Computer Networks最新文献

英文中文

IF 4.6 2区计算机科学 Q1 COMPUTER SCIENCE, HARDWARE & ARCHITECTURE

Computer Networks

Pub Date : 2026-01-01

引用次数: 0

IF 4.6 2区计算机科学 Q1 COMPUTER SCIENCE, HARDWARE & ARCHITECTURE

Computer Networks

Pub Date : 2026-01-01

引用次数: 0

IF 4.6 2区计算机科学 Q1 COMPUTER SCIENCE, HARDWARE & ARCHITECTURE

Computer Networks

Pub Date : 2026-01-01

引用次数: 0

IF 4.6 2区计算机科学 Q1 COMPUTER SCIENCE, HARDWARE & ARCHITECTURE

Computer Networks

Pub Date : 2026-01-01

引用次数: 0

Adaptive gradient sparsification with layer and stage-wise for accelerating distributed DNN training 基于层和阶段的自适应梯度稀疏化加速分布式DNN训练

IF 4.6 2区计算机科学 Q1 COMPUTER SCIENCE, HARDWARE & ARCHITECTURE

Computer Networks

Pub Date : 2025-12-31 DOI: 10.1016/j.comnet.2025.111983

Wai-Xi Liu , Jun Cai , Yue Yin , Zhen-Xin Zhang , Kong-Yang Chen , Jian-Tao Fu , Wen-Li Shang

The scalability of distributed Deep Neural Network (DNN) training is critically limited by its substantial communication latency, arising from the massive volume of gradient data and frequent synchronization rounds. While existing techniques like gradient sparsification and quantization aim to alleviate this bottleneck, they often compromise model accuracy or introduce significant computational costs. We observe that gradients of different layers of DNN exhibit varying sensitivities to model performance and drastically different distribution patterns at different training stages. Moreover, the computational overhead of gradient clustering itself is predicTable Motivated by these insights, we propose an Adaptive Gradient Sparsification with Layer and Stage-wise (AGSLS) method, based on a gradient clustering idea. We propose a novel efficiency-aware scheme that first identifies which gradient layers should be sparsified and which are better transmitted directly, ensuring that each sparsification operation contributes positively to training acceleration. Furthermore, AGSLS introduces a layer-wise adaptive gradient sensing scheme that tailors the number of clusters for each layer at different training stages, thereby minimizing communication traffic without sacrificing model accuracy. We evaluate AGSLS extensively across multiple widely-used datasets and models, spanning image classification (ResNet-18, VGGNet-16, and AlexNet on CIFAR-10/100) and natural language processing (BERT on SST-2). The results demonstrate that AGSLS significantly outperforms existing approaches, including Bulk Synchronous Parallel (BSP), STL-SGD, DGC, and RedSync, reducing training time by up to 86.67%, 56.6%, 52.1%, and 57.1%, respectively.

分布式深度神经网络（DNN）训练的可扩展性受到其巨大的通信延迟的严重限制，这是由大量的梯度数据和频繁的同步轮引起的。虽然现有的梯度稀疏化和量化等技术旨在缓解这一瓶颈，但它们往往会损害模型的准确性或引入大量的计算成本。我们观察到DNN不同层的梯度对模型性能表现出不同的敏感性，并且在不同的训练阶段表现出截然不同的分布模式。此外，梯度聚类本身的计算开销是可预测的。基于这些见解，我们提出了一种基于梯度聚类思想的自适应分层分层梯度稀疏化（AGSLS）方法。我们提出了一种新的效率感知方案，首先确定哪些梯度层应该稀疏化，哪些梯度层可以直接传输，确保每个稀疏化操作都对训练加速有积极的贡献。此外，AGSLS引入了一种分层自适应梯度传感方案，该方案在不同的训练阶段为每层定制簇的数量，从而在不牺牲模型精度的情况下最大限度地减少通信流量。我们在多个广泛使用的数据集和模型上对AGSLS进行了广泛的评估，包括图像分类（CIFAR-10/100上的ResNet-18、VGGNet-16和AlexNet）和自然语言处理（SST-2上的BERT）。结果表明，AGSLS显著优于现有的Bulk Synchronous Parallel （BSP）、STL-SGD、DGC和RedSync方法，分别减少了86.67%、56.6%、52.1%和57.1%的训练时间。

