"Article 104125"},"PeriodicalIF":4.8,"publicationDate":"2025-12-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145884344","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 reinforcement learning–based active interception algorithm for wireless networks topology identification 一种基于强化学习的无线网络拓扑识别主动拦截算法

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

Ad Hoc Networks

Pub Date : 2025-12-30 DOI: 10.1016/j.adhoc.2025.104134

Liping Luo , Zhou Peng , Han Xu , Renhai Feng

Accurate Topology Identification (TI) in non-cooperative networks is critical, particularly during various communication engagements that demand low computational overhead. Active interception has proven effective in such scenarios. Specifically, active interception is performed on full-duplex eavesdroppers which cause frequency hopping, thereby obtaining corresponding received signal strength as indicator. However, its interference power adjustment requires numerous iterations and causes overwhelming frequency hopping. This paper proposes a novel Reinforcement Learning-based Active Interception and Node Localization (RLAI-NL) method. RLAI-NL aims to accurately identify network topology. Four different frequency hopping patterns are designed to evaluate the performance of RLAI-NL. Using Reinforcement Learning (RL), an intelligent agent is trained to dynamically adjust its interference power. Through dynamic learning and policy optimization, the agent avoids unnecessary power consumption associated with specially designed search strategies, while adapting effectively to both small- and large-scale networks as well as various communication modes. Simulation results demonstrate that RLAI significantly outperforms traditional active interception methods, achieving 99% accuracy with fewer frequency hops and iterations, thereby reducing computational complexity and power consumption.

在非合作网络中，精确的拓扑识别（TI）是至关重要的，特别是在需要低计算开销的各种通信约定中。在这种情况下，主动拦截已被证明是有效的。具体来说，对引起跳频的全双工窃听器进行主动拦截，从而获得相应的接收信号强度作为指标。但其干扰功率调整需要多次迭代，且会造成压倒性的跳频。提出了一种基于强化学习的主动拦截与节点定位（RLAI-NL）方法。RLAI-NL旨在准确识别网络拓扑。设计了四种不同的跳频模式来评估RLAI-NL的性能。利用强化学习（RL），训练智能体动态调整其干扰功率。通过动态学习和策略优化，智能体避免了特殊设计的搜索策略所带来的不必要的功耗，同时有效地适应小型和大型网络以及各种通信模式。仿真结果表明，RLAI显著优于传统的主动拦截方法，以更少的跳频和迭代次数达到99%的准确率，从而降低了计算复杂度和功耗。

{"title":"A reinforcement learning–based active interception algorithm for wireless networks topology identification","authors":"Liping Luo , Zhou Peng , Han Xu , Renhai Feng","doi":"10.1016/j.adhoc.2025.104134","DOIUrl":"10.1016/j.adhoc.2025.104134","url":null,"abstract":"<div><div>Accurate Topology Identification (TI) in non-cooperative networks is critical, particularly during various communication engagements that demand low computational overhead. Active interception has proven effective in such scenarios. Specifically, active interception is performed on full-duplex eavesdroppers which cause frequency hopping, thereby obtaining corresponding received signal strength as indicator. However, its interference power adjustment requires numerous iterations and causes overwhelming frequency hopping. This paper proposes a novel Reinforcement Learning-based Active Interception and Node Localization (RLAI-NL) method. RLAI-NL aims to accurately identify network topology. Four different frequency hopping patterns are designed to evaluate the performance of RLAI-NL. Using Reinforcement Learning (RL), an intelligent agent is trained to dynamically adjust its interference power. Through dynamic learning and policy optimization, the agent avoids unnecessary power consumption associated with specially designed search strategies, while adapting effectively to both small- and large-scale networks as well as various communication modes. Simulation results demonstrate that RLAI significantly outperforms traditional active interception methods, achieving 99% accuracy with fewer frequency hops and iterations, thereby reducing computational complexity and power consumption.</div></div>","PeriodicalId":55555,"journal":{"name":"Ad Hoc Networks","volume":"184 ","pages":"Article 104134"},"PeriodicalIF":4.8,"publicationDate":"2025-12-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146023874","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

MUFO: Multi-UAV flight optimization for enhancing connectivity in remote driving services MUFO：多无人机飞行优化，增强远程驾驶服务的连通性

