Computer Networks最新文献

{"title":"MP-CreditINT: Enhancing multi-path RDMA transport with credit-based congestion control and in-band network telemetry","authors":"Yi Pan,&nbsp;Jiali You","doi":"10.1016/j.comnet.2025.111665","DOIUrl":"10.1016/j.comnet.2025.111665","url":null,"abstract":"<div><div>Remote Direct Memory Access (RDMA) has played a critical role in recent Large-Language-Model (LLM) training workloads by enabling low-latency, high-throughput communication across GPUs. To further improve network efficiency, multi-path RDMA transport has received increasing attention. Given the limited on-chip resources of RDMA NICs, MP-RDMA stands out as a state-of-the-art multi-path transport by adopting memory-efficient mechanisms for congestion and out-of-order control. However, MP-RDMA relies on ECN, which provides only coarse-grained binary congestion signals, resulting in limited congestion control capabilities. Our experimental observations reveal intrinsic limitations of MP-RDMA, such as slow bandwidth probing, large oscillations in queue buildup, and transient congestion, etc. These limitations reduce the network efficiency, especially under All-To-All communication patterns that are becoming increasingly dominant with the evolution of Mixture-of-Experts (MoE) models and Expert Parallelism.</div><div>To address these limitations and retain the memory-efficient property of MP-RDMA, we propose the MP-CreditINT. This approach re-architects MP-RDMA using credit-based congestion control and in-band network telemetry (INT). We systematically enumerate and address the architectural and algorithmic challenges arising from this transformation, including explicit path control and path symmetry, INT-based data-clocking credit window control, and robustness against feedback loop breakage. Then, we evaluate MP-CreditINT using micro-benchmarks, heavy incast traffic, and LLM training workloads. Simulation results demonstrate that when compared to MP-RDMA, MP-CreditINT achieves 6–38× faster ramp-up speed and 8–16× faster fairness convergence, while maintaining near-zero out-of-order degree. In heavy incast scenarios, it achieves superior fairness and near-zero queue buildup, whereas MP-RDMA exhibits exponential queue growth with increasing incast scale. Finally, under two representative LLM training workloads, All-Reduce and All-To-All, MP-CreditINT reduces completion time by 5%–8% and 7%–13% respectively when compared to MP-RDMA, demonstrating its benefits in LLM training.</div></div>","PeriodicalId":50637,"journal":{"name":"Computer Networks","volume":"272 ","pages":"Article 111665"},"PeriodicalIF":4.6,"publicationDate":"2025-09-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145158550","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}

{"title":"Network traffic anomaly detection method based on stacked fusion time features","authors":"Tianhao Hou ,&nbsp;Zheng Zhang ,&nbsp;Qiuling Wu ,&nbsp;Yan Yan ,&nbsp;Hao Li","doi":"10.1016/j.comnet.2025.111729","DOIUrl":"10.1016/j.comnet.2025.111729","url":null,"abstract":"<div><div>The rapid growth of Network Traffic (NT) necessitates reliable anomaly detection to mitigate security threats and ensure system stability. An accurate NT prediction strategy is key in this process. However, current network traffic anomaly detection (NTAD) methods are limited in accuracy due to their inability to adequately balance the modeling of short-term and long-term temporal dependencies in network traffic. This paper proposes an NTAD method based on a Stacked Fusion Time Feature (SFTF) framework to overcome this limitation. Specifically, a stacked time feature encoder is first constructed to capture time-series patterns across multiple resolutions, generating hierarchical feature sequences. These sequences are then fed into a multi-timescale feature fusion module based on a temporal convolutional network to integrate local and global temporal features. In addition, an interquartile range-based detection mechanism is established to identify anomalies from the prediction results. Experiments are conducted on two representative datasets, Yahoo S5 and SMD. On Yahoo S5, SFTF achieves an average AUC of 0.9647 and an <span><math><msub><mi>F</mi><mn>1</mn></msub></math></span> score of 0.9750; on SMD, SFTF attains an <span><math><msub><mi>F</mi><mn>1</mn></msub></math></span> score of 0.9713, with precision and recall of 0.9803 and 0.9622, respectively. These results demonstrate that SFTF can effectively identify abnormal states and accurately locate network anomalies across datasets with different temporal characteristics. The proposed method offers a robust and accurate solution for NTAD in large-scale, low-latency environments, with promising applications in bandwidth optimization and cybersecurity.</div></div>","PeriodicalId":50637,"journal":{"name":"Computer Networks","volume":"272 ","pages":"Article 111729"},"PeriodicalIF":4.6,"publicationDate":"2025-09-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145158499","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}