{"title":"Adaptive gradient sparsification with layer and stage-wise for accelerating distributed DNN training","authors":"Wai-Xi Liu , Jun Cai , Yue Yin , Zhen-Xin Zhang , Kong-Yang Chen , Jian-Tao Fu , Wen-Li Shang","doi":"10.1016/j.comnet.2025.111983","DOIUrl":"10.1016/j.comnet.2025.111983","url":null,"abstract":"<div><div>The scalability of distributed Deep Neural Network (DNN) training is critically limited by its substantial communication latency, arising from the massive volume of gradient data and frequent synchronization rounds. While existing techniques like gradient sparsification and quantization aim to alleviate this bottleneck, they often compromise model accuracy or introduce significant computational costs. We observe that gradients of different layers of DNN exhibit varying sensitivities to model performance and drastically different distribution patterns at different training stages. Moreover, the computational overhead of gradient clustering itself is predicTable Motivated by these insights, we propose an Adaptive Gradient Sparsification with Layer and Stage-wise (AGSLS) method, based on a gradient clustering idea. We propose a novel efficiency-aware scheme that first identifies which gradient layers should be sparsified and which are better transmitted directly, ensuring that each sparsification operation contributes positively to training acceleration. Furthermore, AGSLS introduces a layer-wise adaptive gradient sensing scheme that tailors the number of clusters for each layer at different training stages, thereby minimizing communication traffic without sacrificing model accuracy. We evaluate AGSLS extensively across multiple widely-used datasets and models, spanning image classification (ResNet-18, VGGNet-16, and AlexNet on CIFAR-10/100) and natural language processing (BERT on SST-2). The results demonstrate that AGSLS significantly outperforms existing approaches, including Bulk Synchronous Parallel (BSP), STL-SGD, DGC, and RedSync, reducing training time by up to 86.67%, 56.6%, 52.1%, and 57.1%, respectively.</div></div>","PeriodicalId":50637,"journal":{"name":"Computer Networks","volume":"276 ","pages":"Article 111983"},"PeriodicalIF":4.6,"publicationDate":"2025-12-31","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145927421","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"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

Exploring the future of telecommunications through a Techno-Economic lens 通过技术经济视角探索电信的未来

IF 4.6 2区计算机科学 Q1 COMPUTER SCIENCE, HARDWARE & ARCHITECTURE

Computer Networks

Pub Date : 2025-12-31 DOI: 10.1016/j.comnet.2025.111984

Edoardo Meraviglia , Mattia Magnaghi , Antonio Capone , Nicola Blefari Melazzi , Marta Valsecchi , Luca Dozio

The telecommunications industry is undergoing a major transformation driven by new technologies alongside economic challenges that are reshaping the entire ecosystem. The increasing digitalization of industries and sectors demands more advanced network infrastructures tailored to diverse market verticals and supporting businesses in their digital transformation. However, the ecosystem’s future remains uncertain, influenced by operators’ strategic choices, business model innovation, national and industrial policies, regulations, and cross-industry investments. This paper extends previous work by not only outlining possible evolutionary scenarios of the telecommunications ecosystem but also analyzing the reciprocal relationship between technological advances and these scenarios. Specifically, it investigates how emerging market, ecosystem, and economic conditions influence the evolution and application of technology, while technology itself drives changes in these domains. Focusing on the European ecosystem, with implications for global contexts, this integrated perspective aims to support market players and policymakers in navigating transformation and making informed decisions. This work responds to the Grand Challenge “Creating a vision of the future evolution of telecommunications” within the RESTART research program, the most important public R&D initiative ever launched in the Italian telecommunications sector, funded by the European Union / Italian Ministry of University and Research, with an investment of 116 million euros. Consistently with the elements outlined above, it focuses on the trajectory of the telecommunications ecosystem, providing a comprehensive and forward looking assessment of the key business, technology and regulatory drivers at both national and European level.

