IEEE Transactions on Cloud Computing最新文献_第3页

An Efficient NTRU-Type MPHE Scheme for Sensitive Data in Cloud Computing 云计算中敏感数据的高效ntru型MPHE方案

IF 5.3 2区计算机科学

IEEE Transactions on Cloud Computing Pub Date : 2026-07-01 Epub Date: 2026-03-18 DOI: 10.1109/TCC.2026.3694356

Jianxin Gao;Ying Gao

{"title":"An Efficient NTRU-Type MPHE Scheme for Sensitive Data in Cloud Computing","authors":"Jianxin Gao;Ying Gao","doi":"10.1109/TCC.2026.3694356","DOIUrl":"https://doi.org/10.1109/TCC.2026.3694356","url":null,"abstract":"The development of cloud computing has heightened demand for sensitive data privacy. Multi-party homomorphic encryption (MPHE) enables homomorphic evaluation on encrypted data from multiple users, providing a crucial solution for secure multi-user computation in cloud computing models. NTRU-type MPHE schemes are particularly promising due to their simplicity and efficiency; however, existing solutions face three key limitations: a lack of threshold decryption (<inline-formula><tex-math>$t$</tex-math></inline-formula>-out-of-<inline-formula><tex-math>$N$</tex-math></inline-formula>), reliance on trusted third parties, and vulnerability to subfield lattice attacks. In this paper, we propose GTS, a novel NTRU-type MPHE scheme. Compared to existing schemes, GTS offers several advantages. On the one hand, we design a <inline-formula><tex-math>$t$</tex-math></inline-formula>-out-of-<inline-formula><tex-math>$N$</tex-math></inline-formula> threshold access structure, allowing GTS to facilitate threshold decryption with predefined participants. On the other hand, we combine distributed key generation and Lagrange interpolation, enabling GTS to operate without relying on trusted third parties and enhancing its resistance to subfield lattice attacks. These features effectively address the aforementioned challenges. We provide the correctness and security proofs for GTS, along with comprehensive analyses of computational and communication overheads. Experimental evaluations demonstrate significant efficiency improvements: compared to state-of-the-art methods, GTS achieves total execution time speedups of up to <inline-formula><tex-math>$2.4times$</tex-math></inline-formula> and <inline-formula><tex-math>$3.7times$</tex-math></inline-formula> at 128-bit security with <inline-formula><tex-math>$N=20$</tex-math></inline-formula> parties.","PeriodicalId":13202,"journal":{"name":"IEEE Transactions on Cloud Computing","volume":"14 3","pages":"1407-1420"},"PeriodicalIF":5.3,"publicationDate":"2026-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148903111","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

AdaNUMA: A Stability-Aware NUMA Scheduling Framework for Multi-Tenant Cloud Servers AdaNUMA：用于多租户云服务器的稳定性感知NUMA调度框架

IF 5.3 2区计算机科学

IEEE Transactions on Cloud Computing Pub Date : 2026-07-01 Epub Date: 2026-06-23 DOI: 10.1109/TCC.2026.3706533

