首页 > 最新文献

Ad Hoc Networks最新文献

英文 中文
MobiAuth: Blockchain-driven decentralized authentication for enhanced security and privacy in mobile networks MobiAuth:区块链驱动的分散身份验证,用于增强移动网络的安全性和隐私性
IF 4.8 3区 计算机科学 Q1 COMPUTER SCIENCE, INFORMATION SYSTEMS Pub Date : 2026-02-01 Epub Date: 2025-10-30 DOI: 10.1016/j.adhoc.2025.104064
Narendra K. Dewangan , Gauri Shankar
Decentralized authentication in dynamic mobile networks faces significant challenges due to high node mobility, resource constraints, and vulnerabilities to side-channel attacks. In this work, we present MobiAuth, a blockchain-driven framework based on Hyperledger Iroha and OMNET++ that enables secure, peer-to-peer authentication using compact Ed25519 signatures and ephemeral session keys. Our protocol eliminates single points of failure by distributing trust across a permissioned ledger and employs constant-time cryptographic operations to thwart timing and power-analysis attacks. We validate MobiAuth through co-simulation in OMNET++ integrated with Iroha via a Python gRPC bridge and benchmark its performance with Hyperledger Caliper.
Simulation yields 95% packet delivery with an authentication latency ranging from 12 ms in the only OMNeT++ and baseline to 20–150ms in the full ledger-integrated system, and a ledger write throughput of 250tps. Comparative experiments demonstrate a 33% reduction in communication overhead and robust operation under random Control Point failures and Byzantine Access Node behavior. Analysis of on-device ledger synchronization further highlights practical storage growth and bandwidth requirements for long-term deployment. These results indicate that MobiAuth achieves strong security and privacy with modest energy impact, scalable performance, and compatibility with mobile devices in real-world network environments.
动态移动网络中的去中心化认证由于节点的高移动性、资源的有限性以及易受侧信道攻击而面临着巨大的挑战。在这项工作中,我们提出了MobiAuth,这是一个基于Hyperledger Iroha和omnet++的区块链驱动框架,它使用紧凑的Ed25519签名和临时会话密钥实现安全的点对点身份验证。我们的协议通过在许可的分类账上分配信任来消除单点故障,并采用恒定时间加密操作来阻止定时和功率分析攻击。我们通过Python gRPC桥在与Iroha集成的omnet++中进行联合仿真验证了MobiAuth,并使用Hyperledger Caliper对其性能进行了基准测试。仿真结果显示95%的数据包交付,身份验证延迟范围从仅有的omnet++和基线中的12 ms到完整的分类账集成系统中的20-150ms,分类账写吞吐量为250tps。对比实验表明,在随机控制点故障和拜占庭访问节点行为下,通信开销降低了33%,并且具有鲁棒性。对设备上分类账同步的分析进一步强调了长期部署的实际存储增长和带宽需求。这些结果表明,MobiAuth实现了强大的安全性和隐私性,具有适度的能源影响、可扩展的性能以及与现实网络环境中的移动设备的兼容性。
{"title":"MobiAuth: Blockchain-driven decentralized authentication for enhanced security and privacy in mobile networks","authors":"Narendra K. Dewangan ,&nbsp;Gauri Shankar","doi":"10.1016/j.adhoc.2025.104064","DOIUrl":"10.1016/j.adhoc.2025.104064","url":null,"abstract":"<div><div>Decentralized authentication in dynamic mobile networks faces significant challenges due to high node mobility, resource constraints, and vulnerabilities to side-channel attacks. In this work, we present <strong>MobiAuth</strong>, a blockchain-driven framework based on Hyperledger Iroha and OMNET<span>++</span> that enables secure, peer-to-peer authentication using compact Ed25519 signatures and ephemeral session keys. Our protocol eliminates single points of failure by distributing trust across a permissioned ledger and employs constant-time cryptographic operations to thwart timing and power-analysis attacks. We validate MobiAuth through co-simulation in OMNET<span>++</span> integrated with Iroha via a Python gRPC bridge and benchmark its performance with Hyperledger Caliper.</div><div>Simulation yields 95% packet delivery with an authentication latency ranging from 12 ms in the only OMNeT<span>++</span> and baseline to 20–150ms in the full ledger-integrated system, and a ledger write throughput of 250tps. Comparative experiments demonstrate a 33% reduction in communication overhead and robust operation under random Control Point failures and Byzantine Access Node behavior. Analysis of on-device ledger synchronization further highlights practical storage growth and bandwidth requirements for long-term deployment. These results indicate that MobiAuth achieves strong security and privacy with modest energy impact, scalable performance, and compatibility with mobile devices in real-world network environments.</div></div>","PeriodicalId":55555,"journal":{"name":"Ad Hoc Networks","volume":"181 ","pages":"Article 104064"},"PeriodicalIF":4.8,"publicationDate":"2026-02-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145468555","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-Technology Interference awareness for multi-user OFDMA scheduling in IEEE 802.11ax IEEE 802.11ax中多用户OFDMA调度的跨技术干扰感知
IF 4.8 3区 计算机科学 Q1 COMPUTER SCIENCE, INFORMATION SYSTEMS Pub Date : 2026-02-01 Epub Date: 2025-10-27 DOI: 10.1016/j.adhoc.2025.104057
Thijs Havinga, Xianjun Jiao, Wei Liu, Baiheng Chen, Robbe Gaeremynck, Ingrid Moerman
Cross-Technology Interference (CTI) significantly degrades the performance of heterogeneous wireless communication systems operating within a shared spectrum. Traditional mitigation techniques, such as Clear Channel Assessment (CCA), often fail due to, amongst others, varying bandwidth and detection threshold. Orthogonal Frequency Division Multiple Access (OFDMA), introduced to Wi-Fi in IEEE 802.11ax (Wi-Fi 6), allows multiple users to be served simultaneously using distinct subcarrier sets, known as Resource Units (RUs), providing enhanced flexibility in the frequency domain. This paper explores and evaluates several Wi-Fi 6 compliant methods for multi-user OFDMA scheduling with CTI awareness. Through simulations, we assess the benefits of different techniques in various scenarios, in terms of either total throughput or average latency. To effectively apply the mitigation techniques, we propose a methodology that incorporates CTI feedback from stations and real-time CCA per RU. Given that commercial Wi-Fi 6 access points lack control over low-level OFDMA features, we use openwifi, a full-stack Wi-Fi transceiver running on software-defined radio, to implement the CTI-aware OFDMA scheduler. Real-life experiments validate the effectiveness of the scheduler and confirm its real-time performance capabilities.