在新技术和经济挑战的推动下，电信行业正在经历一场重大变革，这些挑战正在重塑整个生态系统。行业和部门的日益数字化需要更先进的网络基础设施，为不同的垂直市场量身定制，并支持企业的数字化转型。然而，受运营商的战略选择、商业模式创新、国家和行业政策、法规以及跨行业投资的影响，油气生态系统的未来仍然不确定。本文不仅概述了电信生态系统可能的进化情景，而且分析了技术进步与这些情景之间的相互关系，从而扩展了以前的工作。具体来说，它研究了新兴市场、生态系统和经济条件如何影响技术的发展和应用，而技术本身推动了这些领域的变化。这一综合视角着眼于欧洲生态系统，并对全球环境产生影响，旨在支持市场参与者和政策制定者引导转型并做出明智的决策。这项工作响应了RESTART研究计划中的“创造电信未来发展愿景”大挑战，该计划是意大利电信部门有史以来最重要的公共研发计划，由欧盟/意大利大学和研究部资助，投资额为1.16亿欧元。与上述要素一致，本报告侧重于电信生态系统的发展轨迹，对国家和欧洲层面的关键业务、技术和监管驱动因素进行了全面和前瞻性的评估。

{"title":"Exploring the future of telecommunications through a Techno-Economic lens","authors":"Edoardo Meraviglia , Mattia Magnaghi , Antonio Capone , Nicola Blefari Melazzi , Marta Valsecchi , Luca Dozio","doi":"10.1016/j.comnet.2025.111984","DOIUrl":"10.1016/j.comnet.2025.111984","url":null,"abstract":"<div><div>The telecommunications industry is undergoing a major transformation driven by new technologies alongside economic challenges that are reshaping the entire ecosystem. The increasing digitalization of industries and sectors demands more advanced network infrastructures tailored to diverse market verticals and supporting businesses in their digital transformation. However, the ecosystem’s future remains uncertain, influenced by operators’ strategic choices, business model innovation, national and industrial policies, regulations, and cross-industry investments. This paper extends previous work by not only outlining possible evolutionary scenarios of the telecommunications ecosystem but also analyzing the reciprocal relationship between technological advances and these scenarios. Specifically, it investigates how emerging market, ecosystem, and economic conditions influence the evolution and application of technology, while technology itself drives changes in these domains. Focusing on the European ecosystem, with implications for global contexts, this integrated perspective aims to support market players and policymakers in navigating transformation and making informed decisions. This work responds to the Grand Challenge “Creating a vision of the future evolution of telecommunications” within the RESTART research program, the most important public R&D initiative ever launched in the Italian telecommunications sector, funded by the <span>European Union</span> / Italian Ministry of University and Research, with an investment of 116 million euros. Consistently with the elements outlined above, it focuses on the trajectory of the telecommunications ecosystem, providing a comprehensive and forward looking assessment of the key business, technology and regulatory drivers at both national and European level.</div></div>","PeriodicalId":50637,"journal":{"name":"Computer Networks","volume":"276 ","pages":"Article 111984"},"PeriodicalIF":4.6,"publicationDate":"2025-12-31","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145927422","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"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

CAN-BiGRUBERT: Unveiling automotive vehicle intruders by profiling and characterizing anomalies in controller area network CAN-BiGRUBERT：通过分析和表征控制器局域网中的异常来揭示汽车入侵者

IF 4.6 2区计算机科学 Q1 COMPUTER SCIENCE, HARDWARE & ARCHITECTURE

Computer Networks

Pub Date : 2025-12-30 DOI: 10.1016/j.comnet.2025.111963

Shaila Sharmin , Arash Habibi Lashkari , Hafizah Mansor , Andi Fitriah Abdul Kadir

In-vehicle Controller Area Networks (CAN) are vulnerable to various injection attacks that can compromise the safety of vehicle occupants and result in financial losses. While a substantial body of work on CAN intrusion detection exists, it lacks multiclass attack classification models. Current multiclass models do not encompass all attack types or account for the vehicle’s state, i.e., whether the car is stationary or in motion. This work addresses these limitations by proposing CAN-BiGRUBERT, a multiclass CAN intrusion detection model that jointly predicts the vehicle state and attack class from CAN traffic windows. CAN-BiGRUBERT employs Bidirectional Encoder Representations from Transformers (BERT) to capture spatial dependencies within individual CAN frames, and a Bidirectional Gated Recurrent Unit (BiGRU) network to capture temporal dependencies across multiple frames in a window. For training and evaluating CAN-BiGRUBERT, we comprehensively reviewed current CAN intrusion datasets to select the HCRL Attack & Defense dataset, which contains all injection attacks executed in both vehicle states. We implemented CAN-BiGRUBERT and compared its performance with other variants and state-of-the-art CAN attack classification models, based on individual CAN frames, arbitration identifier (AID) sequences, and windows of complete frames. Compared to the baseline models, the proposed model achieved higher accuracy and F1-score, indicating its superior ability to predict the vehicle state and attack class simultaneously. Specifically excelling in detecting replay attacks and discriminating between driving and stationary states, CAN-BiGRUBERT represents a promising enhanced, informative intrusion detection method for in-vehicle CAN.