WenTao Li;Jie Zhao;ZhiWen Chen;SongQuan Zhu;Zheng Zhang;Xing Chen;He Li;Ming Hui

{"title":"AdaNUMA: A Stability-Aware NUMA Scheduling Framework for Multi-Tenant Cloud Servers","authors":"WenTao Li;Jie Zhao;ZhiWen Chen;SongQuan Zhu;Zheng Zhang;Xing Chen;He Li;Ming Hui","doi":"10.1109/TCC.2026.3706533","DOIUrl":"https://doi.org/10.1109/TCC.2026.3706533","url":null,"abstract":"Multi-socket NUMA servers are widely deployed in cloud data centers to improve throughput and resource efficiency. However, under multi-tenant consolidation and co-location, CPU load, free memory on NUMA nodes, and memory-bandwidth contention become highly time-varying. Static placement policies, such as local binding and interleaving, often fail to track workload phase changes and runtime pressure. This can degrade locality, create node hotspots, and increase tail-latency variability. OS mechanisms such as AutoNUMA improve locality through reactive page migration, but they may respond late or oscillate under transient interference. This paper presents AdaNUMA, a lightweight host-level adaptive NUMA scheduling framework. AdaNUMA formulates memory placement as a joint node–policy decision. It selects both the target NUMA node and the corresponding memory-placement policy, such as binding or interleaving. Admitted updates are applied through controlled epoch-level reconfiguration, rather than through in-run kernel page migration. For online decision making, AdaNUMA trains lightweight gradient-boosted decision tree (GBDT) predictors from offline runtime traces. The implementation uses Light Gradient Boosting Machine (LightGBM) and a compact online feature subset to score candidate node–policy actions. To ensure stable adaptation under noisy multi-tenant dynamics, AdaNUMA introduces a Threshold–Window–Rollback (TR) controller. TR gates candidate updates, validates sustained benefits, and suppresses unsafe decisions through cooldown and rollback. We implement AdaNUMA on an eight-node NUMA server and evaluate it using microbenchmarks, macrobenchmarks, and co-located interference workloads. Across our benchmark suite, AdaNUMA reduces execution time by 18%–25%, improves sustainable memory bandwidth by more than 20%, and reduces tail-latency variability by up to 50% compared with fixed placement policies. It keeps runtime overhead about 3%.","PeriodicalId":13202,"journal":{"name":"IEEE Transactions on Cloud Computing","volume":"14 3","pages":"1374-1391"},"PeriodicalIF":5.3,"publicationDate":"2026-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148903156","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

BRAV: Blockchain-Enabled Revocable Authentication for UAV-Assisted Internet of Vehicles BRAV：无人机辅助车联网的区块链可撤销认证

IF 5.3 2区计算机科学

IEEE Transactions on Cloud Computing Pub Date : 2026-07-01 Epub Date: 2026-08-04 DOI: 10.1109/TCC.2026.3720306

Jiayin Wang;Chang Xu;Liehuang Zhu;Xinyue Ke;Xinbo Liu

{"title":"BRAV: Blockchain-Enabled Revocable Authentication for UAV-Assisted Internet of Vehicles","authors":"Jiayin Wang;Chang Xu;Liehuang Zhu;Xinyue Ke;Xinbo Liu","doi":"10.1109/TCC.2026.3720306","DOIUrl":"https://doi.org/10.1109/TCC.2026.3720306","url":null,"abstract":"Uncrewed aerial vehicle (UAV)-assisted Internet of Vehicles (IoV) improves the robustness of vehicle-to-Road Side Unit (RSU) connectivity in settings with sparse roadside infrastructure and severe urban blockage. However, introducing airborne relays and open wireless links also substantially enlarges the attack surface. Existing authentication schemes for this scenario typically provide only session-level anonymity, while revocation is either omitted or delegated to a single trusted authority, making the system vulnerable to well-known attacks and yielding revocation views that are neither scalable nor consistent across RSUs. To address these issues, we propose BRAV (Blockchain-enabled Revocable Authentication for UAV-assisted Internet of Vehicles). BRAV enhances self-updating hash chains to enable cross-session pseudonym evolution with public-transcript unlinkability against external observers. Moreover, BRAV introduces a Dual Cuckoo Filter revocation mechanism that leverages the globally confirmed state of a consortium blockchain to support fast, compact, and verifiable revocation queries, while maintaining a consistent revocation view among RSUs. We prove session-key security of BRAV in the Real-or-Random model and validate its authentication and secrecy properties using Scyther. Experimental results show that BRAV maintains practical computation and communication overhead while providing stronger security coverage than lightweight baselines. Its filter-based revocation significantly improves query efficiency over conventional certificate revocation lists (CRLs) and a single-filter baseline, and reduces storage consumption by about 87% compared to CRLs. Finally, NS-3 and blockchain simulations further demonstrate the applicability of BRAV in UAV-assisted IoV deployments.","PeriodicalId":13202,"journal":{"name":"IEEE Transactions on Cloud Computing","volume":"14 3","pages":"1519-1534"},"PeriodicalIF":5.3,"publicationDate":"2026-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148903181","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

Efficient and Privacy-Preserving Multi-Label Nearest Neighbors Query for Location-Based Recommendation Services 基于位置推荐服务的高效且隐私保护的多标签最近邻查询

