Journal of Network and Computer Applications最新文献

{"title":"Enhancement and optimization of FlexE technology within metro transport networks","authors":"Mu Liang ,&nbsp;Chen Zhang ,&nbsp;Tao Huang","doi":"10.1016/j.jnca.2025.104365","DOIUrl":"10.1016/j.jnca.2025.104365","url":null,"abstract":"<div><div>Flexible Ethernet (FlexE) technology represents a groundbreaking solution for addressing diverse service requirements and network slicing demands in 5G networks, enabling high-bandwidth, low-latency, and efficient multi-service transmission. However, the current FlexE technology suffers from inefficient bandwidth adjustment, primarily due to its slow overhead insertion mechanism, particularly evident in metro transport networks (MTNs). This inefficiency not only prolongs service reconfiguration time but also leads to significant bandwidth resource wastage along end-to-end network paths. Furthermore, the latency overhead configuration necessitates substantial buffer capacity at network nodes to store pending data, imposing considerable storage pressure on network equipment. In this study, we propose an innovative overhead frame insertion mechanism that addresses these critical limitations while maintaining full compliance with FlexE standards. The proposed method features a streamlined overhead block structure that enables simultaneous and continuous transmission of all overhead information, significantly accelerating service-to-timeslot mapping and reducing link establishment time. Moreover, the proposed mechanism seamlessly integrates with the alignment marker insertion in Physical Coding Sublayer (PCS) and maintains full compatibility with IEEE 802.3 standard, simplifying overhead block extraction and data processing at the receiving end. Simulation results demonstrate that compared to existing FlexE technology, our solution achieves up to a 20-fold improvement in bandwidth adjustment time while substantially reducing buffer requirements and optimizing bandwidth utilization across the entire network infrastructure.</div></div>","PeriodicalId":54784,"journal":{"name":"Journal of Network and Computer Applications","volume":"244 ","pages":"Article 104365"},"PeriodicalIF":8.0,"publicationDate":"2025-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145364131","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":"Autoformer-based mobility and handoff-aware prediction for QoE enhancement in adaptive video streaming in 4G/5G networks","authors":"Maram Helmy ,&nbsp;Mohamed S. Hassan ,&nbsp;Mahmoud H. Ismail ,&nbsp;Usman Tariq","doi":"10.1016/j.jnca.2025.104324","DOIUrl":"10.1016/j.jnca.2025.104324","url":null,"abstract":"<div><div>Traditional Adaptive Bitrate (ABR) algorithms in Dynamic Adaptive Streaming over HTTP (DASH) rely on basic throughput estimation techniques that often struggle to quickly adapt to network fluctuations. As users move across different transportation modes or change from one access point to another (e.g., Wi-Fi to cellular networks or between 4G/5G cells), available bandwidth can vary sharply, causing interruptions, abrupt quality shifts, which impact the ability of conventional ABR algorithms to provide seamless playback and maintain high quality-of-experience (QoE). To address these issues, this paper introduces a novel and comprehensive framework that significantly enhances the adaptability and intelligence of ABR algorithms. The proposed solution integrates three key components: a transformer-based throughput prediction model, a Mobility-Aware Throughput Prediction engine (MATH-P), and a Handoff-Aware Throughput Prediction engine (HATH-P). The transformer-based model outperforms state-of-the-art approaches in predicting throughput for both 4G and 5G networks, leveraging its ability to capture complex temporal patterns and long-term dependencies. The MATH-P engine adapts throughput predictions to varying mobility scenarios, while the HATH-P one manages seamless transitions by accurately predicting 4G/5G handoff events and selecting the appropriate throughput prediction model. The proposed systems were integrated into existing ABR algorithms, replacing traditional throughput estimation techniques. Experimental results demonstrate that the MATH-P and HATH-P engines significantly improve video streaming performance, reducing stall durations, enhancing video quality, and ensuring smoother playback.