跨技术干扰(CTI)会显著降低在共享频谱内运行的异构无线通信系统的性能。传统的缓解技术,如清晰通道评估(CCA),往往由于带宽和检测阈值的变化等原因而失败。在IEEE 802.11ax (Wi-Fi 6)中引入的正交频分多址(OFDMA)允许使用不同的子载波集(称为资源单元(ru))同时为多个用户提供服务,从而在频域提供增强的灵活性。本文探索并评估了几种符合Wi-Fi 6的具有CTI感知的多用户OFDMA调度方法。通过模拟,我们从总吞吐量或平均延迟方面评估了不同技术在各种场景中的优势。为了有效地应用缓解技术,我们提出了一种方法,该方法结合了来自站点的CTI反馈和每个RU的实时CCA。鉴于商用Wi-Fi 6接入点缺乏对低级OFDMA功能的控制,我们使用openwifi,一种运行在软件定义无线电上的全栈Wi-Fi收发器,来实现cti感知的OFDMA调度器。实际实验验证了调度程序的有效性,并证实了其实时性能。
{"title":"Cross-Technology Interference awareness for multi-user OFDMA scheduling in IEEE 802.11ax","authors":"Thijs Havinga,&nbsp;Xianjun Jiao,&nbsp;Wei Liu,&nbsp;Baiheng Chen,&nbsp;Robbe Gaeremynck,&nbsp;Ingrid Moerman","doi":"10.1016/j.adhoc.2025.104057","DOIUrl":"10.1016/j.adhoc.2025.104057","url":null,"abstract":"<div><div>Cross-Technology Interference (CTI) significantly degrades the performance of heterogeneous wireless communication systems operating within a shared spectrum. Traditional mitigation techniques, such as Clear Channel Assessment (CCA), often fail due to, amongst others, varying bandwidth and detection threshold. Orthogonal Frequency Division Multiple Access (OFDMA), introduced to Wi-Fi in IEEE 802.11ax (Wi-Fi 6), allows multiple users to be served simultaneously using distinct subcarrier sets, known as Resource Units (RUs), providing enhanced flexibility in the frequency domain. This paper explores and evaluates several Wi-Fi 6 compliant methods for multi-user OFDMA scheduling with CTI awareness. Through simulations, we assess the benefits of different techniques in various scenarios, in terms of either total throughput or average latency. To effectively apply the mitigation techniques, we propose a methodology that incorporates CTI feedback from stations and real-time CCA per RU. Given that commercial Wi-Fi 6 access points lack control over low-level OFDMA features, we use <em>openwifi</em>, a full-stack Wi-Fi transceiver running on software-defined radio, to implement the CTI-aware OFDMA scheduler. Real-life experiments validate the effectiveness of the scheduler and confirm its real-time performance capabilities.</div></div>","PeriodicalId":55555,"journal":{"name":"Ad Hoc Networks","volume":"181 ","pages":"Article 104057"},"PeriodicalIF":4.8,"publicationDate":"2026-02-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145419843","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
Enhancement of fuzzy trust evaluation via autoencoders in game-theory based security distributed Wireless Sensor Networks 基于博弈论的安全分布式无线传感器网络中自编码器增强模糊信任评价
IF 4.8 3区 计算机科学 Q1 COMPUTER SCIENCE, INFORMATION SYSTEMS Pub Date : 2026-02-01 Epub Date: 2025-11-04 DOI: 10.1016/j.adhoc.2025.104068
Xuegang Wu , Xu Zhang , Kun Gou , Jianfeng Wang
With the rapid development of Internet of Things (IoT) technology, Wireless Sensor Networks (WSNs) have been widely applied in various fields, such as environmental monitoring, health detection, and smart homes. Since wireless sensor nodes are typically deployed in open and dynamically changing environments, the data are susceptible to interference and outliers, directly affecting the network’s reliability and decision-making accuracy. Therefore, effectively detecting and processing anomalous data has become a critical issue for the regular operation of WSNs.
This paper proposes techniques to enhance trust evaluation in game-theory based WSNs using autoencoders. These techniques effectively address the security challenges inherent in WSNs. The protocol first assigns a trust value to each node through fuzzy trust evaluation and recommendation mechanisms to differentiate between reliable and abnormal data. Subsequently, the game theory is used to select the optimal cluster head to optimize inter-node interactions and resource allocation, thereby enhancing overall network performance and robustness. Finally, an autoencoder is employed for dimensionality reduction and data reconstruction, with abnormal data being precisely detected by comparing the reconstruction error against a preset threshold and incorporating historical trust information.
Experimental results demonstrate that the proposed methods achieve high detection accuracy with low false positive or false negative rates in practical applications. Compared with traditional approaches, the proposed techniques exhibit stronger robustness and higher detection efficiency in complex environments, effectively meeting the demands of large-scale WSNs applications.