IF 5.3 2区计算机科学

IEEE Transactions on Cloud Computing Pub Date : 2026-07-01 Epub Date: 2026-06-24 DOI: 10.1109/TCC.2026.3706686

Fengwei Wang;Xinyu Wang;Hui Zhu;Yandong Zheng;Songnian Zhang;Hui Li

{"title":"Efficient and Privacy-Preserving Multi-Label Nearest Neighbors Query for Location-Based Recommendation Services","authors":"Fengwei Wang;Xinyu Wang;Hui Zhu;Yandong Zheng;Songnian Zhang;Hui Li","doi":"10.1109/TCC.2026.3706686","DOIUrl":"https://doi.org/10.1109/TCC.2026.3706686","url":null,"abstract":"In recent years, Location-Based Recommendation Services (LBRSs) have played a pivotal role in the fields of navigation, social interaction, travelling, and so on. However, in LBRSs, query users and the Points of Interest (POI) data owner should inevitably share their sensitive location data with the cloud, which brings serious privacy leakage concerns. In this paper, we propose an efficient and privacy-preserving multi-label nearest neighbors query for LBRSs, named EPmNN. Specifically, through combining the voronoi diagram and k-d tree, a Vo2d-tree data structure is first constructed to achieve efficient Nearest Neighbor (NN) query, which eliminates the backtracking process in traditional k-d tree based NN query. Then, based on the overlapping relationship of voronoi diagrams, an overlapping table for POI datasets with different labels is designed, which achieves multi-label nearest neighbors query through performing a single tree search. Finally, based on a symmetric homomorphic encryption algorithm, a series of privacy-preserving computation protocols are designed for securely searching the Vo2d-tree, with which the security of sensitive location information of query users and the data owner is guaranteed. Security proof demonstrates the location data protection strength of EPmNN. Furthermore, we evaluated the performance of EPmNN using both real-world location datasets and synthetic datasets. The results demonstrate that EPmNN is indeed efficient.","PeriodicalId":13202,"journal":{"name":"IEEE Transactions on Cloud Computing","volume":"14 3","pages":"1745-1759"},"PeriodicalIF":5.3,"publicationDate":"2026-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148902632","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

Causality-Aware Graph Mixture of Experts for Accurate Multi-Cloud Workload Prediction 用于多云工作负载准确预测的因果关系感知图混合专家

IF 5.3 2区计算机科学

IEEE Transactions on Cloud Computing Pub Date : 2026-07-01 Epub Date: 2026-03-04 DOI: 10.1109/TCC.2026.3689897

Yongcan Luo;Zhihao Yu;Jiahao Zheng;Zhengjie Yang;Lei Sun;Dapeng Wu

{"title":"Causality-Aware Graph Mixture of Experts for Accurate Multi-Cloud Workload Prediction","authors":"Yongcan Luo;Zhihao Yu;Jiahao Zheng;Zhengjie Yang;Lei Sun;Dapeng Wu","doi":"10.1109/TCC.2026.3689897","DOIUrl":"https://doi.org/10.1109/TCC.2026.3689897","url":null,"abstract":"Multi-cloud deployments are becoming increasingly necessary to meet data locality and compliance requirements, reduce latency for AI services, and mitigate the risk of single-provider failure. However, multi-cloud telemetry forms high-dimensional, heterogeneous and nonstationary time series where CPU, memory, disk, and LAN/WAN I/O exhibit time-varying, directional lead–lag effects across VMs and regions. Previous approaches, from RNNs to recent attention/GNN models, either assume linear stationarity or learn correlation-driven, largely symmetric dependencies that blur directed causal influence. To better leverage temporal dependencies and leverage multi-cloud deployments, we propose CAGMoE, a causality-aware dual-router Graph Mixture-of-Experts for multi-cloud workload forecasting. First, CAGMoE constructs two complementary graphs per window: an inter-metric graph derived from a transfer-entropy proxy of Granger causality to encode directed cross-metric influence, and an intra-temporal graph to capture local temporal continuity. Furthermore, a shared graph encoder produces token states and path summaries that drive dual routers to form Top-K sparse mixtures over experts. Finally, each expert is a FiLM-conditioned feed-forward network that injects a window-level causal vector to generate the final prediction. Experiments on real-world multi-cloud dataset MUCEP, Google Cluster, and Ali traces demonstrate that our method robustly compares with previous baselines, which improve in both accuracy and reliability, and demonstrate potential for industrial use.","PeriodicalId":13202,"journal":{"name":"IEEE Transactions on Cloud Computing","volume":"14 3","pages":"1569-1584"},"PeriodicalIF":5.3,"publicationDate":"2026-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148902649","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

