Journal of Network and Computer Applications最新文献

{"title":"PE-Chain: An efficient hierarchical sharding architecture for dynamically participating node models","authors":"Junfeng Tian ,&nbsp;Yujian Wu ,&nbsp;Jin Tian ,&nbsp;Liuling Qi","doi":"10.1016/j.jnca.2025.104415","DOIUrl":"10.1016/j.jnca.2025.104415","url":null,"abstract":"<div><div>Sharding is a widely adopted technique for enhancing blockchain scalability by partitioning the network into multiple sub-shards, thereby distributing the workload. However, traditional dynamic sharding schemes often suffer from delayed adjustments to the shard count, hindering rapid convergence to an optimal workload distribution in practical deployments and limiting overall scalability. Furthermore, the dynamic participation of nodes is frequently overlooked. To address these challenges, this paper proposes PolyembryonyChain (PE-Chain), a novel and efficient hierarchical sharding architecture designed for environments with dynamically participating nodes, with the goal of achieving elastic scalability. Its core innovations are twofold. First, it introduces a dynamic threshold sharding algorithm that adaptively adjusts the number of shards and optimizes the network topology to enable elastic expansion. Second, it incorporates a validator assignment and reconfiguration scheme, specifically tailored for dynamic hierarchical environments, to ensure system security. A comprehensive security analysis and simulation results demonstrate that PE-Chain significantly outperforms state-of-the-art baselines, achieving approximately 28% higher throughput while maintaining low latency, which underscores its superior scalability and practical value.</div></div>","PeriodicalId":54784,"journal":{"name":"Journal of Network and Computer Applications","volume":"246 ","pages":"Article 104415"},"PeriodicalIF":8.0,"publicationDate":"2026-02-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145785051","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":"ARProof: A cross-protocol approach to detect and mitigate ARP-spoofing attacks in smart home networks","authors":"Md Mizanur Rahman ,&nbsp;Faycal Bouhafs ,&nbsp;Sayed Amir Hoseini ,&nbsp;Frank den Hartog","doi":"10.1016/j.jnca.2025.104396","DOIUrl":"10.1016/j.jnca.2025.104396","url":null,"abstract":"<div><div>Smart homes are increasingly vulnerable to cyberattacks that lead to network instability, causing homeowners to lodge complaints with their Broadband Service Providers (BSPs). Therefore, effective and timely detection of cyberattacks is crucial for both customers and BSPs. Address Resolution Protocol (ARP) spoofing is one of the most common attacks that can facilitate larger and more severe follow-up attacks. Unfortunately, there are currently no methods that can effectively detect and mitigate ARP spoofing in smart homes from a BSP’s perspective. Current Machine Learning (ML)-based methods often rely on a single dataset from a controlled lab environment designed to mimic a single home, assuming that the results will generalize to all smart homes. Our findings indicate that this assumption is flawed. They are also unsuitable for smart homes from a BSP’s perspective, as they require custom applications, introduce additional overhead, and often rely on the injection of probing traffic into the network. To address these issues, we developed an algorithm that can detect ARP spoofing in smart home networks, regardless of the network structure or connected devices. It uses a cross-protocol strategy by correlating ARP packets with Dynamic Host Configuration Protocol (DHCP) messages to validate address bindings. We evaluated our method using four public datasets and two real-world testbeds, achieving 100% detection accuracy in all scenarios. In addition, the algorithm requires only little computational overhead, confirming its suitability for use by BSPs to detect and mitigate ARP spoofing attacks in smart homes.