Future Generation Computer Systems-The International Journal of Escience最新文献

{"title":"PPTADI: A privacy-preserving training and accelerated distributed inference framework in low-resource AIoT scenarios","authors":"Haoyang Meng,&nbsp;Yizhong Hu,&nbsp;Kexian Liu,&nbsp;Jianfeng Guan","doi":"10.1016/j.future.2025.108084","DOIUrl":"10.1016/j.future.2025.108084","url":null,"abstract":"<div><div>Artificial intelligence of things (AIoT) is a technology that combines AI and IoT, which realizes the interconnection and gives them more intelligent features between devices. However, there are some challenges in the data process. In this paper, we propose a new framework called PPTADI to train AI models while preserving privacy and to accelerate the inference process. Experiments show that PPTADI can effectively prevent label leakage, gradient attacks and model inversion attacks compared to the conventional split federated learning frameworks. In the meanwhile, PPTADI reduces the total inference delay by up to 35 % and the transmission delay by up to 65 % comparing with some SOTA schemes for distributed inference.</div></div>","PeriodicalId":55132,"journal":{"name":"Future Generation Computer Systems-The International Journal of Escience","volume":"175 ","pages":"Article 108084"},"PeriodicalIF":6.2,"publicationDate":"2025-08-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144898495","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":"A granular approach for enhancing node representation in heterogeneous graph learning","authors":"Ying Sun ,&nbsp;Hongjiang Ye ,&nbsp;Feiyi Xu ,&nbsp;Zhenjiang Dong ,&nbsp;Yanfei Sun ,&nbsp;Jin Qi","doi":"10.1016/j.future.2025.108080","DOIUrl":"10.1016/j.future.2025.108080","url":null,"abstract":"<div><div>Heterogeneous graph learning aims to generate meaningful node representations for graph-structured data with diverse node types and complex relations, facilitating downstream tasks such as node classification and clustering. However, existing methods often emphasize either coarse-grained relational structures or fine-grained node attributes, paying limited attention to the other, which constrains their ability to fully capture the intricate interplay between nodes and relations. To address this limitation, we propose a novel Granular Interaction Heterogeneous Graph Auto-Encoder (GIHGAE), which effectively balances granular fusion and interactions in heterogeneous graph learning. Specifically, GIHGAE employs a relation-level encoder as the primary structure extractor to capture coarse-grained relational dependencies across the graph. Complementarily, we design a node-level encoder that integrates fine-grained contextual details from diverse node attributes, refining representations. These multi-granular features are fused into holistic node embeddings. Additionally, to ensure seamless integration of fine-grained and coarse-grained information, we introduce a global-level decoder to model interactions between nodes and relations explicitly. Finally, to further enhance GIHGAE, we incorporate a dual-loss mechanism, combining reconstruction loss for feature preservation and prediction loss to enhance downstream task performance. Extensive experimental evaluations in heterogeneous graph learning tasks highlight the strong performance of GIHGAE, which consistently outperforms current state-of-the-art methods in classification accuracy, clustering quality, and link prediction performance.</div></div>","PeriodicalId":55132,"journal":{"name":"Future Generation Computer Systems-The International Journal of Escience","volume":"175 ","pages":"Article 108080"},"PeriodicalIF":6.2,"publicationDate":"2025-08-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144864309","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":"Blind and bidirectional ownership verification for deduplicated cloud computing systems","authors":"Jay Dave,&nbsp;Kamalesh Ram R,&nbsp;Pratik Patil,&nbsp;Himanshu Patil,&nbsp;Sarvesh Borole,&nbsp;Chinni Vamshi Krushna,&nbsp;Suyash Patil","doi":"10.1016/j.future.2025.108082","DOIUrl":"10.1016/j.future.2025.108082","url":null,"abstract":"<div><div>Cloud storage systems provide several benefits, such as scalable storage capacity, cost efficiency with pay-as-you-go pricing models, easy access from any location with an internet connection, and robust data backup options. These advantages drive the growing popularity of cloud storage, resulting in a rapid increase in the volume of data stored on the cloud. Deduplication is an effective data management technique used in these systems