Computer Communications最新文献

{"title":"Streaming algorithm and hardware accelerator for high-throughput entropy estimation of network flows in sliding windows","authors":"Yaime Fernández ,&nbsp;Javier E. Soto ,&nbsp;Carolina Gallardo-Pavesi ,&nbsp;Yasmany Prieto ,&nbsp;Cecilia Hernández ,&nbsp;Miguel Figueroa","doi":"10.1016/j.comcom.2026.108446","DOIUrl":"10.1016/j.comcom.2026.108446","url":null,"abstract":"<div><div>Shannon’s empirical entropy is widely used in network monitoring for anomaly detection and resource management. Since recent traffic is most relevant for predicting network behavior, entropy estimation is usually performed over sliding time windows. In modern links this task involves high-speed data streams, where both throughput and memory efficiency are critical. Hardware accelerators provide low-latency, energy-efficient processing within the network, but face strict on-chip memory limits. We propose a streaming algorithm for entropy estimation in sliding windows with discrete steps, tailored to these hardware constraints. The algorithm combines frequency and cardinality sketches, storing the most frequent flows and estimating the rest under a set of observations of the sorted log–log scaled histogram of the flows. Flow frequencies for a window <span><math><mi>W</mi></math></span> are reconstructed by merging ten non-sliding frequency sketches, each covering an interval <span><math><mrow><mi>s</mi><mo>=</mo><mi>W</mi><mo>/</mo><mn>10</mn></mrow></math></span>. We also design a Field-Programmable Gate Array (FPGA) accelerator architecture that implements this algorithm. On eight real traffic traces with two million flows per window, the algorithm achieves mean relative errors below 0.45% with a standard deviation of 0.14%. The accelerator, implemented on a Xilinx Virtex XCU55 UltraScale+ FPGA, sustains packet processing at over 160 Gbps while using 56% of the available BRAM, leaving capacity for additional tasks. These results show that accurate entropy estimation in sliding windows with discrete steps enables scalable, high-speed monitoring of modern networks.</div></div>","PeriodicalId":55224,"journal":{"name":"Computer Communications","volume":"250 ","pages":"Article 108446"},"PeriodicalIF":4.3,"publicationDate":"2026-03-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146175259","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}

{"title":"A metaheuristic study: A partial multitask offloading strategy for vehicular fog computing based on adaptive particle swarm optimization","authors":"Fuqi Zhang ,&nbsp;Huilin Jiang ,&nbsp;Fu Liu ,&nbsp;Tao Hou ,&nbsp;Yujia Liu","doi":"10.1016/j.comcom.2026.108455","DOIUrl":"10.1016/j.comcom.2026.108455","url":null,"abstract":"<div><div>To satisfy the demand for low perceived latency in 5G vehicular services, a metaheuristic latency-aware optimization strategy based on the adaptive particle swarm optimization (PSO) algorithm is proposed. First, a single road side unit intensive short-range urban traffic scenario is modelled. Second, to address the described nondeterministic polynomial-time-hard problem, a nonlinear programming method is employed to schedule the idle vehicular fog nodes within a one-hop communication range. Finally, the adaptive PSO strategy is applied to optimize the total and average latencies of task processing by balancing the loads and perceived latency levels of the cloud and vehicular fog nodes. In the numerical simulation experiments, the total processing latency of the proposed strategy is reduced by up to 11.6% compared with that of the classic PSO algorithm, which is one of the most well-known bio-inspired approaches. The total processing latency is reduced by at least 38.0% compared with that of the other test schemes, which are demonstrated effective in this scenario. The experimental results indicate that the proposed method can be used to balance the loads of cloud and vehicular fog nodes and improve the quality of vehicle services and user satisfaction, which provides a suitable solution for the vehicle-fog-cloud scenario of urban smart transportation.</div></div>","PeriodicalId":55224,"journal":{"name":"Computer Communications","volume":"250 ","pages":"Article 108455"},"PeriodicalIF":4.3,"publicationDate":"2026-03-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146175260","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}