车载控制器区域网络（CAN）容易受到各种注入攻击，这些攻击可能危及车辆乘员的安全并导致经济损失。虽然存在大量的CAN入侵检测工作，但缺乏多类攻击分类模型。目前的多类别模型并没有涵盖所有的攻击类型，也没有考虑到车辆的状态，即汽车是静止的还是运动的。这项工作通过提出CAN- bigrubert来解决这些限制，CAN- bigrubert是一个多类CAN入侵检测模型，可以从CAN流量窗口共同预测车辆状态和攻击类别。CAN- bigrubert使用来自变形器的双向编码器表示（BERT）来捕获单个CAN帧内的空间依赖性，并使用双向门控循环单元（BiGRU）网络来捕获窗口中多个帧之间的时间依赖性。为了训练和评估CAN- bigrubert，我们全面审查了当前的CAN入侵数据集，以选择HCRL攻击和防御数据集，其中包含在两种车辆状态下执行的所有注入攻击。我们实现了CAN- bigrubert，并将其性能与基于单个CAN帧、仲裁标识符（AID）序列和完整帧窗口的其他变体和最先进的CAN攻击分类模型进行了比较。与基线模型相比，该模型获得了更高的准确率和f1分，表明其具有更强的同时预测车辆状态和攻击类别的能力。特别是在检测重放攻击和区分驾驶和静止状态方面表现出色，CAN- bigrubert代表了一种有前途的增强的车载CAN信息入侵检测方法。

{"title":"CAN-BiGRUBERT: Unveiling automotive vehicle intruders by profiling and characterizing anomalies in controller area network","authors":"Shaila Sharmin , Arash Habibi Lashkari , Hafizah Mansor , Andi Fitriah Abdul Kadir","doi":"10.1016/j.comnet.2025.111963","DOIUrl":"10.1016/j.comnet.2025.111963","url":null,"abstract":"<div><div>In-vehicle Controller Area Networks (CAN) are vulnerable to various injection attacks that can compromise the safety of vehicle occupants and result in financial losses. While a substantial body of work on CAN intrusion detection exists, it lacks multiclass attack classification models. Current multiclass models do not encompass all attack types or account for the vehicle’s state, i.e., whether the car is stationary or in motion. This work addresses these limitations by proposing CAN-BiGRUBERT, a multiclass CAN intrusion detection model that jointly predicts the vehicle state and attack class from CAN traffic windows. CAN-BiGRUBERT employs Bidirectional Encoder Representations from Transformers (BERT) to capture spatial dependencies within individual CAN frames, and a Bidirectional Gated Recurrent Unit (BiGRU) network to capture temporal dependencies across multiple frames in a window. For training and evaluating CAN-BiGRUBERT, we comprehensively reviewed current CAN intrusion datasets to select the HCRL Attack & Defense dataset, which contains all injection attacks executed in both vehicle states. We implemented CAN-BiGRUBERT and compared its performance with other variants and state-of-the-art CAN attack classification models, based on individual CAN frames, arbitration identifier (AID) sequences, and windows of complete frames. Compared to the baseline models, the proposed model achieved higher accuracy and F1-score, indicating its superior ability to predict the vehicle state and attack class simultaneously. Specifically excelling in detecting replay attacks and discriminating between driving and stationary states, CAN-BiGRUBERT represents a promising enhanced, informative intrusion detection method for in-vehicle CAN.</div></div>","PeriodicalId":50637,"journal":{"name":"Computer Networks","volume":"276 ","pages":"Article 111963"},"PeriodicalIF":4.6,"publicationDate":"2025-12-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145886496","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"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

Hawkeyes: An intelligent honeypot allocation strategy for cyber deception using reinforcement learning 鹰眼：基于强化学习的网络欺骗智能蜜罐分配策略

IF 4.6 2区计算机科学 Q1 COMPUTER SCIENCE, HARDWARE & ARCHITECTURE

Computer Networks

Pub Date : 2025-12-30 DOI: 10.1016/j.comnet.2025.111982

Hien Do Hoang , Trong-Nghia To , Ngo Duc Hoang Son , Khoa Ngo-Khanh , Nguyen Tan Cam , Van-Hau Pham