to reduce storage costs and enhance efficiency through the elimination of redundant data. However, in a deduplication system, a hash digest, i.e., a small piece of information, is used as ownership proof of the entire file. Therefore, a malicious user can gain access to a sensitive file already stored on the cloud by obtaining and presenting the hash digest of that file. On the other hand, data stored in the cloud may be susceptible to loss or damage due to various accidental or intentional reasons. Hence, there is a need for an ownership verification protocol where both the user and server can verify each other’s file ownership without revealing details about the file. Some existing state-of-the-art schemes consider the server as a trusted entity and focus solely on verifying the ownership of the user, while others emphasize bidirectional ownership verification but do not incorporate obliviousness in their solutions. In this paper, we propose a novel bidirectional and oblivious ownership verification scheme for deduplication systems. We cryptographically prove that adversaries lacking complete ownership of the file, cannot successfully pass ownership verification with non-negligible probability. Additionally, we show that adversaries cannot gain any knowledge about the file through the ownership verification process. We implement our scheme in two real cloud scenarios and analyze performance compared to the recent state-of-the-art schemes. The experimental results demonstrate that our approach incurs moderate computational, communication, storage, and energy overheads while achieving ownership authentication and maintaining obliviousness in deduplicated cloud storage systems.</div></div>","PeriodicalId":55132,"journal":{"name":"Future Generation Computer Systems-The International Journal of Escience","volume":"175 ","pages":"Article 108082"},"PeriodicalIF":6.2,"publicationDate":"2025-08-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144864304","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":"A dual-stream parallel architecture for robust visual tracking using scale-aware region proposals","authors":"Sudha SK,&nbsp;Aji S","doi":"10.1016/j.future.2025.108079","DOIUrl":"10.1016/j.future.2025.108079","url":null,"abstract":"<div><div>Visual tracking in dynamic environments faces significant challenges such as occlusions, scale variations, and abrupt motion changes, particularly in traffic scenarios. Tracking multi-scale objects and maintaining the temporal correlations across video sequences is essential for accurate tracking. These challenges motivated us to present a novel method that captures long-term dependencies in motion cues using a scale-aware region proposal (SARPro) network that uses a Faster R-CNN pipeline to predict high-quality region proposals for effective multi-scale video object detection and tracking (VODT). The proposed method uses robust feature extraction through a dual-stream feature pyramid network (DS-FPN) that captures spatial and temporal patterns. The SARPro generates precise bounding box proposals, addressing object scale variations. An iterative approach incorporating an LSTM fine-tunes the bounding boxes. A low-confidence track filter (LCTFilter) is integrated into the DeepSORT tracking algorithm to filter out the least confident tracks. The SARPro is designed to operate within multi-threaded parallel computing with GPU acceleration (MTPC-GPU) to optimize simultaneous detection and tracking. Experiments on benchmark datasets demonstrate that SARPro significantly enhances accuracy, achieving robust detection and tracking of small objects in complex video sequences while ensuring real-time performance. SARPro attains mAP scores of 91.35 % with 57.2 FPS on UA-DETRAC and 88.57 % with 41.9 FPS on BDD100K datasets.</div></div>","PeriodicalId":55132,"journal":{"name":"Future Generation Computer Systems-The International Journal of Escience","volume":"175 ","pages":"Article 108079"},"PeriodicalIF":6.2,"publicationDate":"2025-08-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144864308","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":"SoKV: Scan performance optimization for KV separation with adaptive dynamic grouping and GC-based LSM-tree management","authors":"Yixiang Cai ,&nbsp;Yubiao Pan ,&nbsp;Xinwei Lin ,&nbsp;Jie Xu ,&nbsp;Huizhen Zhang ,&nbsp;Mingwei Lin","doi":"10.1016/j.future.2025.108083","DOIUrl":"10.1016/j.future.2025.108083","url":null,"abstract":"<div><div>Key-value (KV) storage becomes a foundational technology for system software, enabling fast data