</div></div>","PeriodicalId":54784,"journal":{"name":"Journal of Network and Computer Applications","volume":"243 ","pages":"Article 104324"},"PeriodicalIF":8.0,"publicationDate":"2025-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145121249","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":"ARMBoost+: Empowering stacking, ensemble, and boosting models for network intrusion detection with dynamic rule repository","authors":"Vullikanti Vivek ,&nbsp;Bharadwaj Veeravalli","doi":"10.1016/j.jnca.2025.104292","DOIUrl":"10.1016/j.jnca.2025.104292","url":null,"abstract":"<div><div>As network security threats become increasingly complex, the need for efficient and effective network intrusion detection systems (NIDS) is more important than ever. Machine learning (ML) has emerged as a promising solution for NIDS due to its ability to analyze large volumes of network traffic data and detect suspicious patterns. In this paper, we propose ARMBoost+ a novel integrated approach for NIDS using dynamic rule repository building with a combination of stacking, ensemble, and boosting ML models, and associative rule mining (ARM) and bloom filter techniques. ARMBoost+ approach involves generating frequent feature sets using ARM and building a feature repository using bloom filter to avoid duplicate patterns. We then use the feature repository to train the ML models, which are tested on live network traffic data to generate dynamic rules for the rule repository. The live traffic data allowed us to assess the performance and robustness of our NIDS under dynamic and unpredictable network scenarios. The dynamic rule repository is continuously updated with new attack patterns, ensuring that the NIDS is always up-to-date with the latest security threats. To evaluate the effectiveness of ARMBoost+, we conducted experiments using a publicly available datasets and compared the results to existing NIDS approaches. We tested our approach under various scenarios, including simulating ML models without ARM and without automated feature dropping, and using ARM and bloom filter. We employed several ML models, including Stacking Classifier (with logistic regression (LR), random forest (RF), and support vector machine (SVM)), Ensemble with SVM, AdaBoost with Decision Tree, Gradient Boosting, and XGBoosting. Our experimental results demonstrate that the proposed novel ARMBoost+ integrated approach outperforms existing NIDS approaches in terms of accuracy and detection rates. The combination of stacking, ensemble, and boosting ML models, along with ARM and bloom filter, proved to be highly effective in detecting network intrusions. The dynamic rule repository building approach allowed for continuous updating of the NIDS with the latest attack patterns, resulting in improved performance over time. Furthermore, ARMBoost+ approach showed robustness against various types of attacks, including denial-of-service (DoS) and port scanning attacks. We also observed that the inclusion of ARM and bloom filter resulted notable reduction in the False Positive Rate (FPR) by around 4.07% and improved the efficiency of the feature repository.</div></div>","PeriodicalId":54784,"journal":{"name":"Journal of Network and Computer Applications","volume":"243 ","pages":"Article 104292"},"PeriodicalIF":8.0,"publicationDate":"2025-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144921477","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":"Enhanced energy efficiency in UAV-assisted Mobile Edge Computing through improved hybrid nature-inspired algorithm for task offloading","authors":"Hengyu Li,&nbsp;Hongjian Li","doi":"10.1016/j.jnca.2025.104290","DOIUrl":"10.1016/j.jnca.2025.104290","url":null,"abstract":"<div><div>With the rise of Mobile Edge Computing (MEC) and the increasing number of User Equipments (UE), traditional MEC systems struggle with high UE density. UAVs can assist in offloading tasks from base stations, but their limited resources make deployment and offloading strategies critical. This paper investigates UAV deployment strategies and task offloading policies in scenarios where UE density varies over time. First, we introduce the Maximum Clique in Weighted Graph (MCWG) algorithm, which is designed to calculate the UAV deployment coordinates within a weighted graph. In the task offloading phase, considering the resource constraints of UAVs, we develop a collaborative computation offloading framework involving UE, UAVs, MEC, and cloud servers. Secondly, we propose an Improved Hybrid Nature-Inspired Optimization (IHNIO) algorithm under delay and energy consumption constraints. This algorithm aims to minimize the average delay and energy consumption of UEs in UAV-assisted Mobile Edge Computing. Simulation results show that our approach significantly enhances energy efficiency compared to baseline solutions, with potential improvements in efficiency reaching up to 11%.</div></div>","PeriodicalId":54784,"journal":{"name":"Journal of Network and Computer Applications","volume":"243 ","pages":"Article 104290"},"PeriodicalIF":8.0,"publicationDate":"2025-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144867455","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":"From cloud to edge: dynamic placement optimization of business processes in IIoT networks","authors":"Md Razon Hossain ,&nbsp;Alistair Barros ,&nbsp;Colin Fidge","doi":"10.1016/j.jnca.2025.104317","DOIUrl":"10.1016/j.jnca.2025.104317","url":null,"abstract":"<div><div>Breakthroughs in edge computing offer new prospects for businesses to extend Industrial Internet of Things (IIoT) networks beyond analytics to actionable processing. In particular, cloud-based business processes, which