Honeypot allocation has emerged as a pivotal strategy in cyber deception. However, existing approaches often face scalability issues, limited coordination, and inadequate consideration of intrusion stages, which constrain their effectiveness in complex attack environments. To address these challenges, this study introduces Hawkeyes, a hierarchical multi-agent reinforcement learning (HMARL) framework for adaptive honeypot allocation. The framework combines a high-level policy, which selects strategies based on intrusion progression and asset criticality, with low-level agents that determine precise placements within grouped regions. Scalability is ensured through action-space reduction via grouping and multi-agent decomposition. Network states are modeled as enriched graphs that capture compromised nodes, topology, and host vulnerabilities, enabling stage-aware and context-rich decision making. Experimental results show that Hawkeyes outperforms baseline RL methods, achieving 15–18% higher trapping efficiency and superior stage-aware defense (48.4% vs. 30.6% efficiency at False Negative Rate of 0.05). It also maintains stable performance under topology and vulnerability variations, while ensuring real-time deployment with sub-5s latency.

蜜罐分配已成为网络欺骗的关键策略。然而，现有的方法往往面临着可扩展性问题、有限的协调和对入侵阶段的考虑不足，这限制了它们在复杂攻击环境中的有效性。为了解决这些挑战，本研究引入了Hawkeyes，一种用于自适应蜜罐分配的分层多智能体强化学习（HMARL）框架。该框架结合了基于入侵进程和资产重要性选择策略的高级策略和确定分组区域内精确位置的低级代理。通过分组和多智能体分解的动作空间缩减来保证可扩展性。网络状态被建模为捕获受损节点、拓扑和主机漏洞的丰富图形，从而支持阶段感知和上下文丰富的决策制定。实验结果表明，鹰眼的捕获效率比基线RL方法高15-18%，在假阴性率为0.05的情况下，捕获效率为48.4%，防御效率为30.6%。它还在拓扑和漏洞变化下保持稳定的性能，同时确保以低于5s的延迟进行实时部署。

{"title":"Hawkeyes: An intelligent honeypot allocation strategy for cyber deception using reinforcement learning","authors":"Hien Do Hoang , Trong-Nghia To , Ngo Duc Hoang Son , Khoa Ngo-Khanh , Nguyen Tan Cam , Van-Hau Pham","doi":"10.1016/j.comnet.2025.111982","DOIUrl":"10.1016/j.comnet.2025.111982","url":null,"abstract":"<div><div>Honeypot allocation has emerged as a pivotal strategy in cyber deception. However, existing approaches often face scalability issues, limited coordination, and inadequate consideration of intrusion stages, which constrain their effectiveness in complex attack environments. To address these challenges, this study introduces Hawkeyes, a hierarchical multi-agent reinforcement learning (HMARL) framework for adaptive honeypot allocation. The framework combines a high-level policy, which selects strategies based on intrusion progression and asset criticality, with low-level agents that determine precise placements within grouped regions. Scalability is ensured through action-space reduction via grouping and multi-agent decomposition. Network states are modeled as enriched graphs that capture compromised nodes, topology, and host vulnerabilities, enabling stage-aware and context-rich decision making. Experimental results show that Hawkeyes outperforms baseline RL methods, achieving 15–18% higher trapping efficiency and superior stage-aware defense (48.4% vs. 30.6% efficiency at False Negative Rate of 0.05). It also maintains stable performance under topology and vulnerability variations, while ensuring real-time deployment with sub-5s latency.</div></div>","PeriodicalId":50637,"journal":{"name":"Computer Networks","volume":"276 ","pages":"Article 111982"},"PeriodicalIF":4.6,"publicationDate":"2025-12-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145927473","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"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

Vertex-independent spanning trees in data center network BCDC 数据中心网络BCDC中与顶点无关的生成树

IF 4.6 2区计算机科学 Q1 COMPUTER SCIENCE, HARDWARE & ARCHITECTURE

Computer Networks

Pub Date : 2025-12-30 DOI: 10.1016/j.comnet.2025.111981

Jiakang Ma , Baolei Cheng , Yan Wang , Jianxi Fan , Junkai Zhu

The performance of data center networks largely determines cloud computing efficiency. BCDC is a high-performance data center network whose logical graph is exactly the line graph of the n-dimensional crossed cube (CQ_n). However, there are few studies on its vertex-independent spanning trees (VISTs). Until now, constructing VISTs rooted at an arbitrary vertex in BCDC remains an open question. In this paper, an algorithm is proposed to construct the VISTs in BCDC. Firstly, a parallel algorithm is adopted to construct

n - 1

trees in CQ_n. Then, we transform these trees into

2 n - 2

mutually independent trees in the BCDC. Subsequently, by hanging vertices on these trees,