{"title":"A multiclass cost–latency aware framework for multi-tiered cloud storage optimization via access pattern forecasting","authors":"Flávio A.A. Motta ,&nbsp;Saulo Moraes Villela ,&nbsp;Heder Soares Bernardino ,&nbsp;Glauber Dias Gonçalves ,&nbsp;Alex Borges Vieira","doi":"10.1016/j.comcom.2026.108437","DOIUrl":"10.1016/j.comcom.2026.108437","url":null,"abstract":"<div><div>Efficient management of cloud storage resources requires intelligent tier allocation strategies that balance cost optimization with performance requirements. While previous approaches have focused on binary classification schemes for storage tiering, real-world scenarios demand more granular solutions that can adapt to diverse user preferences and workload characteristics. This paper extends our previous work on access frequency prediction by proposing a comprehensive multiclass machine learning framework for intelligent cloud storage tiering. The proposed framework incorporates a novel three-tier classification system (<em>Cold</em>/<em>Warm</em>/<em>Hot</em>) and integrates user-centric preferences through a cost-weight parameter, enabling dynamic adaptation to varying preferences along the cost–latency spectrum. We demonstrate the framework’s effectiveness through extensive experiments on real-world access patterns, where we assess the performance of thirteen machine learning algorithms under various user preference profiles. The results show that our multiclass approach achieves cost reductions of up to 40% compared to a static tiering strategy, while providing Pareto-optimal solutions for different user profiles. Through comprehensive Pareto frontier analysis, we demonstrate the framework’s ability to provide transparent trade-off visualization, enabling informed decision-making for cloud storage administrators. Our main contributions are: a multiclass classification approach for storage tiering, the integration of user preferences via parameterized optimization, a comparative analysis of multiple algorithms across different preference configurations, and a practical validation of the framework’s applicability in production cloud storage environments.</div></div>","PeriodicalId":55224,"journal":{"name":"Computer Communications","volume":"249 ","pages":"Article 108437"},"PeriodicalIF":4.3,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146174822","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}

{"title":"AI-driven intrusion detection for UAV in Smart Urban ecosystems: A comprehensive survey","authors":"Abdullah Khanfor ,&nbsp;Raby Hamadi ,&nbsp;Noureddine Lasla ,&nbsp;Hakim Ghazzai","doi":"10.1016/j.comcom.2026.108435","DOIUrl":"10.1016/j.comcom.2026.108435","url":null,"abstract":"<div><div>UAVs have the potential to revolutionize urban management and provide valuable services to citizens. They can be deployed across diverse applications, including traffic monitoring, disaster response, environmental monitoring, and numerous other domains. However, this integration introduces novel security challenges that must be addressed to ensure safe and trustworthy urban operations. This paper provides a structured, evidence-based synthesis of UAV applications in smart cities and their associated security challenges as reported in the literature over the last decade, with particular emphasis on developments from 2019 to 2025. We categorize these challenges into two primary classes: (1) cyber-attacks targeting the communication infrastructure of UAVs and (2) unwanted or unauthorized physical intrusions by UAVs themselves. We examine the potential of Artificial Intelligence (AI) techniques in developing intrusion detection mechanisms to mitigate these security threats. We analyze how AI-based methods, such as machine/deep learning for anomaly detection and computer vision for object recognition, can play a pivotal role in enhancing UAV security through unified detection systems that address both cyber and physical threats. Furthermore, we consolidate publicly available UAV datasets across network traffic and vision modalities suitable for Intrusion Detection Systems (IDS) development and evaluation. The paper concludes by identifying ten key research directions, including scalability, robustness, explainability, data scarcity, automation, hybrid detection, large language models, multimodal approaches, federated learning, and privacy preservation. Finally, we discuss the practical challenges of implementing UAV IDS solutions in real-world smart city environments.</div></div>","PeriodicalId":55224,"journal":{"name":"Computer Communications","volume":"249 ","pages":"Article 108435"},"PeriodicalIF":4.3,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146081614","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}