随着物联网(IoT)技术的快速发展,无线传感器网络(WSNs)在环境监测、健康检测、智能家居等各个领域得到广泛应用。由于无线传感器节点通常部署在开放和动态变化的环境中,数据容易受到干扰和异常值的影响,直接影响网络的可靠性和决策精度。因此,有效地检测和处理异常数据已成为无线传感器网络正常运行的关键问题。本文提出了利用自编码器增强基于博弈论的无线传感器网络信任评估的技术。这些技术有效地解决了无线传感器网络固有的安全挑战。该协议首先通过模糊信任评价和推荐机制为每个节点分配一个信任值,区分可靠数据和异常数据。随后,利用博弈论选择最优簇头,优化节点间交互和资源分配,从而提高整体网络性能和鲁棒性。最后,采用自编码器进行降维和数据重构,通过将重构误差与预设阈值进行比较,并结合历史信任信息,精确检测异常数据。实验结果表明,该方法在实际应用中具有较高的检测精度和较低的假阳性或假阴性率。与传统方法相比,该方法在复杂环境下具有更强的鲁棒性和更高的检测效率,有效地满足了大规模WSNs应用的需求。
{"title":"Enhancement of fuzzy trust evaluation via autoencoders in game-theory based security distributed Wireless Sensor Networks","authors":"Xuegang Wu ,&nbsp;Xu Zhang ,&nbsp;Kun Gou ,&nbsp;Jianfeng Wang","doi":"10.1016/j.adhoc.2025.104068","DOIUrl":"10.1016/j.adhoc.2025.104068","url":null,"abstract":"<div><div>With the rapid development of Internet of Things (IoT) technology, Wireless Sensor Networks (WSNs) have been widely applied in various fields, such as environmental monitoring, health detection, and smart homes. Since wireless sensor nodes are typically deployed in open and dynamically changing environments, the data are susceptible to interference and outliers, directly affecting the network’s reliability and decision-making accuracy. Therefore, effectively detecting and processing anomalous data has become a critical issue for the regular operation of WSNs.</div><div>This paper proposes techniques to enhance trust evaluation in game-theory based WSNs using autoencoders. These techniques effectively address the security challenges inherent in WSNs. The protocol first assigns a trust value to each node through fuzzy trust evaluation and recommendation mechanisms to differentiate between reliable and abnormal data. Subsequently, the game theory is used to select the optimal cluster head to optimize inter-node interactions and resource allocation, thereby enhancing overall network performance and robustness. Finally, an autoencoder is employed for dimensionality reduction and data reconstruction, with abnormal data being precisely detected by comparing the reconstruction error against a preset threshold and incorporating historical trust information.</div><div>Experimental results demonstrate that the proposed methods achieve high detection accuracy with low false positive or false negative rates in practical applications. Compared with traditional approaches, the proposed techniques exhibit stronger robustness and higher detection efficiency in complex environments, effectively meeting the demands of large-scale WSNs applications.</div></div>","PeriodicalId":55555,"journal":{"name":"Ad Hoc Networks","volume":"181 ","pages":"Article 104068"},"PeriodicalIF":4.8,"publicationDate":"2026-02-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145468466","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
Decentralized opportunistic crowdsensing task allocation with global and local communication 分散的机会主义众感任务分配与全球和本地通信
IF 4.8 3区 计算机科学 Q1 COMPUTER SCIENCE, INFORMATION SYSTEMS Pub Date : 2026-02-01 Epub Date: 2025-11-01 DOI: 10.1016/j.adhoc.2025.104069
Chunyu Tu , Yanghui Chen , Zhiyong Yu , Fangwan Huang , Yuezhong Wu , Xianwei Guo , Chao Yang , Runhe Huang
Traditional opportunistic crowdsensing is usually managed by a central platform that assigns tasks, which imposes significant demands on the platform’s performance and increases the risk of privacy breaches for participants. To address these issues, this paper proposes a decentralized opportunistic sensing-based solution to achieve task allocation under a global budget constraint and maximize task coverage. Unlike traditional approaches, this solution allows participants to decide whether to join sensing tasks autonomously, forming a collaborative multi-agent system. In this decentralized environment, ensuring efficient task allocation while adhering to global constraints poses the main challenge. Since the communication conditions between participants directly affect the execution and coordination efficiency of global constraints, this paper designs corresponding global and local communication algorithms to address large-scale decentralized task allocation. Under global communication conditions, we adopt the concept of a cellular model and propose a decentralized genetic algorithm (D-GA). This approach utilizes global communication to evaluate the fitness of individuals, enabling coordinated problem-solving among agents. In addition, we consider local communication and design a decentralized soft-constrained probabilistic decision algorithm (DS-PD) to address scenarios where global communication is impractical. D-GA achieves 59.1%–81.2% task coverage, performing within 4.5%–5.8% of centralized integer linear programming while reducing execution time by over 96%. DS-PD significantly outperforms all methods under node failures, maintaining robust performance with only 2.1%–3.4% degradation.
传统的机会主义众感通常由一个分配任务的中央平台管理,这对平台的性能提出了很高的要求,并增加了参与者隐私泄露的风险。为了解决这些问题,本文提出了一种基于分散式机会感知的解决方案,以实现全局预算约束下的任务分配,并最大化任务覆盖率。与传统方法不同,该解决方案允许参与者自主决定是否加入感知任务,形成一个协作的多智能体系统。在这种分散的环境中,在遵守全局约束的同时确保有效的任务分配是主要的挑战。由于参与者之间的通信条件直接影响全局约束的执行和协调效率,本文设计了相应的全局和局部通信算法来解决大规模分散的任务分配问题。在全局通信条件下,采用细胞模型的概念,提出了一种分散的遗传算法(D-GA)。这种方法利用全局通信来评估个体的适应度,使代理之间能够协调解决问题。此外,我们考虑了本地通信,并设计了一个分散的软约束概率决策算法(DS-PD)来解决全球通信不切实际的情况。D-GA实现了59.1%-81.2%的任务覆盖率,在集中式整数线性规划的4.5%-5.8%范围内执行,同时将执行时间减少了96%以上。在节点故障情况下,DS-PD显著优于所有方法,仅以2.1%-3.4%的性能下降保持了稳健的性能。
{"title":"Decentralized opportunistic crowdsensing task allocation with global and local communication","authors":"Chunyu Tu ,&nbsp;Yanghui Chen ,&nbsp;Zhiyong Yu ,&nbsp;Fangwan Huang ,&nbsp;Yuezhong Wu ,&nbsp;Xianwei Guo ,&nbsp;Chao Yang ,&nbsp;Runhe Huang","doi":"10.1016/j.adhoc.2025.104069","DOIUrl":"10.1016/j.adhoc.2025.104069","url":null,"abstract":"<div><div>Traditional opportunistic crowdsensing is usually managed by a central platform that assigns tasks, which imposes significant demands on the platform’s performance and increases the risk of privacy breaches for participants. To address these issues, this paper proposes a decentralized opportunistic sensing-based solution to achieve task allocation under a global budget constraint and maximize task coverage. Unlike traditional approaches, this solution allows participants to decide whether to join sensing tasks autonomously, forming a collaborative multi-agent system. In this decentralized environment, ensuring efficient task allocation while adhering to global constraints poses the main challenge. Since the communication conditions between participants directly affect the execution and coordination efficiency of global constraints, this paper designs corresponding global and local communication algorithms to address large-scale decentralized task allocation. Under global communication conditions, we adopt the concept of a cellular model and propose a decentralized genetic algorithm (D-GA). This approach utilizes global communication to evaluate the fitness of individuals, enabling coordinated problem-solving among agents. In addition, we consider local communication and design a decentralized soft-constrained probabilistic decision algorithm (DS-PD) to address scenarios where global communication is impractical. D-GA achieves 59.1%–81.2% task coverage, performing within 4.5%–5.8% of centralized integer linear programming while reducing execution time by over 96%. DS-PD significantly outperforms all methods under node failures, maintaining robust performance with only 2.1%–3.4% degradation.</div></div>","PeriodicalId":55555,"journal":{"name":"Ad Hoc Networks","volume":"181 ","pages":"Article 104069"},"PeriodicalIF":4.8,"publicationDate":"2026-02-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145468463","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