Efficient and Secure Model Initialization Scheme in Federated Learning 联邦学习中高效安全的模型初始化方案

IF 5.3 2区计算机科学

IEEE Transactions on Cloud Computing Pub Date : 2026-07-01 Epub Date: 2026-03-26 DOI: 10.1109/TCC.2026.3697363

Jianchuan Zeng;Jianchang Lai;Shengmin Xu;Zhuoran Zhang;Chao Sun

{"title":"Efficient and Secure Model Initialization Scheme in Federated Learning","authors":"Jianchuan Zeng;Jianchang Lai;Shengmin Xu;Zhuoran Zhang;Chao Sun","doi":"10.1109/TCC.2026.3697363","DOIUrl":"https://doi.org/10.1109/TCC.2026.3697363","url":null,"abstract":"In Federated Learning (FL), a shared initial global model is distributed by an aggregator through a cloud server to all participating clients for the purpose of local training. To mitigate the threat of a potentially malicious cloud server, consistency verification of the client-received initial model is essential. Public key encryption with equality test (PKEET) enables authorized entities to determine whether two ciphertexts contain an identical message or not without decryption. This property supports consistency verification of the initial global model in FL settings. Identity-based encryption with equality test (IBEET) eliminates the certificate, resolving certificate management issues in PKEET. SM9, as a Chinese identity-based cryptographic standard, has been applied in many fields. However, existing equality test schemes based on SM9 only support equality tests on two ciphertexts, resulting in linear executions of the test algorithm when performing tests on multiple ciphertexts. To enhance efficiency and functionality, we propose an efficient multi-ciphertext equality test scheme based on SM9 (SM9-MET), which is well-suited for verifying the initial model in FL. The proposed SM9-MET scheme supports simultaneous equality testing on multiple ciphertexts while maintaining constant-size ciphertexts. Our method requires only one single execution of the test algorithm. Under the <inline-formula><tex-math>$q$</tex-math></inline-formula>-BDHI assumption in the random oracle model, we prove that the SM9-MET scheme achieves one-way security and indistinguishability against chosen-plaintext attacks. Finally, we compare the SM9-MET scheme to other related schemes, and the results demonstrate considerable advantages in both encryption and equality tests on multiple ciphertexts.","PeriodicalId":13202,"journal":{"name":"IEEE Transactions on Cloud Computing","volume":"14 3","pages":"1760-1771"},"PeriodicalIF":5.3,"publicationDate":"2026-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148902889","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

Temporal-Aware GPU Resource Allocation for Distributed LLM Inference via Reinforcement Learning 基于强化学习的分布式LLM推理的时间感知GPU资源分配

IF 5.3 2区计算机科学

IEEE Transactions on Cloud Computing Pub Date : 2026-07-01 Epub Date: 2026-06-11 DOI: 10.1109/TCC.2026.3702578