provide administrative actions and rules through workflow-sequenced activities, can be streamlined on the edge for low-latency access in physical spaces. Although this advances business controls, particularly for critical events of industrial applications, it faces operational barriers. Edge devices, which support high volume and competing demands from a large number of sensors, vary in capacity, reliability, and proximity to sensors and cloud gateways. This warrants a highly efficient placement of process activities, from cloud to edge, given a variety of constraints, including resource demand, capacity, and compatibility, to satisfy timeliness constraints. In contrast to the related IIoT optimization research underway, including those of singleton service placements, business processes pose new challenges. Not only do sets of dependent activities have to be considered for co-deployment, but the meaning of timing constraints needs to be respected, given alternative, parallel, and iterative control-flow paths in processes. In addition, instantiation (replication) to scale activities for increasing data volumes poses further deployment constraints, i.e., on sets of nodes supporting dynamic instantiation of order-dependent activities. Here we present an optimization strategy for business processes that addresses these challenges. We first conceptualize processes in coherent fragments to precisely derive both responsiveness and throughput execution time heuristics and formulate a multi-objective process placement problem. Next, we develop a genetic algorithm-based process placement procedure. To adapt to fluctuating event frequencies, we support an interplay between scaling algorithms for service instances and process placement optimization. Validation through an industrial safety monitoring use case drawn from the construction industry shows that our approach improves timeliness responses by almost one-third and more than doubles execution throughput compared to existing methods.</div></div>","PeriodicalId":54784,"journal":{"name":"Journal of Network and Computer Applications","volume":"243 ","pages":"Article 104317"},"PeriodicalIF":8.0,"publicationDate":"2025-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145009289","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":"Energy-efficient optimal relay design for wireless sensor network in underground mines","authors":"Md Zahangir Alam ,&nbsp;Mohamed Lassaad Ammari ,&nbsp;Abbas Jamalipour ,&nbsp;Paul Fortier","doi":"10.1016/j.jnca.2025.104303","DOIUrl":"10.1016/j.jnca.2025.104303","url":null,"abstract":"<div><div>The transceiver design for multi-hop multiple-input multiple-output (MIMO) relay is very challenging, and for a large scale network, it is not economical to send the signal through all possible links. Instead, we can find the best path from source-to-destination that gives the highest end-to-end signal-to-noise ratio (SNR). In this paper, we provide a linear minimum mean squared error (MMSE) based multi-hop multi-terminal MIMO non-regenerative half-duplex amplify-and-forward (AF) parallel relay design for a wireless sensor network (WSN) in an underground mines. The transceiver design of such a network becomes very complex. We can simplify a complex multi-terminal parallel relay system into a series of links using selection relaying, where transmission from the source to the relay, relay to relay, and finally relay to the destination will take place using the best relay that provides the best link performance among others. The best relay selection using the traditional technique in our case is not easy, and we need a strategy to find the best path from a large number of hidden paths. We first find the set of simplified series multi-hop MIMO best relays from source to destination using the optimum path selection technique found in the literature. Then we develop a joint optimum design of the source precoder, the relay amplifier, and the receiver matrices using the full channel diagonalizing technique followed by the Lagrange strong duality principle with known channel state information (CSI). Finally, simulation results show an excellent agreement with numerical analysis demonstrating the effectiveness of the proposed framework.</div></div>","PeriodicalId":54784,"journal":{"name":"Journal of Network and Computer Applications","volume":"243 ","pages":"Article 104303"},"PeriodicalIF":8.0,"publicationDate":"2025-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145049249","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 NDN caching: Leveraging dynamic behaviour for enhanced efficiency","authors":"Matta Krishna Kumari ,&nbsp;Nikhil Tripathi","doi":"10.1016/j.jnca.2025.104288","DOIUrl":"10.1016/j.jnca.2025.104288","url":null,"abstract":"<div><div>The TCP/IP architecture has been the backbone of the Internet for decades, but its host-centric design is becoming less suitable for the data-centric communication demands of today. As the demand for efficient content distribution and retrieval grows, Named