{"title":"Perspectives toward biosimilars among oncologists: A Malaysian survey.","authors":"Soon Cien Chong, Retha Rajah, Poh Lee Chow, Hsio Ching Tan, Chin Man Chong, Kar Yee Khor, Wan Ping Lee, Wan Ying Tan","doi":"10.1177/10781552221104773","DOIUrl":"10.1177/10781552221104773","url":null,"abstract":"<p><p>IntroductionBiosimilars confer significant cost-saving advantages and expand patients' access to biologic therapies in cancer care. In line with the increasing availability of antineoplastic biosimilars, it is pertinent to understand the oncologists' view on the adoption of biosimilars in their clinical practice. The study aimed to assess (i) the prevalence of biosimilar use, (ii) perception towards biosimilars, (iii) factors influencing the use of biosimilars and (iv) knowledge about biosimilars among Malaysian oncologists.MethodsA cross-sectional survey was conducted among clinical oncologists and medical oncologists in Malaysia between January 2020 and February 2021 using a structured 31-item questionnaire.ResultsAmong the 121 oncologists registered in the country, 36 responded (response rate  =  30%). A total of 64% of the respondents prescribed biosimilars either often or always. Most oncologists (72%) agreed or strongly agreed that switching will not have a significant effect on the treatment benefit, with lower percentages saying that they agreed or strongly agreed that it will not lead to the emergence of additional adverse effects (56%) or harmful immunogenicity (64%). Patients' preferences (40%) and the non-availability of biosimilars in hospitals (34%) are the major barriers cited to the prescribing of biosimilars. Cost differences and robust pharmacovigilance activities are the two most important factors that would influence the prescribing of biosimilars. The mean score of knowledge in biosimilar among respondents was 3.81 (± 0.86) out of a maximum possible score of 6.ConclusionsThe identified gap in prescribing and the use of biosimilars among Malaysian oncologists warrant educational intervention and robust pharmacovigilance activities to facilitate the prescribing of biosimilars and ultimately increase the accessibility to biologics in cancer treatment.</p>","PeriodicalId":55224,"journal":{"name":"Computer Communications","volume":"12 1","pages":"189-199"},"PeriodicalIF":4.3,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"75120995","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}

{"title":"A unified utility-based framework for joint scheduling and buffer management in Delay Tolerant Networks","authors":"Tuan Le","doi":"10.1016/j.comcom.2026.108432","DOIUrl":"10.1016/j.comcom.2026.108432","url":null,"abstract":"<div><div>Data delivery in Delay Tolerant Networks (DTNs) is fundamentally constrained by two scarce resources: node buffer space and contact duration. Conventional approaches typically address message scheduling and buffer management as isolated optimization problems, resulting in suboptimal resource utilization where high-utility messages are often dropped to accommodate lower-value traffic. To bridge this gap, this paper proposes a unified, utility-based framework that jointly optimizes these decisions by formulating the contact opportunity as a Multidimensional 0/1 Knapsack Problem (MKP). We derive rigorous, closed-form marginal utility functions for three distinct objectives: maximizing delivery probability, minimizing average latency, and a composite metric balancing both. Unlike static heuristics, these metrics are derived from a deadline-constrained probabilistic model that explicitly quantifies the marginal benefit of replication relative to the message’s remaining Time-to-Live (TTL). To solve the resulting NP-hard joint allocation problem in real time, we introduce a computationally efficient greedy heuristic based on Utility Density. Trace-driven simulations using real-world vehicular mobility datasets (San Francisco and Rome) demonstrate that our unified policy outperforms state-of-the-art baselines, including ReAR and OBSBM. Beyond enabling flexible Quality of Service (QoS) enforcement via a tunable weighting coefficient, the overall analysis demonstrates that the proposed framework effectively resolves resource contention, achieving up to a 40% improvement in Delivery Ratio in sparse environments and a <span><math><mrow><mn>3</mn><mo>×</mo></mrow></math></span> reduction in Average Latency when optimized for speed compared to existing techniques.</div></div>","PeriodicalId":55224,"journal":{"name":"Computer Communications","volume":"249 ","pages":"Article 108432"},"PeriodicalIF":4.3,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146071086","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}