Chengze Du;Zhiwei Yu;Heng Xu;Haojie Wang;Bo Liu;Jialong Li

{"title":"Temporal-Aware GPU Resource Allocation for Distributed LLM Inference via Reinforcement Learning","authors":"Chengze Du;Zhiwei Yu;Heng Xu;Haojie Wang;Bo Liu;Jialong Li","doi":"10.1109/TCC.2026.3702578","DOIUrl":"https://doi.org/10.1109/TCC.2026.3702578","url":null,"abstract":"The rapid growth of large language model (LLM) services imposes increasing demands on distributed GPU inference infrastructure. Most existing scheduling systems follow a reactive paradigm, relying solely on the current system state to make decisions, without considering how task demand and resource availability evolve over time. This lack of temporal awareness in reactive approaches leads to inefficient GPU utilization, high task migration overhead, and poor system responsiveness under dynamic workloads. In this work, we identify the fundamental limitations of these instantaneous-state-only scheduling approaches and propose Temporal Optimal Resource scheduling via Two-layer Architecture (TORTA). TORTA introduces a spatiotemporal scheduling framework that captures both long-term workload patterns and short-term execution constraints. It adopts a two-layer design: a macro-level scheduler leverages reinforcement learning and optimal transport to coordinate inter-region task distribution, while a micro-level allocator refines task-to-server assignments within each region to reduce latency and switching costs. Experimental results across multiple network topologies show that TORTA reduces average inference response time by up to 15%, improves load balance by approximately 4-5%, and cuts total operational cost by 10-20% compared to state-of-the-art baseline methods.","PeriodicalId":13202,"journal":{"name":"IEEE Transactions on Cloud Computing","volume":"14 3","pages":"2181-2194"},"PeriodicalIF":5.3,"publicationDate":"2026-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148902971","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

HEART: Achieving Timely Multi-Model Training for Vehicle-Edge-Cloud-Integrated Hierarchical Federated Learning 实现车辆边缘云集成分层联邦学习的及时多模型训练

IF 5.3 2区计算机科学

IEEE Transactions on Cloud Computing Pub Date : 2026-07-01 Epub Date: 2026-03-28 DOI: 10.1109/TCC.2026.3698093

Xiaohong Yang;Minghui Liwang;Xianbin Wang;Zhipeng Cheng;Seyyedali Hosseinalipour;Huaiyu Dai;Zhenzhen Jiao

{"title":"HEART: Achieving Timely Multi-Model Training for Vehicle-Edge-Cloud-Integrated Hierarchical Federated Learning","authors":"Xiaohong Yang;Minghui Liwang;Xianbin Wang;Zhipeng Cheng;Seyyedali Hosseinalipour;Huaiyu Dai;Zhenzhen Jiao","doi":"10.1109/TCC.2026.3698093","DOIUrl":"https://doi.org/10.1109/TCC.2026.3698093","url":null,"abstract":"The rapid growth of AI-enabled Internet of Vehicles (IoV) calls for efficient Machine Learning (ML) solutions that can handle high vehicular mobility and decentralized data. This has motivated the emergence of Hierarchical Federated Learning over vehicle-edge-cloud architectures (VEC-HFL). Nevertheless, one aspect which is underexplored in the literature on VEC-HFL is that vehicles often need to execute multiple ML tasks simultaneously, where this multi-model training environment introduces crucial challenges. First, improper aggregation rules can lead to model obsolescence and prolonged training times. Second, vehicular mobility may result in inefficient data utilization by preventing the vehicles from returning their models to the network edge. Third, achieving a balanced resource allocation across diverse tasks becomes of paramount importance as it majorly affects the effectiveness of collaborative training. We take one of the first steps towards addressing these challenges via proposing a framework for multi-model training in dynamic VEC-HFL with the goal of minimizing global training latency while ensuring balanced training across various tasks, a problem that turns out to be NP-hard. To facilitate timely model training, we introduce a hybrid synchronous-asynchronous aggregation rule. Building on this, we present a novel method called Hybrid Evolutionary And gReedy allocaTion (HEART). The framework operates in two stages: first, it achieves balanced task scheduling through a hybrid heuristic approach that combines improved Particle Swarm Optimization (PSO) and Genetic Algorithms (GA); second, it employs a low-complexity greedy algorithm to determine the training priority of assigned tasks on vehicles. Experiments on real-world datasets demonstrate the superiority of HEART over existing methods.","PeriodicalId":13202,"journal":{"name":"IEEE Transactions on Cloud Computing","volume":"14 3","pages":"1888-1903"},"PeriodicalIF":5.3,"publicationDate":"2026-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148903163","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

DAGless: Automating End-to-End Support for Serverless DAG Lifecycles With Cost-Efficient Execution 无DAG：自动化对无服务器DAG生命周期的端到端支持，具有成本效益的执行

IF 5.3 2区计算机科学

IEEE Transactions on Cloud Computing Pub Date : 2026-07-01 Epub Date: 2026-07-27 DOI: 10.1109/TCC.2026.3717022