</div></div>","PeriodicalId":54784,"journal":{"name":"Journal of Network and Computer Applications","volume":"246 ","pages":"Article 104396"},"PeriodicalIF":8.0,"publicationDate":"2026-02-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145593077","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":"SELB: Self-Evolution Load Balancing Based on Temporal Graph Convolutional Network in Software-Defined Data Center Networks","authors":"Yong Liu ,&nbsp;Guisheng Liu ,&nbsp;Tianyi Yu ,&nbsp;Qian Meng","doi":"10.1016/j.jnca.2025.104401","DOIUrl":"10.1016/j.jnca.2025.104401","url":null,"abstract":"<div><div>Software-Defined Networking (SDN) is a network architecture that separates the control plane and data plane of the traditional data center network, resulting in enhanced network scalability and flexibility. The conventional Equal Cost MultiPath (ECMP) load balancing algorithm, which relies on static hash mapping, has limitations when applied to data center networks, leading to issues such as hash conflicts and congestion between mouse and elephant flows. Therefore, load balancing based on flowlet granularity has been proposed. This approach divides flows into flowlets, leveraging the burstiness of traffic to enhance load balancing capabilities. However, these approaches encounter several challenges, such as the lack of real-time feedback on network load situations, the inability of static flowlet timeouts to adapt to dynamic changes in the network, and inadequate consideration of load distribution. To address these challenges, we propose a novel load balancing strategy called Self-Evolution Load Balancing (SELB) based on Temporal Graph Convolutional Network (T-GCN). SELB utilizes the T-GCN to dynamically predict the network load state for real-time feedback. Meanwhile, the adaptive flow splitting algorithm is employed to dynamically adjust the timeout of flowlets, effectively adapting to changes in network dynamics. Moreover, SELB incorporates a load-aware route planning strategy that considers the overall network load distribution. By doing so, it can intelligently route flowlets along equivalent multipaths, enhancing load balancing capabilities. The simulation results demonstrate that SELB effectively reduces Flow Completion Time (FCT), enhances average throughput, and improves load balancing performance in comparison to existing schemes.</div></div>","PeriodicalId":54784,"journal":{"name":"Journal of Network and Computer Applications","volume":"246 ","pages":"Article 104401"},"PeriodicalIF":8.0,"publicationDate":"2026-02-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145657738","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":"Towards IT/OT integration in industry digitalization: A comprehensive survey","authors":"Riccardo Venanzi,&nbsp;Giuseppe Di Modica,&nbsp;Luca Foschini,&nbsp;Paolo Bellavista","doi":"10.1016/j.jnca.2025.104373","DOIUrl":"10.1016/j.jnca.2025.104373","url":null,"abstract":"<div><div>According to both academic and industry perspectives, the Fourth Industrial Revolution has brought about a paradigm shift in the manufacturing sector enabling companies to enhance their competitiveness in the global market. To achieve this goal, manufacturing companies will need to undertake a deep digital transformation, primarily by introducing advanced Information Technology (IT) into traditionally less digitalized departments, such as shop