2 n - 2

VISTs rooted at an arbitrary vertex in BCDC are obtained. Finally, we used Python’s Matplotlib and NumPy packages for simulation and obtained results showing that the discrepancy between the average path length and the network diameter remains within 0.5, and the communication success rate stays above 60% even under a 30% vertex failure rate, which verifies the high efficiency and strong security of the network in data transmission.

数据中心网络的性能在很大程度上决定了云计算的效率。BCDC是一种高性能数据中心网络，其逻辑图正是n维交叉立方体（CQn）的线形图。然而，对其不依赖于顶点的生成树（vist）的研究却很少。到目前为止，构造基于BCDC任意顶点的vist仍然是一个悬而未决的问题。本文提出了一种构造BCDC中vist的算法。首先，采用并行算法在CQn中构造n−1棵树。然后，我们将这些树在BCDC中转化为2n−2个相互独立的树。然后，通过在这些树上挂起顶点，得到2n−2根于BCDC中任意顶点的vist。最后，我们使用Python的Matplotlib和NumPy包进行仿真，得到的结果表明，平均路径长度与网络直径的差异保持在0.5以内，即使在30%的顶点失败率下，通信成功率也保持在60%以上，验证了网络在数据传输方面的高效率和强安全性。

{"title":"Vertex-independent spanning trees in data center network BCDC","authors":"Jiakang Ma , Baolei Cheng , Yan Wang , Jianxi Fan , Junkai Zhu","doi":"10.1016/j.comnet.2025.111981","DOIUrl":"10.1016/j.comnet.2025.111981","url":null,"abstract":"<div><div>The performance of data center networks largely determines cloud computing efficiency. BCDC is a high-performance data center network whose logical graph is exactly the line graph of the <em>n</em>-dimensional crossed cube (<em>CQ<sub>n</sub></em>). However, there are few studies on its vertex-independent spanning trees (VISTs). Until now, constructing VISTs rooted at an arbitrary vertex in BCDC remains an open question. In this paper, an algorithm is proposed to construct the VISTs in BCDC. Firstly, a parallel algorithm is adopted to construct <span><math><mrow><mi>n</mi><mo>−</mo><mn>1</mn></mrow></math></span> trees in <em>CQ<sub>n</sub></em>. Then, we transform these trees into <span><math><mrow><mn>2</mn><mi>n</mi><mo>−</mo><mn>2</mn></mrow></math></span> mutually independent trees in the BCDC. Subsequently, by hanging vertices on these trees, <span><math><mrow><mn>2</mn><mi>n</mi><mo>−</mo><mn>2</mn></mrow></math></span> VISTs rooted at an arbitrary vertex in BCDC are obtained. Finally, we used Python’s Matplotlib and NumPy packages for simulation and obtained results showing that the discrepancy between the average path length and the network diameter remains within 0.5, and the communication success rate stays above 60% even under a 30% vertex failure rate, which verifies the high efficiency and strong security of the network in data transmission.</div></div>","PeriodicalId":50637,"journal":{"name":"Computer Networks","volume":"276 ","pages":"Article 111981"},"PeriodicalIF":4.6,"publicationDate":"2025-12-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145927418","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"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

Closed-Form Analytics of Multicell Massive MIMO System Using M-MMSE and TPE Techniques in Correlated Environment 相关环境下基于M-MMSE和TPE技术的多小区大规模MIMO系统封闭分析