Data Networking (NDN) emerges as a promising alternative. NDN shifts the focus from host-centric to data-centric networking, with packets routed based on content names rather than IP addresses. A key feature of NDN is in-network caching, which attempts to reduce latency, alleviate network congestion and enhance content availability. However, the known NDN caching schemes do not consider the dynamic content demand that changes with respect to time and location. This causes the end users to encounter relatively higher content access latency. To address this challenge, in this paper, we propose a novel dynamic behaviour strategy that can be integrated into the known NDN caching schemes. This strategy can enable the NDN routers to make cooperative decisions and move the content copy to an edge router that requests the content most frequently. We comprehensively evaluate the performance of state-of-the-art NDN caching schemes with and without our proposed dynamic strategy using several real-world topologies. Our experimental results show that incorporating dynamic behaviour into these schemes leads to significantly better outcomes in terms of CHR, content latency, and path stretch. Specifically, the best improvements include a threefold increase in CHR, an 80% reduction in content access latency, and nearly a 45% decrease in path stretch. As an aside, we also develop a framework for the Icarus simulator to automate the process of performance assessment of different NDN caching schemes on a large number of real-world topologies.</div></div>","PeriodicalId":54784,"journal":{"name":"Journal of Network and Computer Applications","volume":"243 ","pages":"Article 104288"},"PeriodicalIF":8.0,"publicationDate":"2025-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144889351","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":"PRAETOR:Packet flow graph and dynamic spatio-temporal graph neural network-based flow table overflow attack detection method","authors":"Kaixi Wang ,&nbsp;Yunhe Cui ,&nbsp;Guowei Shen ,&nbsp;Chun Guo ,&nbsp;Yi Chen ,&nbsp;Qing Qian","doi":"10.1016/j.jnca.2025.104333","DOIUrl":"10.1016/j.jnca.2025.104333","url":null,"abstract":"<div><div>The flow table overflow attack on SDN switches is considered to be a destructive attack in SDN. By exhausting the computing and storage resources of SDN switches, this attack severely disrupts the normal communication functions of SDN networks. Graph neural networks are now being employed to detect flow table overflow attacks in SDN. When a flow graph is constructed, flow features are commonly utilized as nodes to represent the characteristics of flow table overflow attacks. However, a graph solely relying on these nodes and attributes may not fully encompass all the nuances of the flow table overflow attack. Additionally, GNN model may be difficult in capturing the graph information between different flow graphs over time, thus decreasing the detection accuracy of packet flow graph. To address these issues, we introduce PRAETOR, a detection method for flow table overflow attacks that leverages a packet flow graph and a dynamic spatio-temporal graph neural network. More particularly, The PaFlo-Graph algorithm and the EGST model are introduced by PRAETOR. The PaFlo-Graph algorithm generates a packet flow graph for each flow. It utilizes packet information to construct the graph with more detail, thereby better reflecting the characteristics of flow table overflow attacks. The EGST model is a dynamic spatio-temporal graph convolutional network designed to detect flow table overflow attacks by analyzing packet flow graphs. Experiments were conducted under two network topologies, where we used tcpreplay to replay packets from the bigFlow dataset to simulate SDN network flow. We also employed sFlow to sample packet features. Based on the sampled data, two datasets were constructed, each containing 1,760 network flows. For each packet, eight key features were extracted to represent its characteristics. The evaluation metrics include TPR, TNR, accuracy, precision, recall, F1-score, confusion matrix, ROC curves, and PR curves. Experimental results show that the proposed PaFlo-Graph algorithm generates more detailed flow graphs compared to KNN and CRAM, resulting in an average improvement of 6.49% in accuracy and 8.7% in precision. Furthermore, the overall detection framework, PRAETOR, achieves detection accuracies of 99.66% and 99.44% on Topo1 and Topo2, respectively. The precision scores reach 99.32% and 99.72%, and the F1-scores are 99.57% and 100%, respectively, indicating superior detection performance compared to other methods.