processing and high-performance applications across various workloads and scenarios. In KV separation storage systems, where values are stored separately from the LSM-tree, scan operations necessitate traversal of the LSM-tree to retrieve addresses of desired values before accessing the corresponding KV pairs. Consequently, the organization of KV pairs and the size of the LSM-tree significantly impact scan performance. Recognizing this, we devised two strategies: Adaptive Dynamic Grouping and GC-based LSM-tree Management, to enhance scan performance by expediting the restoration of orderliness in frequently accessed KV pairs and reducing LSM-tree size. Finally, we implemented our prototype system called SoKV. Experimental results show that the scan throughput of SoKV is 2.66<span><math><mo>×</mo></math></span> that of RocksDB, 10.38<span><math><mo>×</mo></math></span> that of Parallax, 1.88<span><math><mo>×</mo></math></span> that of WiscKey, 2.27<span><math><mo>×</mo></math></span> that of HashKV and 1.39<span><math><mo>×</mo></math></span> that of FenceKV. Additionally, due to the reduction in the size of the LSM-tree, SoKV also outperforms the all other systems in terms of update performance.</div></div>","PeriodicalId":55132,"journal":{"name":"Future Generation Computer Systems-The International Journal of Escience","volume":"175 ","pages":"Article 108083"},"PeriodicalIF":6.2,"publicationDate":"2025-08-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144889186","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":"ReMEC: Reliability-aware scheduling of mixed-criticality IoT tasks in DVFS-enabled Multi-tier Edge Computing","authors":"Akhirul Islam ,&nbsp;Suchetana Chakraborty ,&nbsp;Manojit Ghose","doi":"10.1016/j.future.2025.108074","DOIUrl":"10.1016/j.future.2025.108074","url":null,"abstract":"<div><div>The Internet of Things (IoT) has witnessed significant growth, driving innovation across a wide range of application domains. Many IoT applications are characterized by their high resource demands and stringent latency requirements. Multi-tier edge computing has emerged, addressing these needs, where the application is scheduled across IoT devices, edge servers, and the cloud. However, ensuring reliable application performance remains a key challenge, particularly in transient IoT device failures caused by electromagnetic interference, battery depletion, hardware failures, or software crashes. In this work, we consider task execution reliability by incorporating failure of the user device, while the previous work primarily focuses on server-side reliability and overlooks user-centric limitations. We also include the user budget constraint while enhancing the task execution reliability by task replication. Additionally, we consider mixed criticality tasks in our application model, reflecting real-world scenarios more accurately, an aspect largely overlooked in existing works. To achieve task execution reliability while ensuring user budget and task latency deadline, we introduce ReMEC, a fuzzy logic-based reliable hybrid task offloading framework that relies on a distributed message queuing strategy to preserve execution state during device failures, and a fixed-point iterative method for optimizing DVFS frequencies to improve energy efficiency without violating task deadlines or compromising reliability. Our comprehensive benchmarking, which rigorously compares ReMEC against two state-of-the-art strategies (RMEAC and FP-TOSM) and three baseline approaches (BR-greedy, LE-greedy, and Random-RR), demonstrates that ReMEC outperforms all of them, achieving average improvements of 26.19 % in latency, 31.49 % in energy consumption, and 72.16 % in application failure rate, thereby demonstrating its practical applicability in real-world IoT scenarios.</div></div>","PeriodicalId":55132,"journal":{"name":"Future Generation Computer Systems-The International Journal of Escience","volume":"175 ","pages":"Article 108074"},"PeriodicalIF":6.2,"publicationDate":"2025-08-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144880009","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":"Increasing the reliability of citizen science campaign data for deforestation detection in tropical forests","authors":"Hugo Resende ,&nbsp;Álvaro L. Fazenda ,&nbsp;Fábio A.M. Cappabianco ,&nbsp;Fabio A. Faria","doi":"10.1016/j.future.2025.108081","DOIUrl":"10.1016/j.future.2025.108081","url":null,"abstract":"<div><div>In recent years, citizen science (CS) campaigns leveraging crowdsourcing have proven effective in generating large