{"title":"ARPDR++: Exploiting local-global temporal modeling for smartphone-based indoor pedestrian localization","authors":"Xiaoqiang Teng ,&nbsp;Shibiao Xu ,&nbsp;Deke Guo ,&nbsp;Yulan Guo ,&nbsp;Pengfei Xu ,&nbsp;Runbo Hu ,&nbsp;Hua Chai","doi":"10.1016/j.comnet.2025.111710","DOIUrl":"10.1016/j.comnet.2025.111710","url":null,"abstract":"<div><div>The increasing prevalence of mobile computing has made Pedestrian Dead Reckoning (PDR) one of the most promising and attractive indoor localization techniques for ubiquitous applications. However, existing PDR approaches are either sensitive to various users or suffer from accumulated errors that cause position drifts. To address these issues, this paper proposes ARPDR++, an accurate and robust PDR approach that improves the accuracy and robustness of indoor localization methods. ARPDR++ introduces a novel step counting algorithm based on motion models that deeply exploits inertial sensor data. We combine step counting with adaptive thresholding to personalize the PDR system for different users. Furthermore, we propose a novel stride-heading model with a deep neural network to predict stride lengths and walking orientations, which significantly reduces displacement errors. Experimental results on public datasets demonstrate that ARPDR++ outperforms the state-of-the-art PDR methods.</div></div>","PeriodicalId":50637,"journal":{"name":"Computer Networks","volume":"272 ","pages":"Article 111710"},"PeriodicalIF":4.6,"publicationDate":"2025-09-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145220816","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}

{"title":"Evaluating MPQUIC schedulers in dynamic wireless networks with 2D and 3D mobility","authors":"Minh Hai Vu ,&nbsp;Thanh Trung Nguyen ,&nbsp;Thi Ha Ly Dinh ,&nbsp;Thanh-Hung Nguyen ,&nbsp;Phi Le Nguyen ,&nbsp;Kien Nguyen ,&nbsp;Hiroo Sekiya","doi":"10.1016/j.comnet.2025.111689","DOIUrl":"10.1016/j.comnet.2025.111689","url":null,"abstract":"<div><div>This study evaluates the performance of Multipath QUIC (MPQUIC) schedulers, which allow mobile devices to use multiple wireless networks for better throughput and reliability. Previous evaluations of MPQUIC schedulers are mostly limited to simple, two-dimensional (2D) scenarios, which do not capture the complexities of three-dimensional (3D) environments involving mobile or aerial devices. To address this, we implemented three 3D mobility models—Random Waypoint 3D, Reference Point Group Mobility 3D, and Gauss–Markov 3D—adapted from existing 2D models. We then assessed three non-learning MPQUIC schedulers (i.e., minRTT, BLEST, and ECF) and two learning-based schedulers (i.e., Peekaboo and Q-ReLeS) under varied conditions. Our results indicate that movement patterns, particularly random mobility, significantly affect scheduler performance. In stable network conditions, learning-based schedulers like Q-ReLeS outperform non-learning ones in download time and packet loss, but as conditions worsen, their advantages decrease, suggesting a need for further optimization in dynamic environments.</div></div>","PeriodicalId":50637,"journal":{"name":"Computer Networks","volume":"272 ","pages":"Article 111689"},"PeriodicalIF":4.6,"publicationDate":"2025-09-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145158548","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}