</div></div>","PeriodicalId":54784,"journal":{"name":"Journal of Network and Computer Applications","volume":"243 ","pages":"Article 104333"},"PeriodicalIF":8.0,"publicationDate":"2025-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145121251","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":"ST-MemA: Leveraging Swin Transformer and memory-enhanced LSTM for encrypted traffic classification","authors":"Zhiyuan Li ,&nbsp;Yujie Jin","doi":"10.1016/j.jnca.2025.104329","DOIUrl":"10.1016/j.jnca.2025.104329","url":null,"abstract":"<div><div>Traffic classification is essential for effective intrusion detection and network management. However, with the pervasive use of encryption technologies, traditional machine learning-based and deep learning-based methods often fall short in capturing the fine-grained details in encrypted traffic. To address these limitations, we propose a memory-enhanced LSTM model based on Swin Transformer for multi-class encrypted traffic classification. Our approach first reconstructs raw encrypted traffic by converting each flow into single-channel images. A hierarchical attention network, incorporating both byte-level and packet-level attention, then performs comprehensive feature extraction on these traffic images. The resulting feature maps are subsequently classified to identify traffic flow categories. By combining the long-term dependency capabilities of LSTM with the Swin Transformer’s strengths in feature extraction, our model effectively captures global features across diverse traffic types. Furthermore, we enhance LSTM with memory attention, enabling the model to focus on more fine-grained information. Experimental results on three public datasets—USTC-TFC2016, ISCX-VPN2016, and CIC-IoT2022 show that our model, ST-MemA, improves the classification accuracy to 99.43%, 98.96% and 98.21% and <span><math><msub><mrow><mi>F</mi></mrow><mrow><mn>1</mn></mrow></msub></math></span>-score to 0.9936, 0.9826 and 0.9746, respectively. The results also demonstrate that our proposed model outperforms current state-of-the-art models in classification accuracy and computational efficiency.</div></div>","PeriodicalId":54784,"journal":{"name":"Journal of Network and Computer Applications","volume":"243 ","pages":"Article 104329"},"PeriodicalIF":8.0,"publicationDate":"2025-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145094148","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":"MTRC: A self-supervised network intrusion detection framework based on multiple Transformers enabled data reconstruction with contrastive learning","authors":"Yufeng Wang ,&nbsp;Hao Xu ,&nbsp;Jianhua Ma ,&nbsp;Qun jin","doi":"10.1016/j.jnca.2025.104300","DOIUrl":"10.1016/j.jnca.2025.104300","url":null,"abstract":"<div><div>Nowadays, Network Intrusion Detection System (NIDS) is essential for identifying and mitigating network threats in increasingly complex and dynamic network environments. Due to the benefits of automatic feature extraction and powerful expressive capability, Deep Neural Networks (DNN) based NIDS has witnessed great deployment. Considering the extremely high annotation cost, i.e., the extreme difficulty of labeling anomalous samples in supervised DNN based NIDS schemes, practically, many NIDS schemes are unsupervised. which either use generative-based approaches, such as encoder-decoder structure to identify deviated samples without the labeled intrusion data, or employ discriminative-based methods by designing pretext tasks to construct additional supervisory signals from the given data. However, the former only generates a single reconstruction version for each input sample, lacking a holistic view of the latent distribution of input sample, while the latter focuses on learning the global perspective of samples, often neglecting internal structures. To address these issues, this paper proposes a novel self-supervised NIDS framework based on multiple Transformers enabled data reconstruction with contrastive learning, MTRC, through combining generative-based and discriminative-based paradigms. In detail, our paper's contributions are threefold. First, a cross-feature correlation module is proposed to convert each tabular network traffic record into an original data view that effectively captures the cross-feature correlations. Second, inspired by the idea of the multiple-view reconstruction and contrastive learning, multiple Encoder-Decoder structured Transformers are used to generate different views for each original data view, which intentionally make each reconstructed view semantically similar to the original data view, and while these reconstructed views diversified between each other, aiming to holistically capture the latent features of normal data samples. Experimental results on multiple real network traffic datasets demonstrate that MTRC outperforms state-of-the-art unsupervised and self-supervised NIDS schemes, achieving superior performance in terms of AUC-ROC, AUC-PR, and F1-score metrics. The MTRC source code is publicly available at: <span><span>https://github.com/sunyifen/MTRC</span><svg><path></path></svg></span>.</div></div>","PeriodicalId":54784,"journal":{"name":"Journal of Network and Computer Applications","volume":"243 ","pages":"Article 104300"},"PeriodicalIF":8.0,"publicationDate":"2025-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144908813","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}