floors, where Operational Technology (OT) currently dominate. For the full achievement of Industry 4.0 revolution objectives, practitioners believe in the strong requirement of a progressive and tight integration between IT and OT departments. In the depicted scenario, communication technologies are expected to play a pivotal role in facilitating the integration process, but other more recent and advanced IT have also proven helpful. In particular, the topic of IT/OT integration has attracted significant attention from various research communities that have sought to identify both the opportunities and challenges associated with its implementation. Although some good surveys of those works have appeared in the literature, to the best of our knowledge, no comprehensive review has yet been conducted that is fully dedicated to the topic of IT/OT convergence. In this paper, we propose a holistic approach to examine the various dimensions of IT/OT integration, which we classify into five interconnected realms, Communication, IT-Driven Support to OT, Human Centricity, Advanced Industrial Control Systems, and cybersecurity. Furthermore, we develop a realm-oriented taxonomy to organize the surveyed works in a structured manner, offering readers a clear overview of the current state of the literature, along with insights into unexplored opportunities and future directions for IT/OT integration.</div></div>","PeriodicalId":54784,"journal":{"name":"Journal of Network and Computer Applications","volume":"245 ","pages":"Article 104373"},"PeriodicalIF":8.0,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145441548","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":"Dynamic searchable symmetric encryption with efficient conjunctive query and non-interactive real deletion","authors":"Zhengwei Ren ,&nbsp;Pei He ,&nbsp;Rongwei Yu ,&nbsp;Li Deng ,&nbsp;Yan Tong ,&nbsp;Shiwei Xu","doi":"10.1016/j.jnca.2025.104387","DOIUrl":"10.1016/j.jnca.2025.104387","url":null,"abstract":"<div><div>Dynamic searchable symmetric encryption (DSSE) enables users to perform update and search operations over encrypted data on cloud servers. However, many DSSE schemes are unable to efficiently perform conjunctive queries containing multiple keywords, limiting their search capabilities. Those DSSE schemes supporting conjunctive query fail to achieve real deletion, affecting the efficiencies of subsequent searches. In this paper, we propose a DSSE scheme supporting conjunctive query and non-interactive real deletion simultaneously. For a conjunctive query containing multiple keywords, we adjust the positions of these keywords so that the keyword contained by the least number of document(s) is at the forefront of the conjunctive query. The document(s) containing this keyword are then located, and on the basis of the document(s) the remaining keywords are checked to obtain the final search result. Moreover, cuckoo filter is adopted to store the ciphertext to be searched, making the conjunctive query efficient. We deploy two search databases on the cloud server to achieve non-interactive real deletion. Benefiting from these two databases, the deleted ciphertext will be physically removed from the cloud server with no impact on subsequent searches, improving search efficiencies of subsequent searches. Our scheme only utilizes a few hash functions and a pseudorandom function, while the forward privacy and backward privacy are still achieved. We conduct a formal security analysis and extensive experimental evaluations, showing that our scheme has efficiency advantages in both update and search processes.