datasets across various fields such as environmental monitoring, and astronomy. However, the quality of volunteer-contributed data remains a challenge, as inconsistent responses often arise from inattentiveness and rapid analyses. To increase reliability in the generation of labeled datasets in citizen science campaigns, this paper proposes the combination of outlier detection techniques (Z-Score, Tukey and Median Absolute Deviation) to remove unreliable voluntary contributions, followed by exclusion of tasks with high Shannon entropy, that is, without consensus of volunteers. To validate this methodology, a case study was conducted using three CS campaigns from the ForestEyes project, which employs citizen science and machine learning to detect deforested areas. The results showed that applying those statistical techniques to filter contributions based on response time of the volunteers joining with median entropy filter led to a growth of up to 20 % of accuracy in campaigns, highlighting the importance of integrating statistical techniques and variability to improve the CS results.</div></div>","PeriodicalId":55132,"journal":{"name":"Future Generation Computer Systems-The International Journal of Escience","volume":"175 ","pages":"Article 108081"},"PeriodicalIF":6.2,"publicationDate":"2025-08-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144889185","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":"Digital twin platform for real-time data communication in UAV environment","authors":"Changhui Bae ,&nbsp;Euteum Choi ,&nbsp;Sungjoo Kang ,&nbsp;Sungsoo Ahn ,&nbsp;Seongjin Lee","doi":"10.1016/j.future.2025.108078","DOIUrl":"10.1016/j.future.2025.108078","url":null,"abstract":"<div><div>Real-time data communication is essential for controlling objects from virtual space to physical space through digital twins. However, existing digital twin platforms for UAV environments primarily focus on data modeling, prediction, and simulation rather than real-time performance and have not been extensively evaluated for real-time data communication, which may limit their applicability in real-world UAV operations. This paper introduces the RC-DT(Real-Time Communication Digital Twin)-Platform, which supports real-time data communication in UAV environments. The RC-DT-Platform’s data communication performance was evaluated by measuring the throughput as bandwidth, and the number of registered items increased. Results show that the RC-DT-Platform can transmit approximately 454 data/sec for 100 bytes data, 119 data/sec for 100 KB data, and 0.7 data/sec for 16 MB of data. Additionally, with 32 registered objects, the RC-DT-Platform can achieve a read throughput of about 3500 data/sec, regardless of data size. The performance of pure Ditto degrades by up to approximately 10 times as the number of registered objects increases up to 32, whereas the RC-DT-Platform maintains a degradation of less than 6.25 times. Thus, the RC-DT-Platform meets the required real-time data communication performance by considering flight speed, data size, and data generation rate.</div></div>","PeriodicalId":55132,"journal":{"name":"Future Generation Computer Systems-The International Journal of Escience","volume":"175 ","pages":"Article 108078"},"PeriodicalIF":6.2,"publicationDate":"2025-08-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144913287","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":"FLaTEC: An efficient federated learning scheme across the Thing-Edge-Cloud environment","authors":"Van An Le,&nbsp;Jason Haga,&nbsp;Yusuke Tanimura,&nbsp;Truong Thao Nguyen","doi":"10.1016/j.future.2025.108073","DOIUrl":"10.1016/j.future.2025.108073","url":null,"abstract":"<div><div>Federated Learning (FL) has become a cornerstone for enabling decentralized model training in mobile edge computing and Internet of Things (IoT) environments and maintaining data privacy by keeping data local to devices. However, the exponential growth in the number of participating devices and the increasing size and complexity of Machine Learning (ML) models amplify FL’s challenges, including high communication overhead, significant computational and energy constraints on edge devices, and the issue of heterogeneous data distribution, i.e., non-Independent and Identically Distributed (non-IID) data across clients. To address these challenges, we propose FLaTEC, a novel FL system tailored for the Thing-Edge-Cloud (TEC) architecture. First, FLaTEC introduces a split-training architecture that divides the global model into three components: a lightweight base