{"title":"Secrecy performance optimization for UAV-based cognitive relay NOMA system with friendly jamming","authors":"Thanh Trung Nguyen ,&nbsp;Tran Manh Hoang ,&nbsp;Le The Dung ,&nbsp;Phuong T. Tran","doi":"10.1016/j.comnet.2025.111599","DOIUrl":"10.1016/j.comnet.2025.111599","url":null,"abstract":"<div><div>This paper investigates the secrecy performance of an unmanned aerial vehicle (UAV)-based cognitive relay non-orthogonal multiple access system and friendly jamming. Specifically, by utilizing the radio spectrum licensed to primary users, a UAV serves as an aerial relay to facilitate signal transmission from a secondary ground source to secondary ground users. To prevent the relayed signals from being intercepted by secondary ground eavesdroppers, a second UAV is deployed to transmit friendly jamming signals. The average minimum secrecy rate (AMSR) of the secondary users is maximized through the joint optimization of UAVs’ flight trajectories and the transmit powers of the system nodes, while explicitly accounting for uncertainties in the eavesdroppers’ locations, the adverse effects of friendly jamming on both secondary and primary users, and the imperfections in successive interference cancellation (SIC) at both the UAV and secondary users. A block coordinate descent (BCD) technique is employed to tackle the resulting non-convex AMSR maximization problem and obtain an efficient solution. Simulation results at the 70-<span><math><mrow><mi>t</mi><mi>h</mi></mrow></math></span> time slot demonstrate that the proposed system achieves approximately 83.3% higher AMSR compared to a system without friendly jamming, around 1000% improvement over a non-optimized, non-jamming system, and also outperforms other benchmark schemes.</div></div>","PeriodicalId":50637,"journal":{"name":"Computer Networks","volume":"272 ","pages":"Article 111599"},"PeriodicalIF":4.6,"publicationDate":"2025-09-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145158547","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}

{"title":"Group-based windows scheduling method for non-deterministic periodic flows in time-sensitive networks","authors":"Zhe Zhao ,&nbsp;Yongjun Li ,&nbsp;Xiang Wang ,&nbsp;Qin Tian ,&nbsp;Ke Wang ,&nbsp;Kai Zhang ,&nbsp;Weitao Pan","doi":"10.1016/j.comnet.2025.111719","DOIUrl":"10.1016/j.comnet.2025.111719","url":null,"abstract":"<div><div>Time-triggered (TT) flows are usually periodic in time-sensitive networks. However, nondeterministic end systems can generate TT flow frames with significant jitter (i.e., jittery TT flows). Jitter can cause frames to miss the TT windows scheduled for the current period, resulting in excessive access delays, which in turn affect the end-to-end deterministic transmission of the TT flows. In our previous study, we proposed the use of a dynamic multiwindow approach to achieve deterministic access to jittery TT flows; however, its window schedule computation is too slow, and this method is only suitable for small networks with a few TT flows. We therefore propose a group-based, fast scheduling method for accessing and transmitting the windows of jittery TT flows based on multiple windows. A combination of heuristic algorithms and solvers, including the establishment of TT window groups, division of the solution region, and integrated parallel and serial incremental coarse- and fine-grained computations, significantly improves the efficiency of TT window scheduling. For coarse-grained scheduling, by establishing large window clusters and central alignment, the complexity of scheduling is considerably reduced while keeping success rates high. Furthermore, the integer linear programming constraints and objective functions for this method are provided. Compared with the conventional dynamic multiwindow approach, the proposed approach reduces the scheduling time for TT windows by two orders of magnitude for a small star network with a small number of jittery TT flows. Moreover, the reduction in scheduling time becomes more pronounced as the network topology complexity and number of jittery TT flows increase. Finally, the scheduling time performance of the proposed method is verified in commonly used star, tree, and bus networks. Evaluations demonstrate that the access and transmission windows for 500 jittery TT flows can be scheduled in these networks, enabling deterministic access and significantly improving scheduling efficiency.</div></div>","PeriodicalId":50637,"journal":{"name":"Computer Networks","volume":"272 ","pages":"Article 111719"},"PeriodicalIF":4.6,"publicationDate":"2025-09-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145158525","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}