Optimization of ship engine room communication based on ray tracing method and ZigBee technology 基于光线追踪法和ZigBee技术的船舶机舱通信优化
IF 4.8 3区 计算机科学 Q1 COMPUTER SCIENCE, INFORMATION SYSTEMS Pub Date : 2026-02-01 Epub Date: 2025-10-28 DOI: 10.1016/j.adhoc.2025.104067
Lige Yuan , Yangzhou Hao , Yinghua Li
The communication quality in the engine room of the ship is not ideal because of the complicated internal environment of the ship during the voyage at sea. The communication quality of the existing communication methods is not high, which is affected by the environment of the engine room, and the signal is unstable and the transmission fails. In view of this situation, a communication optimization method for ship engine room based on ZigBee technology and ray tracing method is proposed. Firstly, ZigBee technology was used to design the networking structure, and simulation software and ray final method were used to simulate the ship scene and radio wave transmission. In addition, the state of the communication signal is predicted by the short-time memory network and the communication scheme is adjusted in time. Genetic algorithm is used to optimize the antenna parameters. Finally, according to the simulation model, the design network structure and node layout parameters are optimized, and the communication optimization of the ship engine room is finally realized. The experimental results show that the transmission rate, data packet transmission delay time and transmission error rate indicators of the communication system after using the communication optimization scheme designed in the research are 214.05kbps, 20.14ms and 0.52 % respectively. And its node coverage exceeds 90 %. It can be seen that the communication optimization method designed in this study can effectively improve the quality of communication transmission data, improve the transmission efficiency, and provide reliable communication technology support for ship navigation. The research method provides a new idea for the design of wireless communication system in similar complex environment.
船舶在海上航行时,由于船舶内部环境复杂,导致机舱通信质量不理想。现有通信方式的通信质量不高,受机舱环境影响,信号不稳定,传输失败。针对这种情况,提出了一种基于ZigBee技术和光线追踪方法的船舶机舱通信优化方法。首先,采用ZigBee技术进行组网结构设计,利用仿真软件和ray final方法对船舶场景和无线电波传输进行仿真。此外,利用短时记忆网络预测通信信号的状态,及时调整通信方案。采用遗传算法对天线参数进行优化。最后,根据仿真模型对设计网络结构和节点布局参数进行优化,最终实现船舶机舱通信优化。实验结果表明,采用本研究设计的通信优化方案后,通信系统的传输速率、数据包传输延迟时间和传输错误率指标分别为214.05kbps、20.14ms和0.52%。节点覆盖率超过90%。可见,本研究设计的通信优化方法可以有效提高通信传输数据的质量,提高传输效率,为船舶导航提供可靠的通信技术支持。该研究方法为类似复杂环境下的无线通信系统设计提供了一种新的思路。
{"title":"Optimization of ship engine room communication based on ray tracing method and ZigBee technology","authors":"Lige Yuan ,&nbsp;Yangzhou Hao ,&nbsp;Yinghua Li","doi":"10.1016/j.adhoc.2025.104067","DOIUrl":"10.1016/j.adhoc.2025.104067","url":null,"abstract":"<div><div>The communication quality in the engine room of the ship is not ideal because of the complicated internal environment of the ship during the voyage at sea. The communication quality of the existing communication methods is not high, which is affected by the environment of the engine room, and the signal is unstable and the transmission fails. In view of this situation, a communication optimization method for ship engine room based on ZigBee technology and ray tracing method is proposed. Firstly, ZigBee technology was used to design the networking structure, and simulation software and ray final method were used to simulate the ship scene and radio wave transmission. In addition, the state of the communication signal is predicted by the short-time memory network and the communication scheme is adjusted in time. Genetic algorithm is used to optimize the antenna parameters. Finally, according to the simulation model, the design network structure and node layout parameters are optimized, and the communication optimization of the ship engine room is finally realized. The experimental results show that the transmission rate, data packet transmission delay time and transmission error rate indicators of the communication system after using the communication optimization scheme designed in the research are 214.05kbps, 20.14ms and 0.52 % respectively. And its node coverage exceeds 90 %. It can be seen that the communication optimization method designed in this study can effectively improve the quality of communication transmission data, improve the transmission efficiency, and provide reliable communication technology support for ship navigation. The research method provides a new idea for the design of wireless communication system in similar complex environment.</div></div>","PeriodicalId":55555,"journal":{"name":"Ad Hoc Networks","volume":"181 ","pages":"Article 104067"},"PeriodicalIF":4.8,"publicationDate":"2026-02-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145468462","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
Stochastic Fractal Search with Reinforcement Learning for dynamic routing in resource-constrained IoT networks 基于强化学习的随机分形搜索在资源受限的物联网网络中的动态路由
IF 4.8 3区 计算机科学 Q1 COMPUTER SCIENCE, INFORMATION SYSTEMS Pub Date : 2026-02-01 Epub Date: 2025-11-04 DOI: 10.1016/j.adhoc.2025.104083
Anupama Mondal , Dipankar Ch. Barman , Nabajyoti Mazumdar
The Internet of Things (IoT) enables pervasive connectivity and large-scale data exchange among heterogeneous devices, driving advancements in intelligent automation and decision-making. However, the resource-constrained nature of IoT nodes and the increasing data traffic in dense deployments demand energy-efficient and adaptive communication strategies. This paper presents a hybrid framework that integrates Stochastic Fractal Search (SFS) for energy-efficient, trust-aware clustering with Q-Learning-based Reinforcement Learning (RL) for dynamic multi-hop routing in IoT and edge-assisted environments. The SFS-based clustering mechanism optimizes cluster-head selection by minimizing intra-cluster energy consumption and ensuring balanced load distribution, while the RL-based routing module adaptively learns optimal forwarding paths based on residual energy, link distance, and trust metrics. A dual trust model, incorporating both energy trust and data trust, is employed to detect and isolate unreliable nodes, thereby enhancing network reliability and security. The proposed SFSRL framework effectively reduces redundant transmissions, balances energy utilization, and enhances scalability across heterogeneous IoT-edge infrastructures. Simulation results validate that SFSRL outperforms existing protocols such as TEC-SFS, RINA, GATERP, and MODRL in terms of energy efficiency, network lifetime, Packet Delivery Ratio (PDR), data rate, throughput, and end-to-end delay, establishing it as a robust and scalable solution for next-generation IoT applications.
