Concurrency and Computation-Practice & Experience最新文献

{"title":"Explainable Graph Neural Networks for Optimized Allocation in Spatio-Temporal Bike Sharing Demand Prediction","authors":"Anand Polamarasetti","doi":"10.1002/cpe.70202","DOIUrl":"https://doi.org/10.1002/cpe.70202","url":null,"abstract":"<div>\u0000 \u0000 <p>This research proposes a novel approach to improving the allocation strategies in spatio-temporal bike sharing demand prediction by enhancing Explainable Graph Neural Networks (X-GNNs). Traditional models often struggle to capture the intricate relationships between spatially distributed bike stations and their temporal demand patterns. In response to this challenge, our proposed framework employs Graph Neural Networks (GNNs) to model the complex interactions within the bike sharing network. Additionally, we optimize interpretability and decision-making by incorporating explainability mechanisms into the model. The X-GNN architecture integrates attention mechanisms and graph attention networks to effectively capture spatial dependencies and temporal dynamics. This not only enables accurate prediction of short-term demand but also provides a clear understanding of the factors influencing the predictions. The attention mechanisms allow the model to focus on crucial nodes and temporal patterns, offering insights into the spatial and temporal features that contribute most significantly to the demand fluctuations. The explainability aspect of the model facilitates transparency in decision-making processes related to resource allocation and station management. Importantly, the explainability aspect of the model facilitates transparency in decision-making processes related to resource allocation and station management. By revealing the underlying drivers of demand—such as weather conditions, peak usage hours, or specific high-demand locations—decision-makers can deploy bikes more efficiently, plan for demand surges, and design proactive redistribution strategies. This bridges the gap between advanced AI models and practical, real-world applications in urban mobility systems. To validate the effectiveness of our proposed framework, extensive experiments were conducted on spatio-temporal bike sharing datasets. The dataset contains weather information, the number of bikes rented per hour, and date information. The results demonstrate superior prediction accuracy compared to baseline models, along with optimized interpretability.</p>\u0000 </div>","PeriodicalId":55214,"journal":{"name":"Concurrency and Computation-Practice & Experience","volume":"37 21-22","pages":""},"PeriodicalIF":1.5,"publicationDate":"2025-07-31","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144751735","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"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":"Optimal Control of Information Dissemination Based on Model-Free Reinforcement Learning","authors":"Dan Xia,&nbsp;Enyu Xu,&nbsp;Yuan Tian,&nbsp;Qiusha Min","doi":"10.1002/cpe.70213","DOIUrl":"https://doi.org/10.1002/cpe.70213","url":null,"abstract":"<div>\u0000 \u0000 <p>With the proliferation of mobile social networks, controlling information dissemination faces challenges in dynamic environments where state transition models are often unavailable. To address the optimal control problem under such model-free constraints, this paper aims to minimize the cumulative cost of epidemic-based information dissemination by innovatively integrating temporal difference (TD) learning with real-time capacity adaptation. The proposed method dynamically identifies optimal control signal timing, eliminating dependency on predefined state matrices. Experimental results demonstrate a 7.0% reduction in cumulative network cost and a 52.6% improvement in s-controllability compared to dynamic programming baselines. This model-free framework not only enhances robustness in time-varying networks but also offers scalability for large-scale applications, advancing real-time control in social media analysis and public opinion management.</p>\u0000 </div>","PeriodicalId":55214,"journal":{"name":"Concurrency and Computation-Practice & Experience","volume":"37 21-22","pages":""},"PeriodicalIF":1.5,"publicationDate":"2025-07-31","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144751731","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"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":"Toward a Reinforcement Learning Approach for Balancing Risk and Cost in Cloud-Based IoT-Driven Business Processes","authors":"Amina Ahmed Nacer,&nbsp;Mohammed Riyadh Abdmeziem","doi":"10.1002/cpe.70212","DOIUrl":"https://doi.org/10.1002/cpe.70212","url":null,"abstract":"<div>\u0000 \u0000 <p>Secure and cost-efficient deployment of Internet of Things (IoT)-driven business processes (BPs) in multicloud environments is a complex task, particularly under dynamic workloads and strict risk or budgetary constraints. We address this challenge by leveraging deep reinforcement learning (DRL) to determine optimal task-to-cloud allocations that comply with user-defined cost or risk thresholds. Our approach introduces a <i>Q</i>-learning model that integrates a confidentiality risk metric and a cost evaluation function into its learning process. Unlike traditional heuristics, the DRL agent adapts to evolving constraints through trial-and-error interaction with its environment. Experimental results on a diverse set of deployment configurations show that our approach achieves 25%–30% reductions in both risk and cost compared to heuristic baselines, while satisfying thresholds in 75% of cases. It also demonstrates strong adaptability to dynamic changes, including task additions and resource fluctuations. These findings highlight the potential of reinforcement learning for reliable, constraint-aware deployment in cloud-based IoT systems.