IF 4.6 2区计算机科学 Q1 COMPUTER SCIENCE, HARDWARE & ARCHITECTURE

Computer Networks

Pub Date : 2025-12-29 DOI: 10.1016/j.comnet.2025.111965

Harleen Kaur, Ankush Kansal

This work computes the average ergodic user rate for the multicell massive Multiple-Input Multiple-Output (mMIMO) system based on Multicell Minimum Mean Squared Error (M-MMSE) and Truncated Polynomial Expansion (TPE) techniques. By applying Random Matrix Theory (RMT) and large system analysis, the deterministic expression for the system's Signal-to-Interference plus Noise Ratio (SINR) with the M-MMSE scheme in uplink and downlink mode is computed, leading to the system's average user rate calculation. The M-MMSE scheme involves gram matrix inversion, increasing the system's lag and complexity. Therefore, the problem is solved by approximating the inverse of the matrix using TPE, which involves simple operations that can parallelize. Also, the complexity of the TPE technique depends only on the TPE order rather than the system's dimensions. Based on the RMT theory, the deterministic equivalents required for SINRs of the TPE scheme in uplink and downlink modes are derived. These deterministic equivalents for TPE SINRs are optimized to compute the average user rate for the system, matching the M-MMSE technique performance at a lower TPE order. In section 6, the system’s average user rate is validated by varying it with different parameters. The comparison between the M-MMSE and the TPE scheme shows that the TPE scheme achieves the required performance at J=3 TPE order. The theoretical results show the accuracy of derived deterministic equivalents.

本文基于多单元最小均方误差（M-MMSE）和截断多项式展开（TPE）技术计算了多单元大规模多输入多输出（mMIMO）系统的平均遍历用户速率。应用随机矩阵理论（RMT）和大系统分析，计算了M-MMSE方案下上行和下行模式下系统信噪比（SINR）的确定性表达式，从而计算出系统的平均用户速率。M-MMSE方案涉及克矩阵反演，增加了系统的滞后和复杂性。因此，这个问题是通过使用TPE近似矩阵的逆来解决的，这涉及到可以并行化的简单操作。此外，TPE技术的复杂性仅取决于TPE的顺序，而不是系统的维度。基于RMT理论，推导了上行和下行模式下TPE方案sinr所需的确定性等效。TPE sinr的这些确定性等效被优化为计算系统的平均用户速率，在较低的TPE阶上匹配M-MMSE技术性能。在第6节中，通过使用不同的参数改变系统的平均用户率来验证系统的平均用户率。M-MMSE与TPE方案的比较表明，TPE方案在J=3 TPE阶时达到了要求的性能。理论结果表明了所导出的确定性等效的准确性。

{"title":"Closed-Form Analytics of Multicell Massive MIMO System Using M-MMSE and TPE Techniques in Correlated Environment","authors":"Harleen Kaur, Ankush Kansal","doi":"10.1016/j.comnet.2025.111965","DOIUrl":"10.1016/j.comnet.2025.111965","url":null,"abstract":"<div><div>This work computes the average ergodic user rate for the multicell massive Multiple-Input Multiple-Output (mMIMO) system based on Multicell Minimum Mean Squared Error (M-MMSE) and Truncated Polynomial Expansion (TPE) techniques. By applying Random Matrix Theory (RMT) and large system analysis, the deterministic expression for the system's Signal-to-Interference plus Noise Ratio (SINR) with the M-MMSE scheme in uplink and downlink mode is computed, leading to the system's average user rate calculation. The M-MMSE scheme involves gram matrix inversion, increasing the system's lag and complexity. Therefore, the problem is solved by approximating the inverse of the matrix using TPE, which involves simple operations that can parallelize. Also, the complexity of the TPE technique depends only on the TPE order rather than the system's dimensions. Based on the RMT theory, the deterministic equivalents required for SINRs of the TPE scheme in uplink and downlink modes are derived. These deterministic equivalents for TPE SINRs are optimized to compute the average user rate for the system, matching the M-MMSE technique performance at a lower TPE order. In section 6, the system’s average user rate is validated by varying it with different parameters. The comparison between the M-MMSE and the TPE scheme shows that the TPE scheme achieves the required performance at J=3 TPE order. The theoretical results show the accuracy of derived deterministic equivalents.</div></div>","PeriodicalId":50637,"journal":{"name":"Computer Networks","volume":"277 ","pages":"Article 111965"},"PeriodicalIF":4.6,"publicationDate":"2025-12-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146090323","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"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

首页上一页

下一页尾页

类型

全部化学•材料生命科学医学物理工程技术环境•农林材料科学地球科学法学管理学化学环境科学与生态学计算机科学教育学经济学农林科学人文科学生物学数学物理与天体物理心理学综合性期刊其他工业工程理学历史学农学文学信息工程

数据库

全部 ACS Publications Elsevier ieeexplore Springer The Royal Society of Chemistry Wiley

期刊

Computer 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.

﹀