物联网(IoT)实现了异构设备之间的普遍连接和大规模数据交换,推动了智能自动化和决策的进步。然而,物联网节点的资源约束性质和密集部署中不断增加的数据流量需要节能和自适应通信策略。本文提出了一个混合框架,该框架将随机分形搜索(SFS)与基于q学习的强化学习(RL)集成在一起,用于物联网和边缘辅助环境中的动态多跳路由。基于sfs的聚类机制通过最小化簇内能量消耗和均衡负载分配来优化簇头选择,而基于rl的路由模块则根据剩余能量、链路距离和信任指标自适应学习最优转发路径。采用能量信任和数据信任相结合的双重信任模型,检测和隔离不可靠节点,提高网络的可靠性和安全性。提出的SFSRL框架有效地减少了冗余传输,平衡了能源利用,增强了跨异构物联网边缘基础设施的可扩展性。仿真结果验证了SFSRL在能效、网络寿命、分组传输比(PDR)、数据速率、吞吐量和端到端延迟方面优于TEC-SFS、RINA、GATERP和MODRL等现有协议,使其成为下一代物联网应用的强大且可扩展的解决方案。
{"title":"Stochastic Fractal Search with Reinforcement Learning for dynamic routing in resource-constrained IoT networks","authors":"Anupama Mondal ,&nbsp;Dipankar Ch. Barman ,&nbsp;Nabajyoti Mazumdar","doi":"10.1016/j.adhoc.2025.104083","DOIUrl":"10.1016/j.adhoc.2025.104083","url":null,"abstract":"<div><div>The Internet of Things (IoT) enables pervasive connectivity and large-scale data exchange among heterogeneous devices, driving advancements in intelligent automation and decision-making. However, the resource-constrained nature of IoT nodes and the increasing data traffic in dense deployments demand energy-efficient and adaptive communication strategies. This paper presents a hybrid framework that integrates Stochastic Fractal Search (SFS) for energy-efficient, trust-aware clustering with Q-Learning-based Reinforcement Learning (RL) for dynamic multi-hop routing in IoT and edge-assisted environments. The SFS-based clustering mechanism optimizes cluster-head selection by minimizing intra-cluster energy consumption and ensuring balanced load distribution, while the RL-based routing module adaptively learns optimal forwarding paths based on residual energy, link distance, and trust metrics. A dual trust model, incorporating both energy trust and data trust, is employed to detect and isolate unreliable nodes, thereby enhancing network reliability and security. The proposed SFSRL framework effectively reduces redundant transmissions, balances energy utilization, and enhances scalability across heterogeneous IoT-edge infrastructures. Simulation results validate that SFSRL outperforms existing protocols such as TEC-SFS, RINA, GATERP, and MODRL in terms of energy efficiency, network lifetime, Packet Delivery Ratio (PDR), data rate, throughput, and end-to-end delay, establishing it as a robust and scalable solution for next-generation IoT applications.</div></div>","PeriodicalId":55555,"journal":{"name":"Ad Hoc Networks","volume":"181 ","pages":"Article 104083"},"PeriodicalIF":4.8,"publicationDate":"2026-02-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145468554","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 dual-mode framework for indoor localization via temporal learning and knowledge distillation 基于时间学习和知识升华的室内定位双模框架
IF 4.8 3区 计算机科学 Q1 COMPUTER SCIENCE, INFORMATION SYSTEMS Pub Date : 2026-02-01 Epub Date: 2025-11-05 DOI: 10.1016/j.adhoc.2025.104089
Huang Lin, Yan Chen, Shuo Li, Wei Peng
WiFi fingerprinting is a widely used approach for indoor localization, but its effectiveness is limited by signal instability and the challenges of real-time inference. To address these issues, we propose a dual-mode localization framework that combines a sequence-based long short-term memory (LSTM) model with a lightweight multilayer perceptron (MLP) trained through knowledge distillation. This framework supports both sequential and single-snapshot RSS inputs and reformulates localization as a block-wise classification task to enhance robustness. Experiments on two large-scale public datasets show that the distilled MLP model achieves more than 20% improvement in localization error compared to a non-distilled baseline, while maintaining high floor-prediction accuracy. This allows for fast, efficient inference on devices with limited computational resources while maintaining high accuracy levels. The dual-mode design enables adaptive selection based on input availability, which offers a flexible and practical solution for real-world indoor positioning in dynamic environments.