</p>\u0000 </div>","PeriodicalId":55214,"journal":{"name":"Concurrency and Computation-Practice & Experience","volume":"37 21-22","pages":""},"PeriodicalIF":1.5,"publicationDate":"2025-07-31","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144751352","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"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":"Multi-Attribute Many-to-One Elderly Care Service Matching Group Decision-Making Considering Double Intermediaries","authors":"Qi Yue,&nbsp;He Huang,&nbsp;Yuan Tao","doi":"10.1002/cpe.70210","DOIUrl":"https://doi.org/10.1002/cpe.70210","url":null,"abstract":"<div>\u0000 \u0000 <p>Considering the rare application of picture fuzzy numbers in bilateral matching group decision-making, an elderly care service matching group decision-making approach based on TOPSIS and the group consistency is proposed. First, the problem of multi-attribute many-to-one elderly care service matching group decision-making considering double intermediaries in a picture fuzzy environment is described. Second, the comprehensive objective evaluation matrix is obtained by aggregating multiple objective evaluation matrices of online platforms as the first intermediary. According to picture fuzzy subjective evaluation matrices of the elderly and the comprehensive objective evaluation matrix of online platforms, integrative evaluation matrices of the elderly side are obtained. Similarly, according to picture fuzzy subjective evaluation matrices of elderly care institutions and the objective evaluation matrix of the community as the second intermediary, integrative evaluation matrices of the elderly care institutions side are obtained. Third, satisfactions under each attribute are calculated according to TOPSIS, and attribute weights are calculated by the improved BWM. On this basis, global satisfactions are obtained. Furthermore, a one-to-one stable matching model is constructed by introducing virtual subjects. Hence, the best elderly care service matching scheme is obtained by model solution. Finally, an example is given to verify the feasibility and effectiveness of the proposed approach.</p>\u0000 </div>","PeriodicalId":55214,"journal":{"name":"Concurrency and Computation-Practice & Experience","volume":"37 21-22","pages":""},"PeriodicalIF":1.5,"publicationDate":"2025-07-31","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144751359","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"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 Study on Meta-Heuristic-Based Web Service Composition Schemes in Cloud Computing Environments: Classification, Advantages, and Limitations","authors":"Mohammad Ali Nezafat Tabalvandani,&nbsp;Mirsaeid Hosseini Shirvani","doi":"10.1002/cpe.70175","DOIUrl":"https://doi.org/10.1002/cpe.70175","url":null,"abstract":"<div>\u0000 \u0000 <p>Cloud computing offers various services, with web services distinguished by their non-functional attributes like implementation cost, reliability, and availability, which affect solution quality. Combining services to achieve optimal quality is an NP-hard problem, best tackled with meta-heuristic algorithms. A review of the literature identified 61 papers from authentic and well-reputed publications addressing web service composition in cloud environments via such algorithms. This review categorizes these algorithms, evaluated implementation methods, techniques, criteria, and limitations, and compares their effectiveness. Furthermore, it outlines future research directions to enhance solutions for the web service composition challenge in cloud settings.</p>\u0000 </div>","PeriodicalId":55214,"journal":{"name":"Concurrency and Computation-Practice & Experience","volume":"37 21-22","pages":""},"PeriodicalIF":1.5,"publicationDate":"2025-07-31","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144751737","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"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 Comparative Analysis on the Services Offered by Amazon Web Services and Microsoft Azure","authors":"Sara Hameed,&nbsp;Syed Hashim Raza Bukhari,&nbsp;Syeda Umm E Abiha Rizvi,&nbsp;Meerab Tahir,&nbsp;Farhan Aadil","doi":"10.1002/cpe.70215","DOIUrl":"https://doi.org/10.1002/cpe.70215","url":null,"abstract":"<p>Different small or large-scale enterprises came to the conclusion that moving data to the cloud is more convenient. In recent decades, AWS and Azure have tried their best to provide comparatively affordable and instant solutions. Both cloud platforms are extensive to understand, so it is difficult for users to make the right choice because of data confidentiality, better storage capacity, and, most importantly, to select a reasonable platform. This paper is an in-depth analysis of the comparable capabilities and services offered by Azure and AWS. The strengths and weaknesses of both platforms are highlighted considering security, storage, pricing, and machine learning services. It critically examines the implementation of the linear regression model and its performance. Furthermore, a survey was conducted to ensure that the users know which cloud service provider is preferred. On the basis of machine learning services, people with more experience in the cloud preferred Azure Machine Learning Studio over Amazon SageMaker. Azure outperformed AWS, achieving an accuracy of 70%, whereas AWS managed only 60%. Similarly, in cases of security, storage, and pricing, Azure was preferred because of its flexible, easy-to-use services. Therefore, from the methodology, it is concluded that Azure was preferred over AWS. However, the right choice can only be made by considering the business needs.