{"title":"Learning to reallocate: MAPPO-based spectrum and power optimization for UAV–UGV clusters with dynamic reconfiguration","authors":"Wei Wang ,&nbsp;Panfeng He ,&nbsp;Boyu Wan,&nbsp;Yong Chen,&nbsp;Yu Zhang","doi":"10.1016/j.comcom.2026.108434","DOIUrl":"10.1016/j.comcom.2026.108434","url":null,"abstract":"<div><div>This paper proposes a Multi-Agent Reinforcement Learning (MARL) framework for joint spectrum and power optimization in unmanned aerial vehicle (UAV)-unmanned ground vehicle (UGV) clusters during dynamic reconfiguration. The system consists of heterogeneous agents (communication nodes and radars) operating in scenarios with malicious jamming and dynamic inter-cluster node transfers. The joint channel selection and power control problem is formulated as a partially observable Markov decision process (POMDP), with a Multi-Agent Proximal Policy Optimization (MAPPO)-based algorithm developed to address two key challenges: For fixed-topology networks, a MAPPO-based algorithm is proposed to maximum number of operational nodes while avoiding intra-cluster interference; For dynamic reconfiguration scenarios, an estimated maximum node capacity (EMNC)-based algorithm is proposed enabling rapid adaptation to topology changes. Simulation results demonstrate that the proposed approach achieves 93%–97% operational node ratios in static configurations (outperforming baseline methods by 15%–40%) while maintaining 90%–94% operational efficiency during dynamic reconfiguration events — a significant improvement over baseline methods that typically suffer 20%–30% performance degradation during topology changes. The proposed solution uniquely combines real-time decision-making with robust adaptation capabilities, offering a practical approach for resilient resource management in dynamic UAV–UGV networks where conventional methods fail to address both dynamic reconfiguration and adversarial interference simultaneously.</div></div>","PeriodicalId":55224,"journal":{"name":"Computer Communications","volume":"249 ","pages":"Article 108434"},"PeriodicalIF":4.3,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146071112","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}

{"title":"FMSformer: Frequency-aware multi-scale spatio temporal correlation model for edge workload prediction","authors":"M.F. Sun,&nbsp;X.H. Zhao,&nbsp;J.K. Li,&nbsp;W.Q. Liu,&nbsp;Q.T. Wu","doi":"10.1016/j.comcom.2026.108454","DOIUrl":"10.1016/j.comcom.2026.108454","url":null,"abstract":"<div><div>Accurate workload prediction enables better service deployment and mitigates the risk of disruptions and performance degradation caused by limited resources. Existing workload prediction methods mostly rely on how workload data varies over time, while neglecting spatial dependencies and multi-scale interactions among edge servers, which results in degraded accuracy and robustness of the workload prediction. To solve this problem, we propose FMSformer, a frequency-aware multi-scale spatial temporal correlation model, which simultaneously models multi-scale temporal dynamics and spatial dependencies among edge servers. Specially, it jointly models cross-server collaborative evolution at multiple time scales by grouping related servers to capture spatial dependencies, while simultaneously extracting multi-scale temporal correlations within each individual server. Furthermore, we extract high energy frequency components to model global dependencies. Finally, a time–frequency mixed loss is employed to avoid label dependence, thereby improving prediction accuracy. Experiments on three real datasets show that FMSformer reduces Mean Absolute Error by 17.4% and Mean Squared Error by 8.4% compared to state-of-the-art models, demonstrating the effectiveness of method for complex edge workload prediction.</div></div>","PeriodicalId":55224,"journal":{"name":"Computer Communications","volume":"249 ","pages":"Article 108454"},"PeriodicalIF":4.3,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146174317","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}