WiFi指纹识别是一种广泛应用于室内定位的方法,但其有效性受到信号不稳定性和实时推理的挑战。为了解决这些问题,我们提出了一种双模式定位框架,该框架结合了基于序列的长短期记忆(LSTM)模型和通过知识蒸馏训练的轻量级多层感知器(MLP)。该框架支持顺序和单快照RSS输入,并将定位重新定义为块分类任务,以增强鲁棒性。在两个大型公共数据集上的实验表明,与未蒸馏的基线相比,经过蒸馏的MLP模型的定位误差提高了20%以上,同时保持了较高的地板预测精度。这允许在计算资源有限的设备上进行快速、高效的推理,同时保持高精度水平。双模设计支持基于输入可用性的自适应选择,为动态环境下的真实室内定位提供了灵活实用的解决方案。
{"title":"A dual-mode framework for indoor localization via temporal learning and knowledge distillation","authors":"Huang Lin,&nbsp;Yan Chen,&nbsp;Shuo Li,&nbsp;Wei Peng","doi":"10.1016/j.adhoc.2025.104089","DOIUrl":"10.1016/j.adhoc.2025.104089","url":null,"abstract":"<div><div>WiFi fingerprinting is a widely used approach for indoor localization, but its effectiveness is limited by signal instability and the challenges of real-time inference. To address these issues, we propose a dual-mode localization framework that combines a sequence-based long short-term memory (LSTM) model with a lightweight multilayer perceptron (MLP) trained through knowledge distillation. This framework supports both sequential and single-snapshot RSS inputs and reformulates localization as a block-wise classification task to enhance robustness. Experiments on two large-scale public datasets show that the distilled MLP model achieves more than 20% improvement in localization error compared to a non-distilled baseline, while maintaining high floor-prediction accuracy. This allows for fast, efficient inference on devices with limited computational resources while maintaining high accuracy levels. The dual-mode design enables adaptive selection based on input availability, which offers a flexible and practical solution for real-world indoor positioning in dynamic environments.</div></div>","PeriodicalId":55555,"journal":{"name":"Ad Hoc Networks","volume":"181 ","pages":"Article 104089"},"PeriodicalIF":4.8,"publicationDate":"2026-02-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145468552","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
Task offloading in satellite–MEC networks for latency-sensitive IoT applications: A martingale-based game approach 针对延迟敏感物联网应用的卫星- mec网络任务卸载:基于鞅的游戏方法
IF 4.8 3区 计算机科学 Q1 COMPUTER SCIENCE, INFORMATION SYSTEMS Pub Date : 2026-02-01 Epub Date: 2025-10-24 DOI: 10.1016/j.adhoc.2025.104052
Xintong Pei , Zhenjiang Zhang , HanChieh Chao , Zihang Yu , Wenhui Wang
To address the growing demand for latency-sensitive Internet of Things (IoT) applications in remote regions, satellite–integrated mobile edge computing (SMEC) deploys computational resources on Low Earth Orbit (LEO) satellites to provide seamless network access and edge computing services for IoT devices lacking terrestrial network coverage. However, ensuring quality of service (QoS) for latency-sensitive tasks in SMEC systems remains challenging due to two critical limitations: existing deterministic models cannot adequately capture the stochastic nature of discontinuous satellite–ground links, leading to inaccurate queuing delay predictions, and current approaches fail to adapt to dynamic user task preferences that significantly impact offloading effectiveness. In response to these challenges, this paper develops a novel dynamic task offloading framework that integrates martingale-based delay analysis with adaptive game-theoretic optimization. We employ Markov Chain Monte Carlo (MCMC) methods to characterize discontinuous satellite–ground stochastic service processes and apply martingale theory to derive tight statistical delay guarantees that significantly outperform conventional moment generating function approaches. Based on this analytical foundation, we formulate multi-task offloading as an exact potential game and propose the Temporal-Enhanced Stochastic Learning (TESL) algorithm, which leverages historical trend learning and environmental dynamics detection to achieve robust convergence in non-stationary environments. Experimental results demonstrate that TESL achieves a 37% lower delay violation probability compared to the best-performing baseline algorithm, exhibiting superior convergence efficiency and adaptability while improving utility, making it well-suited for practical SMEC deployments.
为了满足偏远地区对延迟敏感的物联网(IoT)应用日益增长的需求,卫星集成移动边缘计算(SMEC)将计算资源部署在低地球轨道(LEO)卫星上,为缺乏地面网络覆盖的物联网设备提供无缝网络接入和边缘计算服务。然而,由于两个关键的限制,确保SMEC系统中延迟敏感任务的服务质量(QoS)仍然具有挑战性:现有的确定性模型不能充分捕捉不连续卫星-地面链路的随机性,导致不准确的排队延迟预测,以及当前的方法不能适应动态用户任务偏好,这将显著影响卸载效率。针对这些挑战,本文开发了一种新的动态任务卸载框架,该框架将基于鞅的延迟分析与自适应博弈论优化相结合。我们采用马尔可夫链蒙特卡罗(MCMC)方法来表征不连续的卫星-地面随机服务过程,并应用鞅理论推导出严格的统计延迟保证,显著优于传统的矩生成函数方法。基于此分析基础,我们将多任务卸载作为一种精确的潜在博弈,并提出了时间增强随机学习(TESL)算法,该算法利用历史趋势学习和环境动态检测来实现非平稳环境下的鲁棒收敛。实验结果表明,与性能最好的基线算法相比,TESL算法的延迟违反概率降低了37%,在提高效用的同时表现出优越的收敛效率和适应性,非常适合实际的SMEC部署。
{"title":"Task offloading in satellite–MEC networks for latency-sensitive IoT applications: A martingale-based game approach","authors":"Xintong Pei ,&nbsp;Zhenjiang Zhang ,&nbsp;HanChieh Chao ,&nbsp;Zihang Yu ,&nbsp;Wenhui Wang","doi":"10.1016/j.adhoc.2025.104052","DOIUrl":"10.1016/j.adhoc.2025.104052","url":null,"abstract":"<div><div>To address the growing demand for latency-sensitive Internet of Things (IoT) applications in remote regions, satellite–integrated mobile edge computing (SMEC) deploys computational resources on Low Earth Orbit (LEO) satellites to provide seamless network access and edge computing services for IoT devices lacking terrestrial network coverage. However, ensuring quality of service (QoS) for latency-sensitive tasks in SMEC systems remains challenging due to two critical limitations: existing deterministic models cannot adequately capture the stochastic nature of discontinuous satellite–ground links, leading to inaccurate queuing delay predictions, and current approaches fail to adapt to dynamic user task preferences that significantly impact offloading effectiveness. In response to these challenges, this paper develops a novel dynamic task offloading framework that integrates martingale-based delay analysis with adaptive game-theoretic optimization. We employ Markov Chain Monte Carlo (MCMC) methods to characterize discontinuous satellite–ground stochastic service processes and apply martingale theory to derive tight statistical delay guarantees that significantly outperform conventional moment generating function approaches. Based on this analytical foundation, we formulate multi-task offloading as an exact potential game and propose the Temporal-Enhanced Stochastic Learning (TESL) algorithm, which leverages historical trend learning and environmental dynamics detection to achieve robust convergence in non-stationary environments. Experimental results demonstrate that TESL achieves a 37% lower delay violation probability compared to the best-performing baseline algorithm, exhibiting superior convergence efficiency and adaptability while improving utility, making it well-suited for practical SMEC deployments.</div></div>","PeriodicalId":55555,"journal":{"name":"Ad Hoc Networks","volume":"181 ","pages":"Article 104052"},"PeriodicalIF":4.8,"publicationDate":"2026-02-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145419847","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
MQTTEEB-D: A high-fidelity benchmark for real-time MQTT anomaly detection using machine learning techniques MQTTEEB-D:使用机器学习技术进行实时MQTT异常检测的高保真基准
IF 4.8 3区 计算机科学 Q1 COMPUTER SCIENCE, INFORMATION SYSTEMS Pub Date : 2026-02-01 Epub Date: 2025-10-27 DOI: 10.1016/j.adhoc.2025.104062
Hamza Allaga , Mohamed Biniz , Abderrazak Farchane
Message Queuing Telemetry Transport (MQTT) is essential for resource-constrained Internet of Things (IoT) environments; however, its widespread adoption has introduced significant security vulnerabilities. Although machine learning (ML) offers a promising solution for anomaly detection, existing models are often hindered by unrealistic data, severe class imbalances, and high computational costs. To address these limitations, we present a comprehensive ML framework for MQTT anomaly detection benchmarked on MQTTEEB-D, a high-fidelity dataset from a physical IoT testbed. Our framework evaluates a diverse suite of algorithms, including tree ensembles and boosting methods, on both original imbalanced and balanced data. We assessed performance using standard metrics, imbalance-stable metrics such as the Matthews Correlation Coefficient (MCC), and a Performance–Efficiency Score (PES) to quantify the trade-off between predictive power and computational cost. Our results establish a new state-of-the-art, with the top models achieving over 98.8% accuracy and F1-score. These models also yielded dramatic efficiency gains, including a 43-fold reduction in training time and a 299-fold speedup in inference latency over previous benchmarks. Critically, we found that a model’s resilience to class imbalance is more vital for real-world deployment than its peak performance on artificially balanced data. Simpler tree-based models remained robust under imbalanced conditions, where more complex algorithms failed. These findings provide a new benchmark and reorient model selection towards efficient, reliable, and deployable IoT security systems.