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

Ad Hoc Networks

Pub Date : 2025-12-30 DOI: 10.1016/j.adhoc.2025.104129

Van-Linh Nguyen , Lan-Huong Nguyen , Ren-Hung Hwang

Besides autonomous driving, remote driving is a typical example of leveraging communications to eliminate high-risk driving situations in which drivers are fatigued. However, remote driving remains challenging, particularly in urban areas where buildings’ absorption and reflection may significantly hamper control signals. This paper proposes MUFO, a deep-reinforcement-learning-based multi-UAV flight-optimization framework whose objectives are twofold: (i) path planning to determine optimal UAV trajectories that sustain stable links for remote-driving vehicles, and (ii) efficient deployment to minimize the number of UAVs and their energy consumption while guaranteeing service continuity with minimum data rate. First, the coverage and flight cost issues are defined in a multi-objective optimization problem with constraints on UAV energy and collision avoidance. Based on a built-in map of weak signal areas, a novel technique is proposed: a multi-agent deep deterministic policy gradient (MADDPG) scheme. The goal is to determine the best flying strategy for the UAVs to fly over weak signal areas, enhance signal strengths, and relay connectivity when the remote vehicles arrive there. The simulation results show that MADDPG in MUFO outperforms state-of-the-art deep learning methods and searches by up to 8% of deployment efficiency (energy savings, number of deployed UAVs), particularly when there is a high density of ground traffic jam areas and UAVs are required to hover at those areas for an unexpected additional time. MUFO’s strength is that it considerably improves the deployment efficiency of UAVs via cumulative learning from many trials or completed missions.

除了自动驾驶之外，远程驾驶是利用通信来消除驾驶员疲劳的高风险驾驶情况的典型例子。然而，远程驾驶仍然具有挑战性，特别是在城市地区，建筑物的吸收和反射可能会严重阻碍控制信号。本文提出了一种基于深度强化学习的多无人机飞行优化框架MUFO，其目标有两个：(i)路径规划，以确定最优无人机轨迹，为远程驾驶车辆维持稳定的链路；（ii）高效部署，以最小化无人机数量及其能耗，同时保证以最小数据速率服务连续性。首先，将覆盖和飞行成本问题定义为无人机能量和避碰约束的多目标优化问题；基于内置的弱信号区域映射，提出了一种新技术：多智能体深度确定性策略梯度（madpg）方案。目标是确定无人机在弱信号区域飞行的最佳飞行策略，增强信号强度，并在远程车辆到达时中继连接。仿真结果表明，在MUFO中，madpg比最先进的深度学习方法和搜索效率（节能，部署的无人机数量）高出8%，特别是当存在高密度的地面交通堵塞区域并且需要无人机在这些区域悬停意想不到的额外时间时。MUFO的优势在于，它通过从许多试验或完成的任务中累积学习，大大提高了无人机的部署效率。

{"title":"MUFO: Multi-UAV flight optimization for enhancing connectivity in remote driving services","authors":"Van-Linh Nguyen , Lan-Huong Nguyen , Ren-Hung Hwang","doi":"10.1016/j.adhoc.2025.104129","DOIUrl":"10.1016/j.adhoc.2025.104129","url":null,"abstract":"<div><div>Besides autonomous driving, remote driving is a typical example of leveraging communications to eliminate high-risk driving situations in which drivers are fatigued. However, remote driving remains challenging, particularly in urban areas where buildings’ absorption and reflection may significantly hamper control signals. This paper proposes MUFO, a deep-reinforcement-learning-based multi-UAV flight-optimization framework whose objectives are twofold: (i) path planning to determine optimal UAV trajectories that sustain stable links for remote-driving vehicles, and (ii) efficient deployment to minimize the number of UAVs and their energy consumption while guaranteeing service continuity with minimum data rate. First, the coverage and flight cost issues are defined in a multi-objective optimization problem with constraints on UAV energy and collision avoidance. Based on a built-in map of weak signal areas, a novel technique is proposed: a multi-agent deep deterministic policy gradient (MADDPG) scheme. The goal is to determine the best flying strategy for the UAVs to fly over weak signal areas, enhance signal strengths, and relay connectivity when the remote vehicles arrive there. The simulation results show that MADDPG in MUFO outperforms state-of-the-art deep learning methods and searches by up to 8% of deployment efficiency (energy savings, number of deployed UAVs), particularly when there is a high density of ground traffic jam areas and UAVs are required to hover at those areas for an unexpected additional time. MUFO’s strength is that it considerably improves the deployment efficiency of UAVs via cumulative learning from many trials or completed missions.</div></div>","PeriodicalId":55555,"journal":{"name":"Ad Hoc Networks","volume":"183 ","pages":"Article 104129"},"PeriodicalIF":4.8,"publicationDate":"2025-12-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145884182","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

On minimizing the energy consumption in NB-Fi networks with restrictions on packet loss rate and duty cycle 基于丢包率和占空比限制的NB-Fi网络能耗最小化研究