{"title":"Network slicing through buffer aware multi-user transmissions","authors":"Ferdinando Marrone,&nbsp;Pasquale Imputato,&nbsp;Stefano Avallone","doi":"10.1016/j.comnet.2025.111717","DOIUrl":"10.1016/j.comnet.2025.111717","url":null,"abstract":"<div><div>Wireless networks are rapidly expanding into diverse industry sectors, each with unique requirements. To address this, network slicing has emerged as a key paradigm, enabling the partitioning of a shared physical network into multiple logical slices, each tailored to specific performance needs. While extensive research has explored network slicing in cellular networks, its application in Wi-Fi, particularly within IEEE 802.11ax, remains underexplored. IEEE 802.11ax introduces Orthogonal Frequency Division Multiple Access (OFDMA), which facilitates multi-user transmissions, presenting new opportunities for network slicing.</div><div>This paper proposes a novel approach for proportional resource allocation in Wi-Fi networks, extending OFDMA scheduling to dynamically allocate radio spectrum among different slices. Additionally, our method supports the adaptive reuse of unused resources, optimizing spectrum utilization. Simulation results using the ns-3 network simulator demonstrate that the proposed approach effectively manages OFDMA scheduling, ensuring that slices meet their performance targets closely for throughput and latency more than the existing state-of-the-art solutions.</div></div>","PeriodicalId":50637,"journal":{"name":"Computer Networks","volume":"272 ","pages":"Article 111717"},"PeriodicalIF":4.6,"publicationDate":"2025-09-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145118202","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}

{"title":"Explainability analysis based on attribution features for optimizing automatic modulation classification","authors":"Bo Xu,&nbsp;Shuo Wang,&nbsp;Uzair Aslam Bhatti,&nbsp;Xiaoyi Zhang,&nbsp;Hao Tang","doi":"10.1016/j.comnet.2025.111696","DOIUrl":"10.1016/j.comnet.2025.111696","url":null,"abstract":"<div><div>With the explosive growth of Internet of Things devices, wireless communication systems face significant challenges in achieving wide-area coverage and service continuity. In highly dynamic and heterogeneous networks, modulation schemes exhibit pronounced time-variability and complexity, and traditional automatic modulation classification methods are severely limited under complex channel conditions and low signal-to-noise ratios. Deep learning can significantly improve recognition accuracy. However, its black-box nature and lack of interpretability hinder reliable model optimization and deployment in safety-critical scenarios. To address this, we propose an explainability-driven optimization framework that integrates feature attribution analysis and attention mechanisms to enhance model reliability and interpretability. Four mainstream interpretability methods (IG, DL, LIME, and SHAP) are applied to amplitude-phase and in-phase-quadrature feature domains, and the robustness and effectiveness of attribution features are evaluated via sliding window and feature deletion experiments, as well as misclassification case studies. Based on the selected effective features, a feature adjustment module and attention mechanism are introduced to guide the model’s focus on key features. An Explainability Metric for Modulation is further constructed to quantitatively assess the consistency between attribution results and physical signal characteristics via constellation-domain alignment analysis. Experimental results demonstrate that the framework improves interpretability and reliability without modifying the model architecture or introducing additional information, with recognition accuracy increasing by approximately 10 % for CNN and 6 % for LSTM, and the optimized LSTM achieving over 92 %, providing a practical and effective solution for automatic modulation classification in complex scenarios.</div></div>","PeriodicalId":50637,"journal":{"name":"Computer Networks","volume":"272 ","pages":"Article 111696"},"PeriodicalIF":4.6,"publicationDate":"2025-09-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145106436","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}