消息队列遥测传输(MQTT)对于资源受限的物联网(IoT)环境至关重要;然而,它的广泛采用带来了重大的安全漏洞。尽管机器学习(ML)为异常检测提供了一个很有前途的解决方案,但现有模型经常受到不现实数据、严重的类不平衡和高计算成本的阻碍。为了解决这些限制,我们提出了一个全面的MQTT异常检测ML框架,该框架以MQTTEEB-D为基准,MQTTEEB-D是来自物理物联网测试平台的高保真数据集。我们的框架评估了一套不同的算法,包括树集成和增强方法,对原始不平衡和平衡数据。我们使用标准指标、不平衡稳定指标(如马修斯相关系数(MCC))和性能效率评分(PES)来评估性能,以量化预测能力和计算成本之间的权衡。我们的研究结果建立了一种新的技术水平,顶级模型的准确率超过98.8%,得分为f1。这些模型还产生了显著的效率提升,包括与以前的基准测试相比,训练时间减少了43倍,推理延迟加快了299倍。关键的是,我们发现模型对类不平衡的弹性对于实际部署来说比它在人为平衡数据上的峰值性能更为重要。在不平衡条件下,更简单的基于树的模型仍然是健壮的,而更复杂的算法则失败了。这些发现提供了一个新的基准,并将模型选择重新定位为高效、可靠和可部署的物联网安全系统。
{"title":"MQTTEEB-D: A high-fidelity benchmark for real-time MQTT anomaly detection using machine learning techniques","authors":"Hamza Allaga ,&nbsp;Mohamed Biniz ,&nbsp;Abderrazak Farchane","doi":"10.1016/j.adhoc.2025.104062","DOIUrl":"10.1016/j.adhoc.2025.104062","url":null,"abstract":"<div><div>Message Queuing Telemetry Transport (MQTT) is essential for resource-constrained Internet of Things (IoT) environments; however, its widespread adoption has introduced significant security vulnerabilities. Although machine learning (ML) offers a promising solution for anomaly detection, existing models are often hindered by unrealistic data, severe class imbalances, and high computational costs. To address these limitations, we present a comprehensive ML framework for MQTT anomaly detection benchmarked on MQTTEEB-D, a high-fidelity dataset from a physical IoT testbed. Our framework evaluates a diverse suite of algorithms, including tree ensembles and boosting methods, on both original imbalanced and balanced data. We assessed performance using standard metrics, imbalance-stable metrics such as the Matthews Correlation Coefficient (MCC), and a Performance–Efficiency Score (PES) to quantify the trade-off between predictive power and computational cost. Our results establish a new state-of-the-art, with the top models achieving over 98.8% accuracy and F1-score. These models also yielded dramatic efficiency gains, including a 43-fold reduction in training time and a 299-fold speedup in inference latency over previous benchmarks. Critically, we found that a model’s resilience to class imbalance is more vital for real-world deployment than its peak performance on artificially balanced data. Simpler tree-based models remained robust under imbalanced conditions, where more complex algorithms failed. These findings provide a new benchmark and reorient model selection towards efficient, reliable, and deployable IoT security systems.</div></div>","PeriodicalId":55555,"journal":{"name":"Ad Hoc Networks","volume":"181 ","pages":"Article 104062"},"PeriodicalIF":4.8,"publicationDate":"2026-02-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145419830","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
Collaborative deployment of Large AI Models on the edge: A microservice approach to heterogeneous training and quantized inference 边缘上大型人工智能模型的协作部署:异构训练和量化推理的微服务方法
IF 4.8 3区 计算机科学 Q1 COMPUTER SCIENCE, INFORMATION SYSTEMS Pub Date : 2026-02-01 Epub Date: 2025-11-03 DOI: 10.1016/j.adhoc.2025.104080
Seyed Hossein Ahmadpanah , Meghdad Mirabi , Sanaz Sobhanloo , Pania Afsharfarnia , Donya Fallah , Mobina Bayati
While Large AI Models (LAMs) offer transformative intelligence, their deployment in real-time Internet-of-Things (IoT) applications is constrained by the heterogeneity of edge networks and high cloud latency. To enable cooperative LAM deployment across diverse edge hardware, we propose a modular, hardware-aware framework. Our approach introduces two key innovations. First, a heterogeneity-aware training framework combines Quantization-Aware Training with parameter-efficient federated fine-tuning, reducing communication overhead by over 98% and allowing devices with different hardware (FP32 vs. INT8) to collaboratively refine a single model with near-native accuracy (84.9% vs. 85.5%). Second, a precision-aware inference architecture virtualizes LAM features, including Chain-of-Thought (CoT) steps and Mixture-of-Experts (MoE) layers, into multi-precision microservices. A dynamic orchestrator selects the optimal microservice for each task, balancing energy, latency, and accuracy. Compared to existing edge deployments, experiments show over 70% reduction in active memory usage and nearly 60% reduction in end-to-end inference latency. This framework provides a scalable, privacy-preserving solution for hardware-aware LAM intelligence on heterogeneous edge networks, effectively overcoming key deployment challenges.