model trained on resource-constrained edge devices, a computationally intensive core model trained on edge servers, and a simplified core model designed for on-device training and inference. Second, FLaTEC adopts a separate training strategy in which feature data is uploaded periodically from devices to edge servers to train the core model, reducing frequent data exchanges and mitigating the non-IID problem in FL. Third, to enhance the performance of the simplified model used for on-device training, FLaTEC applies knowledge distillation from the core model trained at the edge. A cloud server orchestrates the entire system by aggregating the base, core, and simplified core models using a federated averaging algorithm, ensuring consistency and coordination across devices and edge servers. Extensive experiments conducted across multiple datasets and diverse ML tasks validate FLaTEC’s superior performance, demonstrating its ability to achieve high accuracy, reduced communication overhead, and resilience to data heterogeneity compared to state-of-the-art methods.</div></div>","PeriodicalId":55132,"journal":{"name":"Future Generation Computer Systems-The International Journal of Escience","volume":"175 ","pages":"Article 108073"},"PeriodicalIF":6.2,"publicationDate":"2025-08-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144879929","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":"TLHAC: Three-level hierarchical architecture of the controller of the software-defined industrial production network","authors":"Jin Chen ,&nbsp;Ziyang Guo ,&nbsp;Liang Tan ,&nbsp;Kun She","doi":"10.1016/j.future.2025.108076","DOIUrl":"10.1016/j.future.2025.108076","url":null,"abstract":"<div><div>As a key component of the industrial intranet, the production network is the source of data generation and the object of intelligent decision-making. Therefore, it is very important for the management and control of the production network. Currently, software-defined network, as one of the key technologies to break the “two-level and three-level” networking model of factory intranet, provides a centralized control and programmable network management capability for the production network. However, as the number of sensor devices in the production network continues to increase, the current single controller deployed at the industrial intranet router may encounter control latency, single points of failure, and uneven load. For this reason, this paper proposes a three-level hierarchical architecture for Software-Defined Network(SDN) controllers in industrial production networks called TLHAC. TLHAC consists of three levels of hierarchy, with the first level being the primary controller deployed on the router of the production network backbone, the second level being the secondary controllers deployed on the edge gateways of the workshop network, and the third level being the sub-controllers deployed on the wireless sensor nodes in the field. When a secondary controller fails, a control latency optimal migration algorithm based on load capacity limitation called LCL_CDOM is proposed to migrate industrial equipment. In addition, to optimize the deployment of sub-controllers, this paper also proposes a sub-controller deployment strategy based on node importance. The strategy first uses the Technique for Order Preference by Similarity to Ideal Solution(TOPSIS) analysis based on multi-attribute decision-making to comprehensively evaluate the importance of wireless sensor nodes, then uses the improved fuzzy multi-objective particle swarm algorithm (called IFMOBPSO) to optimize the solution and select the optimal deployment position of the sub-controller. This paper conducts simulation experiments on the three-level hierarchical deployment architecture and the optimal deployment strategy of the sub-controller. Simulation results demonstrate that TLHAC reduces the average control latency by 42 %-48 % and the average synchronization latency by 19 %-22 % compared to traditional two-level and Edge-SDN architectures. While IFMOBPSO achieves 8 %-14 % lower average control latency of important nodes and than 9 %-12 % lower average synchronization latency between secondary controllers compare to other meta-heuristic algorithms.</div></div>","PeriodicalId":55132,"journal":{"name":"Future Generation Computer Systems-The International Journal of Escience","volume":"175 ","pages":"Article 108076"},"PeriodicalIF":6.2,"publicationDate":"2025-08-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144885605","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}