</div></div>","PeriodicalId":54784,"journal":{"name":"Journal of Network and Computer Applications","volume":"245 ","pages":"Article 104387"},"PeriodicalIF":8.0,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145531185","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 quantum-secure digital signature-based communication protocol for the Internet of Drones (IoD)","authors":"Jithu Vijay V.P.,&nbsp;Shahanas I.N.,&nbsp;Sabu M. Thampi,&nbsp;Aiswarya S. Nair","doi":"10.1016/j.jnca.2025.104398","DOIUrl":"10.1016/j.jnca.2025.104398","url":null,"abstract":"<div><div>The use of drones is rapidly increasing in areas such as surveillance, defense, and emergency services. As a result, ensuring secure communication and proper authentication has become a major concern in the Internet of Drones, where drones must share data and coordinate their actions in real time. One of the biggest challenges in drone networks is maintaining secure and reliable communication between drones. The dynamic and distributed nature of these networks increases the risk of security breaches. Existing systems mostly rely on cryptographic methods like RSA and ECC. These methods will not remain secure in the future because of advancements in quantum computing. These systems also depend on static data storage and centralized credential management, which make them vulnerable to attacks such as impersonation, replay, and man-in-the-middle. To address these issues, we propose a quantum-secure drone-to-drone authentication and secure communication protocol that utilizes Post Quantum Cryptographic (PQC) algorithms such as, Kyber for encryption and Dilithium for digital signatures. Both are lattice-based lightweight cryptographic algorithms that offer strong resistance against quantum attacks. Instead of storing secret data on drones and to prevent cloning, we use Physical Unclonable Functions (PUF) to generate device specific seeds for authentication and key generation during each session. A Hyperledger Fabric Blockchain is used at the Ground Control Station (GCS) to store drone credentials securely and avoid single point failure. We conducted the formal security analysis using the Burrows–Abadi–Needham (BAN) logic for trust validation and the Scyther tool to formally analyze and verify resistance against classical and quantum-era attacks. In addition to formal proofs, informal analysis confirms that the protocol maintains data integrity and authentication even under active network threats. We implemented the protocol using Raspberry Pi drones and a Linux-based GCS. Performance results show a low computation time of 0.08 s for authentication and 0.12 s for secure communication on Raspberry Pi 5, with minimal memory usage and acceptable communication cost suitable for implementation on resource-constrained drones.</div></div>","PeriodicalId":54784,"journal":{"name":"Journal of Network and Computer Applications","volume":"245 ","pages":"Article 104398"},"PeriodicalIF":8.0,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145609218","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":"Edge-AI: A systematic review on architectures, applications, and challenges","authors":"Himanshu Gauttam ,&nbsp;Garima Nain ,&nbsp;K.K. Pattanaik ,&nbsp;Paulo Mendes","doi":"10.1016/j.jnca.2025.104375","DOIUrl":"10.1016/j.jnca.2025.104375","url":null,"abstract":"<div><div>The evolution of computing technologies and the generation of massive amounts of data fueled the development of <em>Artificial Intelligence</em> (AI), specifically <em>Deep Learning</em> (DL), solutions to extract key patterns from data, and the generation of insights and knowledge useful to achieve optimized service execution. Traditional cloud-based