</p>","PeriodicalId":55214,"journal":{"name":"Concurrency and Computation-Practice & Experience","volume":"37 21-22","pages":""},"PeriodicalIF":1.5,"publicationDate":"2025-07-31","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://onlinelibrary.wiley.com/doi/epdf/10.1002/cpe.70215","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144751732","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"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}

{"title":"Hybrid Quantum Key Distribution Framework: Integrating BB84, B92, E91, and GHZ Protocols for Enhanced Cryptographic Security","authors":"Kaushik Dehingia,&nbsp;Nimisha Dutta","doi":"10.1002/cpe.70221","DOIUrl":"https://doi.org/10.1002/cpe.70221","url":null,"abstract":"&lt;div&gt;\u0000 \u0000 &lt;p&gt;The rapid evolution of quantum computing poses a significant threat to classical cryptographic systems like Rivest-Shamir-Adleman (RSA) and Elliptic Curve Cryptography (ECC), which rely on the computational hardness of problems such as integer factorization and discrete logarithms. Quantum algorithms such as Shor's algorithm can solve these problems quickly, which undermines the foundations on which classical cryptography relies. Quantum Key Distribution (QKD) is an alternative to the classical methods, which promises information-theoretic security based on quantum mechanics. Currently, there are existing QKD protocols, including BB84, B92, E91, and GHZ, all of which exhibit various real-world limitations. For example, these QKD protocols can be vulnerable to a variety of side-channel attacks (e.g., detector blinding, photon-number-splitting) and neglect to consider fluctuating network conditions. Current QKD protocols also fail to accommodate scalability for many-to-many or noise-limited scenarios. Many implementations of the existing QKD protocols and other common forms of networks remain static, relying on arbitrary decisions of fixed values that yield simple linear conclusions that can be predicted and targeted in the real-world environment. To address these omissions, we propose a new framework for dynamic or adaptive hybrid QKD in which we incorporate BB84, B92, E91, and GHZ into one common approach with all protocols selected based on a probability-weighted distribution of (0.3, 0.2, 0.3, 0.2). In the hybrid QKD implementation, the probability weights of protocol selection are assigned with partiality toward BB84, E91, B92, and GHZ, respectively. This may also introduce a higher variety of protocols and diversity in approaches that will further limit cross-protocol possibilities of attack vectors, while increasing the possible flexibility of adaptability in attacked situations. In addition, we incorporate an artificial intelligence (AI)-based optimization module using a neural network to evaluate local environmental noise and quantum bit error rate (QBER) in real time. It adjusts protocol selection probabilities dynamically based on both historical and live operational data to optimize throughput while maintaining low error rates. The system architecture supports modular and parallel operation and has been mapped out and designed to be scalable and compatible with future quantum networks. We test our system using IBM's Qiskit AerSimulator utilizing a 14-qubit register with 100 rounds of a 1% depolarization noise model, which significantly outperformed static hybrids such as Chen et al. in terms of both key rate and QBER. Our system consistently produced an average QBER of 0.02 and a key generation rate of 12 bits per round. E91 consistently produced CHSH violating &lt;span&gt;&lt;/span&gt;&lt;math&gt;\u0000 &lt;semantics&gt;\u0000 &lt;mrow&gt;\u0000 &lt;mi&gt;S&lt;/mi&gt;\u0000 &lt;mo&gt;=&lt;/mo&gt;\u0000 &lt;mn&gt;2&lt;/m","PeriodicalId":55214,"journal":{"name":"Concurrency and Computation-Practice & Experience","volume":"37 21-22","pages":""},"PeriodicalIF":1.5,"publicationDate":"2025-07-31","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144751734","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"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":"Predictive VNF Deployment With Virtual Network Mapping Using SDN/NFV-Enabled UAV Swarms","authors":"Qizhao Zhou,&nbsp;Zhongyu Shi","doi":"10.1002/cpe.70229","DOIUrl":"https://doi.org/10.1002/cpe.70229","url":null,"abstract":"<div>\u0000 \u0000 <p>Unmanned Aerial Vehicle (UAV) networks are emerging as pivotal enablers for supporting Network Function Virtualization (NFV) and Software-Defined Networking (SDN) services, particularly in meeting the diverse and stringent virtual network function (VNF) scheduling demands of future communication networks. However, a fundamental challenge arises from the SDN controller's inability to synchronize resource request information from VNFs in real time, potentially causing significant delays in mapping and scheduling strategies, especially for delay-sensitive UAV network services. To address this challenge, this paper introduces a predictive VNF deployment model, seamlessly integrated with virtual network mapping, designed to operate within constraints such as the ordered sequence of VNFs, delay requirements, and service