虽然大型人工智能模型(lam)提供了变革性的智能,但它们在实时物联网(IoT)应用中的部署受到边缘网络异质性和高云延迟的限制。为了实现跨不同边缘硬件的协作式LAM部署,我们提出了一个模块化的硬件感知框架。我们的方法引入了两个关键的创新。首先,异构感知训练框架将量化感知训练与参数高效的联邦微调相结合,减少了98%以上的通信开销,并允许不同硬件设备(FP32 vs INT8)以接近原生精度(84.9% vs 85.5%)协同改进单个模型。其次,精度感知推理架构将LAM特征(包括思维链(CoT)步骤和专家混合(MoE)层)虚拟化为多精度微服务。动态编排器为每个任务选择最佳的微服务,平衡能量、延迟和准确性。与现有的边缘部署相比,实验表明,活动内存使用减少了70%以上,端到端推理延迟减少了近60%。该框架为异构边缘网络上的硬件感知LAM智能提供了可扩展的隐私保护解决方案,有效地克服了关键的部署挑战。
{"title":"Collaborative deployment of Large AI Models on the edge: A microservice approach to heterogeneous training and quantized inference","authors":"Seyed Hossein Ahmadpanah ,&nbsp;Meghdad Mirabi ,&nbsp;Sanaz Sobhanloo ,&nbsp;Pania Afsharfarnia ,&nbsp;Donya Fallah ,&nbsp;Mobina Bayati","doi":"10.1016/j.adhoc.2025.104080","DOIUrl":"10.1016/j.adhoc.2025.104080","url":null,"abstract":"<div><div>While Large AI Models (LAMs) offer transformative intelligence, their deployment in real-time Internet-of-Things (IoT) applications is constrained by the heterogeneity of edge networks and high cloud latency. To enable cooperative LAM deployment across diverse edge hardware, we propose a modular, hardware-aware framework. Our approach introduces two key innovations. First, a heterogeneity-aware training framework combines Quantization-Aware Training with parameter-efficient federated fine-tuning, reducing communication overhead by over 98% and allowing devices with different hardware (FP32 vs. INT8) to collaboratively refine a single model with near-native accuracy (84.9% vs. 85.5%). Second, a precision-aware inference architecture virtualizes LAM features, including Chain-of-Thought (CoT) steps and Mixture-of-Experts (MoE) layers, into multi-precision microservices. A dynamic orchestrator selects the optimal microservice for each task, balancing energy, latency, and accuracy. Compared to existing edge deployments, experiments show over 70% reduction in active memory usage and nearly 60% reduction in end-to-end inference latency. This framework provides a scalable, privacy-preserving solution for hardware-aware LAM intelligence on heterogeneous edge networks, effectively overcoming key deployment challenges.</div></div>","PeriodicalId":55555,"journal":{"name":"Ad Hoc Networks","volume":"181 ","pages":"Article 104080"},"PeriodicalIF":4.8,"publicationDate":"2026-02-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145468465","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
期刊
Ad Hoc Networks
全部 Acc. Chem. Res. ACS Applied Bio Materials ACS Appl. Electron. Mater. ACS Appl. Energy Mater. ACS Appl. Mater. Interfaces ACS Appl. Nano Mater. ACS Appl. Polym. Mater. ACS BIOMATER-SCI ENG ACS Catal. ACS Cent. Sci. ACS Chem. Biol. ACS Chemical Health & Safety ACS Chem. Neurosci. ACS Comb. Sci. ACS Earth Space Chem. ACS Energy Lett. ACS Infect. Dis. ACS Macro Lett. ACS Mater. Lett. ACS Med. Chem. Lett. ACS Nano ACS Omega ACS Photonics ACS Sens. ACS Sustainable Chem. Eng. ACS Synth. Biol. Anal. Chem. BIOCHEMISTRY-US Bioconjugate Chem. BIOMACROMOLECULES Chem. Res. Toxicol. Chem. Rev. Chem. Mater. CRYST GROWTH DES ENERG FUEL Environ. Sci. Technol. Environ. Sci. Technol. Lett. Eur. J. Inorg. Chem. IND ENG CHEM RES Inorg. Chem. J. Agric. Food. Chem. J. Chem. Eng. Data J. Chem. Educ. J. Chem. Inf. Model. J. Chem. Theory Comput. J. Med. Chem. J. Nat. Prod. J PROTEOME RES J. Am. Chem. Soc. LANGMUIR MACROMOLECULES Mol. Pharmaceutics Nano Lett. Org. Lett. ORG PROCESS RES DEV ORGANOMETALLICS J. Org. Chem. J. Phys. Chem. J. Phys. Chem. A J. Phys. Chem. B J. Phys. Chem. C J. Phys. Chem. Lett. Analyst Anal. Methods Biomater. Sci. Catal. Sci. Technol. Chem. Commun. Chem. Soc. Rev. CHEM EDUC RES PRACT CRYSTENGCOMM Dalton Trans. Energy Environ. Sci. ENVIRON SCI-NANO ENVIRON SCI-PROC IMP ENVIRON SCI-WAT RES Faraday Discuss. Food Funct. Green Chem. Inorg. Chem. Front. Integr. Biol. J. Anal. At. Spectrom. J. Mater. Chem. A J. Mater. Chem. B J. Mater. Chem. C Lab Chip Mater. Chem. Front. Mater. Horiz. MEDCHEMCOMM Metallomics Mol. Biosyst. Mol. Syst. Des. Eng. Nanoscale Nanoscale Horiz. Nat. Prod. Rep. New J. Chem. Org. Biomol. Chem. Org. Chem. Front. PHOTOCH PHOTOBIO SCI PCCP Polym. Chem.
×
引用
GB/T 7714-2015
复制
MLA
复制
APA
复制
导出至
BibTeX EndNote RefMan NoteFirst NoteExpress
×
0
微信
客服QQ
Book学术公众号 扫码关注我们
反馈
×
意见反馈
请填写您的意见或建议
请填写您的手机或邮箱
×
提示
您的信息不完整,为了账户安全,请先补充。
现在去补充
×
提示
您因"违规操作"
具体请查看互助需知
我知道了
×
提示
现在去查看 取消
×
提示
确定
Book学术官方微信
Book学术文献互助
Book学术文献互助群
群 号:604180095
Book学术
文献互助 智能选刊 最新文献 互助须知 联系我们:info@booksci.cn
Book学术提供免费学术资源搜索服务,方便国内外学者检索中英文文献。致力于提供最便捷和优质的服务体验。
Copyright © 2023 Book学术 All rights reserved.
ghs 京公网安备 11010802042870号 京ICP备2023020795号-1