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

Ad Hoc Networks

Pub Date : 2025-12-27 DOI: 10.1016/j.adhoc.2025.104132

Dmitry Bankov, Anastasiia Fedorishcheva, Polina Levchenko, Andrey Lyakhov, Evgeny Khorov

Low-power wide-area networks (LPWANs) are a widely adopted solution for collecting data from remote sensors. One of the possible LPWAN solutions is NB-Fi (Narrow Band Fidelity), which has two options to ensure reliable data delivery. The first one uses acknowledgments. However, the regulation rules of the ISM (Industrial, Scientific, and Medical) bands used by NB-Fi define the minimal duty cycle, which limits the total intensity of packet transmissions by base stations and sensors. As the number of sensors in an NB-Fi network is typically high, the duty cycle restriction may prevent the base station from sending some acknowledgments. Another option to ensure the delivery of sensor data is to use unsolicited retries, which increases not only data reliability but also both the network load and the energy consumption of the sensors. This paper sheds light on how to combine and configure these two options in order to provide the required transmission reliability with minimal sensors’ energy consumption and comply with the duty cycle restrictions. For that, we develop a mathematical model of an NB-Fi network and propose an algorithm based on this model for choosing the ratio of sensors that use acknowledgments and the number of transmission attempts for sensors that use unsolicited retries. Numerical results confirm that the algorithm minimizes the sensors’ energy consumption while satisfying the restrictions on the duty cycle and the packet loss rate.

低功耗广域网（lpwan）是一种广泛采用的远程传感器数据采集解决方案。一种可能的LPWAN解决方案是NB-Fi（窄带保真度），它有两种选择来确保可靠的数据传输。第一个使用确认。然而，NB-Fi使用的ISM（工业、科学和医疗）频段的监管规则定义了最小占空比，这限制了基站和传感器传输数据包的总强度。由于NB-Fi网络中的传感器数量通常很高，占空比限制可能会阻止基站发送一些确认。确保传感器数据传输的另一个选择是使用未请求的重试，这不仅增加了数据可靠性，还增加了网络负载和传感器的能耗。本文阐述了如何结合和配置这两种选择，以提供所需的传输可靠性与最小的传感器的能量消耗，并符合占空比的限制。为此，我们开发了NB-Fi网络的数学模型，并提出了一种基于该模型的算法，用于选择使用确认的传感器比例和使用非请求重试的传感器的传输尝试次数。数值结果表明，该算法在满足占空比和丢包率限制的情况下，最大限度地降低了传感器的能耗。

{"title":"On minimizing the energy consumption in NB-Fi networks with restrictions on packet loss rate and duty cycle","authors":"Dmitry Bankov, Anastasiia Fedorishcheva, Polina Levchenko, Andrey Lyakhov, Evgeny Khorov","doi":"10.1016/j.adhoc.2025.104132","DOIUrl":"10.1016/j.adhoc.2025.104132","url":null,"abstract":"<div><div>Low-power wide-area networks (LPWANs) are a widely adopted solution for collecting data from remote sensors. One of the possible LPWAN solutions is NB-Fi (Narrow Band Fidelity), which has two options to ensure reliable data delivery. The first one uses acknowledgments. However, the regulation rules of the ISM (Industrial, Scientific, and Medical) bands used by NB-Fi define the minimal duty cycle, which limits the total intensity of packet transmissions by base stations and sensors. As the number of sensors in an NB-Fi network is typically high, the duty cycle restriction may prevent the base station from sending some acknowledgments. Another option to ensure the delivery of sensor data is to use unsolicited retries, which increases not only data reliability but also both the network load and the energy consumption of the sensors. This paper sheds light on how to combine and configure these two options in order to provide the required transmission reliability with minimal sensors’ energy consumption and comply with the duty cycle restrictions. For that, we develop a mathematical model of an NB-Fi network and propose an algorithm based on this model for choosing the ratio of sensors that use acknowledgments and the number of transmission attempts for sensors that use unsolicited retries. Numerical results confirm that the algorithm minimizes the sensors’ energy consumption while satisfying the restrictions on the duty cycle and the packet loss rate.</div></div>","PeriodicalId":55555,"journal":{"name":"Ad Hoc Networks","volume":"183 ","pages":"Article 104132"},"PeriodicalIF":4.8,"publicationDate":"2025-12-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145884184","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

Optimizing task allocation in Mobile Crowdsensing with multiple opportunistic users and participatory UAVs 基于多机会用户和参与式无人机的移动众测任务分配优化