Philipp Gritsch;Simon Triendl;Sashko Ristov

{"title":"DAGless: Automating End-to-End Support for Serverless DAG Lifecycles With Cost-Efficient Execution","authors":"Philipp Gritsch;Simon Triendl;Sashko Ristov","doi":"10.1109/TCC.2026.3717022","DOIUrl":"https://doi.org/10.1109/TCC.2026.3717022","url":null,"abstract":"Developing cost-efficient serverless DAGs (Directed Acyclic Graphs) is a challenging process involving (1) the non-trivial FaaSification of monolithic workflows into serverless environments, (2) profiling to determine runtime characteristics, and (3) effectively allocating tasks and resources to serverless functions. Workflows often rely on implicit dependencies (e.g., external systems, files), which introduce requirements that only surface at runtime, complicating the adaptation to serverless architectures. Managing these dependencies within serverless environments requires a holistic approach that integrates information gathered across the entire lifecycle of serverless DAGs. Unfortunately, current solutions often address only isolated parts of this process, typically requiring significant manual effort. To overcome these challenges, we introduce DAGless, a platform that automates end-to-end support for the serverless DAG lifecycle. DAGless transforms the workflow through three models, each representing a specific phase of the lifecycle, iteratively refining the serverless DAG to achieve cost-efficient execution. DAGless estimates costs of the transformed serverless DAG with a minimum accuracy of 91.07% and provides up to 8× cheaper profiling. DAGless dominates state-of-the-art cost optimizers, not only in terms of cost (16.93%), but also in terms of performance (4.27×).","PeriodicalId":13202,"journal":{"name":"IEEE Transactions on Cloud Computing","volume":"14 3","pages":"1673-1687"},"PeriodicalIF":5.3,"publicationDate":"2026-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=11623681","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148903166","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Bayesian Federated Learning With Personalized Parameters for Cloud-Edge Collaboration 面向云边缘协作的个性化参数贝叶斯联邦学习

IF 5.3 2区计算机科学

IEEE Transactions on Cloud Computing Pub Date : 2026-07-01 Epub Date: 2026-04-15 DOI: 10.1109/TCC.2026.3684083

Yunsheng Zhang;Yiheng Ruan;Laurence T. Yang;Honglu Zhao;Xianjun Deng;Zecan Yang

{"title":"Bayesian Federated Learning With Personalized Parameters for Cloud-Edge Collaboration","authors":"Yunsheng Zhang;Yiheng Ruan;Laurence T. Yang;Honglu Zhao;Xianjun Deng;Zecan Yang","doi":"10.1109/TCC.2026.3684083","DOIUrl":"https://doi.org/10.1109/TCC.2026.3684083","url":null,"abstract":"The rapid development of cloud-edge collaboration has facilitated the large-scale generation, transmission, and storage of data across an expanding device network, offering abundant resources for the application of federated learning (FL). However, the non-independent and identically distributed (non-IID) characteristics of data impose considerable challenges on the traditional FL approach, causing difficulties in achieving model convergence and performance. Moreover, the limited training data in edge devices increases the risk of overfitting during local training and also reduces the generalization ability of the model. To address the above challenges, this article proposes a Bayesian federated learning method with personalized parameters (BFedpp). This method employs Bayesian learning to perform probabilistic modeling of FL problems under cloud-edge collaboration, effectively mitigating the impact of statistical heterogeneity across clients. Additionally, it adopts a mixed-effects approach, decomposing the client model into shared fixed-effect parameters and personalized random-effect parameters to better represent local client characteristics. Finally, to enhance model performance, a tensor-enabled adaptive parameter selection method is proposed, which generates a mask tensor with the appropriate personalized parameter distribution for each client based on quantitative sensitivity indicators. Experimental investigations showed that, in the context of limited data availability and substantially unbalanced distributions, BFedpp outperformed existing methods in terms of accuracy on the CIFAR-10 and CIFAR-100 datasets.","PeriodicalId":13202,"journal":{"name":"IEEE Transactions on Cloud Computing","volume":"14 3","pages":"1479-1490"},"PeriodicalIF":5.3,"publicationDate":"2026-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148903230","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