{"title":"From amplifiers to OTN boards: Multi-layer optimization for low-cost optical metro networks","authors":"Aryanaz Attarpour ,&nbsp;Sanaz Ghane ,&nbsp;Memedhe Ibrahimi ,&nbsp;Francesco Musumeci ,&nbsp;Andrea Castoldi ,&nbsp;Andrea Bovio ,&nbsp;Massimo Tornatore","doi":"10.1016/j.comnet.2025.111726","DOIUrl":"10.1016/j.comnet.2025.111726","url":null,"abstract":"<div><div>Optical metro networks interconnect access networks to core networks and must support traffic ranging from aggregation of low-rate end-user requests to high-rate inter-datacenter transfers. To effectively support traffic volumes consisting of heterogeneous flows at extremely different bit-rate, optical metro networks must jointly support coherent (100/200Gbps) and non-coherent (10Gbps) transmission technologies. When deploying these networks, network operators prioritize seeking solutions that consider both scalability and equipment cost minimization. In metro optical networks, different technologies can enable cost savings: at Optical Transport Network (OTN) layer, <em>traffic grooming</em> can be used to reduce equipment cost, while, at Wavelength Division Multiplexing (WDM) layer, <em>filterless optical switching nodes</em>, based on purely passive components, can be used to avoid expensive Wavelength Selective Switches deployment (WSS), and <em>optimized Optical Amplifiers (OA) placement</em> can decrease significantly required amplifiers cost. Joint deployment of these technologies can facilitate significant cost savings, but requires coordination in form of multi-layer optimization, across OTN and WDM network layers to minimize overall equipment cost (from amplifiers at WDM layer, to OTN boards at OTN layer). In this paper, we propose a novel single-step Genetic Algorithm (GA) to jointly optimize OTN-layer equipment cost (OTN boards) and WDM-layer equipment (mainly OAs) cost. We propose two sequential GA approaches, named two-step and three-step. Numerical results, obtained using real network topologies and traffic matrices provided by our industrial collaborators, show that our proposed GA-based approaches can save costs up to 58 % compared to real-world baseline solutions, and that single-step approach outperforms two- and three-step cases up to 10 %.</div></div>","PeriodicalId":50637,"journal":{"name":"Computer Networks","volume":"272 ","pages":"Article 111726"},"PeriodicalIF":4.6,"publicationDate":"2025-09-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145158549","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}

{"title":"Energy efficiency optimization of aerial intelligent reflecting surface-assisted communications based on multi-agent deep reinforcement learning","authors":"Suyue Li,&nbsp;GuangQian Li,&nbsp;YunGuang Xi,&nbsp;Anhong Wang","doi":"10.1016/j.comnet.2025.111722","DOIUrl":"10.1016/j.comnet.2025.111722","url":null,"abstract":"<div><div>Reconfigurable Intelligent Surface (RIS) technology has emerged as a promising solution, garnering extensive attention in millimeter-wave (mmWave) communication systems. This paper explores the application of deploying aerial reconfigurable intelligent surfaces (ARISs) on multiple unmanned aerial vehicles (UAVs) to provide services to ground users in complex environments. However, existing studies seldom address the orientation design of RISs. In fact, the orientation design of multiple ARISs facilitates the establishment of collaborative communication links for users and enhances communication coverage. Therefore, this paper proposes a joint optimization problem for the trajectories, orientations, and phase shifts of ARISs in multi-ARIS-assisted communication systems, aiming to maximize the system’s energy efficiency. To address this issue, a multi-agent deep reinforcement learning (MADRL) approach, namely multi-agent proximal policy optimization (MAPPO), is employed. Simulation results demonstrate that the proposed scheme enhances system energy efficiency by approximately 22 % over the benchmark scheme.</div></div>","PeriodicalId":50637,"journal":{"name":"Computer Networks","volume":"272 ","pages":"Article 111722"},"PeriodicalIF":4.6,"publicationDate":"2025-09-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145109813","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}