execution of DL solutions faces several challenges, such as latency, data privacy, and reliability, while trying to meet service requirements. In contrast, the limited computing and storage resources on the edge pose daunting challenges in executing resource-intensive DL solutions closer to the customer. This scenario led to the birth of an interdisciplinary research field named Edge-AI or Edge-Intelligence, aiming to mitigate the limitations of cloud and edge-based DL executions. In this context, this work proposes a reference layered Edge-AI framework to ensure the successful deployment of the Edge-Intelligence paradigm, encompassing three novel layers for the optimization of edge infrastructure, edge inference, and edge training. The work presents a detailed investigation and analysis of the schemes centered around the above-listed layers of the proposed Edge-AI framework. Furthermore, this work discusses potential application domains for Edge-AI, delving into a set of potential limitations, and ending up identifying future research directions in terms of Edge-AI infrastructure deployment, inference and training, which are functionalities needed to deploy and use robust, sustainable, and efficient intelligent edge networks.</div></div>","PeriodicalId":54784,"journal":{"name":"Journal of Network and Computer Applications","volume":"245 ","pages":"Article 104375"},"PeriodicalIF":8.0,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145461584","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 comprehensive study of the 6LoWSD protocol architecture with respect to scalability and mobility for SDN-enabled IoT networks","authors":"Wanbanker Khongbuh ,&nbsp;Goutam Saha","doi":"10.1016/j.jnca.2025.104370","DOIUrl":"10.1016/j.jnca.2025.104370","url":null,"abstract":"<div><div>The Internet of Things (IoT) and software-defined networks (SDN) have opened up new opportunities for innovation. Many of the limitations of the IoT system can be rectified with the SDN concepts. Thus, the combination of SDN and IoT has tremendous potential in various application domains. As the number of IoT devices is increasing with time, the scalability issues need to be further improved. Another significant challenge in IoT environments is mobility. Maintaining seamless mobility and persistent connectivity for IoT devices operating over large-scale or geographically dispersed environments presents a significant research challenge. But scalability and mobility are complex challenges. Developing scalable, mobile, and adaptive network architectures is crucial for SDN-enabled IoT ecosystems. Using SDN-enabled IoT networks, we introduced a comprehensive approach to address these challenges. Here, a new protocol based on OpenFlow of SDN and 6LoWPAN of the IoT system, namely, 6LoWSD has been proposed. In this investigation, emphasis has been placed on techniques on how the proposed 6LoWSD can improve scalability and mobility issues. In this study, experiments with the proposed protocol were performed using physical devices and a simulated platform. The results were compared with the 6LoWPAN counterpart and were found to be satisfactory.