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

Ad Hoc Networks

Pub Date : 2025-12-26 DOI: 10.1016/j.adhoc.2025.104128

Shi Yu, Bing Shi, Xiao Su, Saisai Li, Shuai Li, Xing Tang

Mobile Crowdsensing (MCS) refers to the use of human users or unmanned aerial vehicles (UAVs) equipped with mobile devices to collect sensing data. However, existing MCS methods suffer from insufficient coverage due to sparse population distribution in some sensing areas and restrictions on UAV flights in densely populated urban regions. To address this issue, we propose a collaborative approach that combines opportunistic users, who are passively engaged and widely distributed across cities, with participatory UAVs, which are capable of covering sparsely populated regions. Specifically, we introduce a hybrid task allocation strategy called MUMUHTA (Multi-User Multi-UAV Hybrid Task Allocation) to optimize sensing coverage. The strategy assumes that grid heat is derived from historical user trajectory data, and UAVs are assigned only tasks without autonomous control. MUMUHTA uses user trajectory prediction and greedy recruitment for opportunistic users, along with per-slot UAV matching based on the Kuhn–Munkres algorithm, while a dynamic switching rule determines when UAVs take over tasks from users. Simulation experiments using the Rome user trajectory dataset and Shanghai Telecom task dataset show that MUMUHTA improves the task completion rate by an average of 29.70%, 8.11%, 9.69%, 5.05%, and 15.38% compared to benchmark strategies: MPU, MOU-Random, DLMV-MPU, DLMV(T)-MPU, and HR-DLVCS.

移动众测（Mobile Crowdsensing， MCS）是指利用人类用户或配备移动设备的无人机（uav）采集传感数据。然而，现有的MCS方法由于在一些传感区域人口分布稀疏，以及在人口密集的城市地区无人机飞行受到限制，存在覆盖不足的问题。为了解决这个问题，我们提出了一种协作方法，将被动参与并广泛分布在城市中的机会主义用户与能够覆盖人口稀少地区的参与式无人机相结合。具体来说，我们引入了一种称为MUMUHTA（多用户多无人机混合任务分配）的混合任务分配策略来优化传感覆盖。该策略假设网格热量来源于历史用户轨迹数据，并且无人机只分配任务而没有自主控制。MUMUHTA使用用户轨迹预测和贪婪招募机会用户，以及基于Kuhn-Munkres算法的每插槽无人机匹配，同时动态切换规则决定无人机何时接管用户的任务。基于罗马用户轨迹数据集和上海电信任务数据集的仿真实验表明，与MPU、mu - random、DLMV-MPU、DLMV(T)-MPU和HR-DLVCS等基准策略相比，MUMUHTA的任务完成率平均提高了29.70%、8.11%、9.69%、5.05%和15.38%。

{"title":"Optimizing task allocation in Mobile Crowdsensing with multiple opportunistic users and participatory UAVs","authors":"Shi Yu, Bing Shi, Xiao Su, Saisai Li, Shuai Li, Xing Tang","doi":"10.1016/j.adhoc.2025.104128","DOIUrl":"10.1016/j.adhoc.2025.104128","url":null,"abstract":"<div><div>Mobile Crowdsensing (MCS) refers to the use of human users or unmanned aerial vehicles (UAVs) equipped with mobile devices to collect sensing data. However, existing MCS methods suffer from insufficient coverage due to sparse population distribution in some sensing areas and restrictions on UAV flights in densely populated urban regions. To address this issue, we propose a collaborative approach that combines opportunistic users, who are passively engaged and widely distributed across cities, with participatory UAVs, which are capable of covering sparsely populated regions. Specifically, we introduce a hybrid task allocation strategy called MUMUHTA (Multi-User Multi-UAV Hybrid Task Allocation) to optimize sensing coverage. The strategy assumes that grid heat is derived from historical user trajectory data, and UAVs are assigned only tasks without autonomous control. MUMUHTA uses user trajectory prediction and greedy recruitment for opportunistic users, along with per-slot UAV matching based on the Kuhn–Munkres algorithm, while a dynamic switching rule determines when UAVs take over tasks from users. Simulation experiments using the Rome user trajectory dataset and Shanghai Telecom task dataset show that MUMUHTA improves the task completion rate by an average of 29.70%, 8.11%, 9.69%, 5.05%, and 15.38% compared to benchmark strategies: MPU, MOU-Random, DLMV-MPU, DLMV(T)-MPU, and HR-DLVCS.</div></div>","PeriodicalId":55555,"journal":{"name":"Ad Hoc Networks","volume":"183 ","pages":"Article 104128"},"PeriodicalIF":4.8,"publicationDate":"2025-12-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145884181","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