{"title":"A survey on CSI-based Wi-Fi sensing datasets and models with a focus on reproducibility","authors":"Idio Guarino ,&nbsp;Damiano Carra ,&nbsp;Marco Cominelli ,&nbsp;Francesco Gringoli ,&nbsp;Renato Lo Cigno","doi":"10.1016/j.comcom.2026.108431","DOIUrl":"10.1016/j.comcom.2026.108431","url":null,"abstract":"<div><div>Wi-Fi sensing based on Channel State Information (CSI) has witnessed considerable research activity in recent years. However, a critical literature analysis reveals that only a limited amount of proposals are potentially reproducible, with many works lacking essential experimental details, publicly available datasets, or accessible analysis code. This may impede the research progress and the subsequent transition of promising findings into practical applications. The objective of this work is to identify CSI-based sensing proposals that are potentially reproducible based on the published information. Our goal is to provide a focused review of resources that can serve as a concrete starting point for researchers and practitioners seeking to experiment with and advance the field of Wi-Fi sensing. We perform a comprehensive analysis of publicly available datasets (encompassing both the collection methodologies and the environmental characteristics) and existing sensing models, accompanied by their code, pre-processing steps, and evaluation procedures. Finally, we discuss what are the minimum requirements for truly verifiable contributions in this field, and outline the best practices for creating and sharing reproducible CSI-based sensing datasets and models.</div></div>","PeriodicalId":55224,"journal":{"name":"Computer Communications","volume":"249 ","pages":"Article 108431"},"PeriodicalIF":4.3,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146174821","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}

{"title":"Dynamic resource allocation for digital twin-enhanced hierarchical federated learning in sustainable internet of things","authors":"Ze Wei ,&nbsp;Rongxi He ,&nbsp;Xiaojing Chen ,&nbsp;Chengzhi Song","doi":"10.1016/j.comcom.2025.108410","DOIUrl":"10.1016/j.comcom.2025.108410","url":null,"abstract":"<div><div>This paper proposes a digital twin (DT)-enhanced hierarchical federated learning framework for sustainable Internet of Things (IoT) networks. In this framework, mobile edge computing servers coordinate collaborative training, while DTs maintain real-time physical-virtual synchronization. Our core contributions are threefold. First, to tackle device heterogeneity, we propose two mechanisms: (1) an elastic time window that dynamically adapts aggregation deadlines based on median training times while incorporating distance-aware resource compensation to mitigate channel degradation, and (2) a DT-enhanced weighting strategy that dynamically balances energy sustainability, channel quality, and model freshness while guaranteeing convergence through closed-loop cross-layer coordination. Second, we derive a convergence bound explicitly linked to the device participation ratio, establishing a direct theoretical connection between resource allocation and learning performance. Then, through theoretical analysis, it can be found that reducing training latency and energy consumption by jointly optimizing computing and communication resources, as well as EH duration, is key to maximizing this ratio without compromising the reliability of the gradients, thereby indirectly enhancing convergence. Third, guided by this insight, we formulate a mixed-integer nonlinear programming problem that aims to maximize the participation ratio while jointly minimizing energy consumption and training latency, by optimizing the energy harvesting time, collaboration ratio, and communication/computation resources. To solve this NP-hard problem, we propose a DT-driven decomposition framework that partitions it into two subproblems, which are then solved by three DT-driven algorithms with provable near-optimality guarantees. Experimental results validate the superiority of our approach, demonstrating significant improvements in convergence performance, latency, energy efficiency, and participant sample rate, while also advancing the sustainability of FL.</div></div>","PeriodicalId":55224,"journal":{"name":"Computer Communications","volume":"248 ","pages":"Article 108410"},"PeriodicalIF":4.3,"publicationDate":"2026-02-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145929270","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}