</div></div>","PeriodicalId":54784,"journal":{"name":"Journal of Network and Computer Applications","volume":"245 ","pages":"Article 104370"},"PeriodicalIF":8.0,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145461588","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":"Adaptive context-aware multi-tab website fingerprinting using hierarchical deep learning","authors":"Faisal Murad ,&nbsp;Jie Cui ,&nbsp;Muhammad Aurangzeb Khan ,&nbsp;Depeng Chen","doi":"10.1016/j.jnca.2025.104374","DOIUrl":"10.1016/j.jnca.2025.104374","url":null,"abstract":"&lt;div&gt;&lt;div&gt;Website fingerprinting aims to infer visited websites from encrypted network traffic. Conventional approaches predominantly assume single-tab browsing, limiting their applicability under realistic multi-tab conditions, where concurrent flows introduce inter-tab interference, temporal overlap, and attribution ambiguity that degrade classification accuracy. This paper presents Adaptive Context-Aware Multi-Tab Fingerprinting, a dynamic framework designed for multi-tab website fingerprinting through context-driven feature modeling and attention adaptation. ACMF integrates three coordinated modules. (1) CBAM employs an attention-augmented LSTM that processes sequences of packet direction, size, and inter-arrival time with tab-switch indicators, producing a session context vector &lt;span&gt;&lt;math&gt;&lt;mi&gt;c&lt;/mi&gt;&lt;/math&gt;&lt;/span&gt;. A self-attention state &lt;span&gt;&lt;math&gt;&lt;msub&gt;&lt;mrow&gt;&lt;mi&gt;z&lt;/mi&gt;&lt;/mrow&gt;&lt;mrow&gt;&lt;mi&gt;t&lt;/mi&gt;&lt;/mrow&gt;&lt;/msub&gt;&lt;/math&gt;&lt;/span&gt; modulates recurrent dynamics, and multi-scale temporal aggregation yields &lt;span&gt;&lt;math&gt;&lt;mrow&gt;&lt;mi&gt;c&lt;/mi&gt;&lt;mo&gt;=&lt;/mo&gt;&lt;msub&gt;&lt;mrow&gt;&lt;mo&gt;∑&lt;/mo&gt;&lt;/mrow&gt;&lt;mrow&gt;&lt;mi&gt;ℓ&lt;/mi&gt;&lt;/mrow&gt;&lt;/msub&gt;&lt;msub&gt;&lt;mrow&gt;&lt;mi&gt;ω&lt;/mi&gt;&lt;/mrow&gt;&lt;mrow&gt;&lt;mi&gt;ℓ&lt;/mi&gt;&lt;/mrow&gt;&lt;/msub&gt;&lt;msup&gt;&lt;mrow&gt;&lt;mi&gt;h&lt;/mi&gt;&lt;/mrow&gt;&lt;mrow&gt;&lt;mrow&gt;&lt;mo&gt;(&lt;/mo&gt;&lt;mi&gt;ℓ&lt;/mi&gt;&lt;mo&gt;)&lt;/mo&gt;&lt;/mrow&gt;&lt;/mrow&gt;&lt;/msup&gt;&lt;/mrow&gt;&lt;/math&gt;&lt;/span&gt;. (2) DTAM uses a Transformer encoder with per-slot gating for variable tab cardinality. For each slot feature &lt;span&gt;&lt;math&gt;&lt;msub&gt;&lt;mrow&gt;&lt;mi&gt;f&lt;/mi&gt;&lt;/mrow&gt;&lt;mrow&gt;&lt;mi&gt;i&lt;/mi&gt;&lt;/mrow&gt;&lt;/msub&gt;&lt;/math&gt;&lt;/span&gt;, a gate &lt;span&gt;&lt;math&gt;&lt;mrow&gt;&lt;msub&gt;&lt;mrow&gt;&lt;mi&gt;g&lt;/mi&gt;&lt;/mrow&gt;&lt;mrow&gt;&lt;mi&gt;i&lt;/mi&gt;&lt;/mrow&gt;&lt;/msub&gt;&lt;mo&gt;=&lt;/mo&gt;&lt;mi&gt;σ&lt;/mi&gt;&lt;mrow&gt;&lt;mo&gt;(&lt;/mo&gt;&lt;msubsup&gt;&lt;mrow&gt;&lt;mi&gt;w&lt;/mi&gt;&lt;/mrow&gt;&lt;mrow&gt;&lt;mi&gt;g&lt;/mi&gt;&lt;/mrow&gt;&lt;mrow&gt;&lt;mo&gt;⊤&lt;/mo&gt;&lt;/mrow&gt;&lt;/msubsup&gt;&lt;mrow&gt;&lt;mo&gt;[&lt;/mo&gt;&lt;mi&gt;c&lt;/mi&gt;&lt;mo&gt;;&lt;/mo&gt;&lt;msub&gt;&lt;mrow&gt;&lt;mi&gt;f&lt;/mi&gt;&lt;/mrow&gt;&lt;mrow&gt;&lt;mi&gt;i&lt;/mi&gt;&lt;/mrow&gt;&lt;/msub&gt;&lt;mo&gt;]&lt;/mo&gt;&lt;/mrow&gt;&lt;mo&gt;+&lt;/mo&gt;&lt;msub&gt;&lt;mrow&gt;&lt;mi&gt;b&lt;/mi&gt;&lt;/mrow&gt;&lt;mrow&gt;&lt;mi&gt;g&lt;/mi&gt;&lt;/mrow&gt;&lt;/msub&gt;&lt;mo&gt;)&lt;/mo&gt;&lt;/mrow&gt;&lt;/mrow&gt;&lt;/math&gt;&lt;/span&gt; scales multi-head attention outputs, normalized by &lt;span&gt;&lt;math&gt;&lt;mrow&gt;&lt;msub&gt;&lt;mrow&gt;&lt;mo&gt;∑&lt;/mo&gt;&lt;/mrow&gt;&lt;mrow&gt;&lt;mi&gt;i&lt;/mi&gt;&lt;/mrow&gt;&lt;/msub&gt;&lt;msub&gt;&lt;mrow&gt;&lt;mi&gt;g&lt;/mi&gt;&lt;/mrow&gt;&lt;mrow&gt;&lt;mi&gt;i&lt;/mi&gt;&lt;/mrow&gt;&lt;/msub&gt;&lt;mo&gt;+&lt;/mo&gt;&lt;mi&gt;ϵ&lt;/mi&gt;&lt;/mrow&gt;&lt;/math&gt;&lt;/span&gt; and followed by a position-wise feed-forward network to produce representation &lt;span&gt;&lt;math&gt;&lt;mi&gt;F&lt;/mi&gt;&lt;/math&gt;&lt;/span&gt;. (3) HMLFE combines dilated 1D CNNs to capture local temporal motifs with a GNN that builds a similarity graph using edge weights &lt;span&gt;&lt;math&gt;&lt;mrow&gt;&lt;mo&gt;exp&lt;/mo&gt;&lt;mrow&gt;&lt;mo&gt;(&lt;/mo&gt;&lt;mo&gt;−&lt;/mo&gt;&lt;mo&gt;‖&lt;/mo&gt;&lt;msub&gt;&lt;mrow&gt;&lt;mi&gt;u&lt;/mi&gt;&lt;/mrow&gt;&lt;mrow&gt;&lt;mi&gt;i&lt;/mi&gt;&lt;/mrow&gt;&lt;/msub&gt;&lt;mo&gt;−&lt;/mo&gt;&lt;msub&gt;&lt;mrow&gt;&lt;mi&gt;u&lt;/mi&gt;&lt;/mrow&gt;&lt;mrow&gt;&lt;mi&gt;j&lt;/mi&gt;&lt;/mrow&gt;&lt;/msub&gt;&lt;msup&gt;&lt;mrow&gt;&lt;mo&gt;‖&lt;/mo&gt;&lt;/mrow&gt;&lt;mrow&gt;&lt;mn&gt;2&lt;/mn&gt;&lt;/mrow&gt;&lt;/msup&gt;&lt;mo&gt;/&lt;/mo&gt;&lt;mi&gt;τ&lt;/mi&gt;&lt;mo&gt;)&lt;/mo&gt;&lt;/mrow&gt;&lt;/mrow&gt;&lt;/math&gt;&lt;/span&gt; and performs attention-based message passing. A graph readout operation generates &lt;span&gt;&lt;math&gt;&lt;msub&gt;&lt;mrow&gt;&lt;mi&gt;h&lt;/mi&gt;&lt;/mrow&gt;&lt;mrow&gt;&lt;mi&gt;G&lt;/mi&gt;&lt;/mrow&gt;&lt;/msub&gt;&lt;/math&gt;&lt;/span&gt;, and the final representation &lt;span&gt;&lt;math&gt;&lt;mrow&gt;&lt;mi&gt;r&lt;/mi&gt;&lt;mo&gt;=&lt;/mo&gt;&lt;mrow&gt;&lt;mo&gt;[&lt;/mo&gt;&lt;mi&gt;F&lt;/mi&gt;&lt;mo&gt;∥&lt;/mo&gt;&lt;msub&gt;&lt;mrow&gt;&lt;mi&gt;h&lt;/mi&gt;&lt;/mrow&gt;&lt;","PeriodicalId":54784,"journal":{"name":"Journal of Network and Computer Applications","volume":"245 ","pages":"Article 104374"},"PeriodicalIF":8.0,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145404578","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":"Task scheduling of cloud computing system by frilled lizard optimization with time varying expansion mixed function oscillation and horned lizard camouflage strategy","authors":"Hao-Ming Song,&nbsp;Si-Wen Zhang,&nbsp;Jie-Sheng Wang,&nbsp;Cheng Xing,&nbsp;Yu-Feng Sun,&nbsp;Yu-Cai Wang,&nbsp;Xiao-Fei Sui","doi":"10.1016/j.jnca.2025.104386","DOIUrl":"10.1016/j.jnca.2025.104386","url":null,"abstract":"<div><div>With the increasing complexity and scale of cloud computing systems, task scheduling optimization has become critical for improving resource utilization, enhancing service reliability, and reducing overall energy consumption. Traditional swarm intelligence algorithms often struggle to achieve an effective balance between global exploration and local exploitation, leading to premature convergence or sub-optimal solutions, particularly in large-scale and high-dimensional problem scenarios. To address these challenges, this study proposes a Time Varying Mixed Function Frilled Lizard Optimization algorithm (TMCFLO) that incorporates a horned lizard-inspired camouflage strategy to increase population diversity and prevent premature convergence, alongside a novel mixed function oscillation mechanism, combining sine, cosine, power, logarithm, and Gaussian functions, to enhance local search precision and convergence efficiency. A time-varying expansion factor is further introduced to dynamically regulate oscillation amplitude, ensuring adaptive adjustment of search behavior throughout the optimization process. Extensive evaluations on the CEC 2022 benchmark set demonstrate that TMCFLO outperforms classical algorithms, including PSO, ACO, WOA, AOA, POA, ZOA, HO, RLLPSO and IHBA, achieving up to 26 percent improvement in optimization accuracy. In practical cloud computing task scheduling experiments with 1500 and 3000 tasks, TMCFLO achieves the lowest single task energy consumption of 0.2196, the lowest total energy consumption of 658.80, and the highest energy efficiency of 4.5569, confirming its effectiveness, scalability, and energy-efficient superiority for complex cloud scheduling problems.</div></div>","PeriodicalId":54784,"journal":{"name":"Journal of Network and Computer Applications","volume":"245 ","pages":"Article 104386"},"PeriodicalIF":8.0,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145528948","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}