arrival time. In recognition of the dynamic nature of UAV services, our framework incorporates VNF live migration and rescheduling. Consequently, we formulate the VNF mapping and scheduling challenge as a predictive long-term lateral resource optimization problem, leveraging Long Short-Term Memory (LSTM) techniques. By employing digital twin (DT)-based virtual network mapping, the SDN controller gains precise insights into the UAV's VNF resource demands, thereby effectively addressing service acceptance issues within VNF mapping and scheduling policies. Our simulation resultsdemonstrate that the proposed method achieves superior outcomes in terms of total benefit, network service acceptance rate, and average delay within the digital twin system. This approach not only enhances the operational efficiency of UAV networks but also ensures robust and timely service delivery in complex network environments, thereby contributing to the advancement of UAV-based NFV and SDN services.</p>\u0000 </div>","PeriodicalId":55214,"journal":{"name":"Concurrency and Computation-Practice & Experience","volume":"37 21-22","pages":""},"PeriodicalIF":1.5,"publicationDate":"2025-07-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144740147","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"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":"Enhancing the Harris Hawks Optimization Algorithm With Ambush-Based Operators for Feature Selection in UAV-Based Intrusion Detection Systems","authors":"Sayed Zabihullah Musawi,&nbsp;Mohammad Farshi,&nbsp;Sepehr Ebrahimi Mood,&nbsp;Alireza Souri","doi":"10.1002/cpe.70207","DOIUrl":"https://doi.org/10.1002/cpe.70207","url":null,"abstract":"<div>\u0000 \u0000 <p>Autonomous vehicles (AVs), including drones, rely on sensors, machine learning algorithms, and large datasets for perception, decision-making, and control. However, the high dimensionality of these datasets increases computational load and hampers real-time performance. In Unmanned Aerial Vehicle (UAV) systems, feature selection is critical for reducing complexity and enhancing processing efficiency, thereby enabling faster and more accurate decision-making. In this study, we enhance the Harris Hawks Optimization (HHO) algorithm by introducing a novel ambush-based operator to regulate selection pressure, resulting in an improved variant named AMHHO. The effectiveness of AMHHO is validated using IEEE CEC2019 benchmark functions and compared against several well-known optimization algorithms. To further evaluate its robustness, ablation studies and sensitivity analyses are conducted to identify the most efficient AMHHO variants. Furthermore, a binary version of AMHHO (BAMHHO) is applied to ten high-dimensional datasets and the UAV-IDS-2020 dataset for feature selection and classification tasks. BAMHHO is assessed based on classification accuracy, fitness value, feature selection ratio, and computation time, demonstrating superior performance across multiple datasets and outperforming state-of-the-art methods. To rigorously evaluate the statistical significance of its results, Wilcoxon Signed-Rank test is applied to compare BAMHHO with other well-known algorithms, confirming the statistical superiority of BAMHHO. In conclusion, BAMHHO not only achieves effective performance on high-dimensional datasets but also achieves 100% classification accuracy on the UAV-IDS-2020 dataset, all while maintaining an optimal balance between feature reduction and computational efficiency. These findings confirm BAMHHO's effectiveness in handling high-dimensional data and highlight its potential for application in UAV-based intrusion detection systems.</p>\u0000 </div>","PeriodicalId":55214,"journal":{"name":"Concurrency and Computation-Practice & Experience","volume":"37 21-22","pages":""},"PeriodicalIF":1.5,"publicationDate":"2025-07-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144740432","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"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":"Profiling and Optimization of Multicard GPU Machine Learning Jobs","authors":"Marcin Lawenda,&nbsp;Kyrylo Khloponin,&nbsp;Krzesimir Samborski,&nbsp;Łukasz Szustak","doi":"10.1002/cpe.70196","DOIUrl":"https://doi.org/10.1002/cpe.70196","url":null,"abstract":"<div>\u0000 \u0000 <p>The article discusses various model optimization techniques, providing a comprehensive analysis of key performance indicators. Several parallelization strategies for image recognition are analyzed, adapted to different hardware and software configurations, including distributed data parallelism and distributed hardware processing. Changing the tensor layout in PyTorch DataLoader from NCHW to NHWC and enabling <i>pin</i>_<i>memory</i> has proven to be very beneficial and easy to implement. Furthermore, the impact of different performance techniques (DPO, LoRA, QLoRA, and QAT) on the tuning process of LLMs was investigated. LoRA allows for faster tuning, while requiring less VRAM compared to DPO. On the other hand, QAT is the most resource-intensive method, with the slowest processing times. A significant portion of LLM tuning time is attributed to initializing new kernels and synchronizing multiple threads when memory operations are not dominant.</p>\u0000 </div>","PeriodicalId":55214,"journal":{"name":"Concurrency and Computation-Practice & Experience","volume":"37 18-20","pages":""},"PeriodicalIF":1.5,"publicationDate":"2025-07-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144681077","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}