Scholarlink

<p>Communication Service Classification (CSC) plays an important role in cyber security and network optimization, particularly in anomaly detection and enhancing service quality. CSC categorizes communication services into various categories based on users’ shared behaviors across different applications. The CSC differs from traditional network traffic classification (NTC) in that it encompasses a broader range of applications, and it can understand the types of services frequently used by users based on these classifications allows for more effective network management. Therefore, the dataset we use differs from publicly available datasets online. We utilized Wireshark software to capture traffic for seven categories of communication services within a campus network environment, thereby constructing our own communication service dataset. We extracted the packet level characteristics and timestamp information for constructing our own communication service dataset. For that, we propose an X-CLNN (extreme gradient boosting classification algorithm based on Convolutional Neural Networks (CNN) and Long Short-Term Memory (LSTM) feature extraction). This algorithm leverages CNN and LSTM to extract spatiotemporal features of network traffic and then uses an extreme gradient boosting tree classifier to categorize the extracted features. Experimental results show that the X-CLNN model achieved an accuracy of 94.8% on our dataset, demonstrating superior classification performance compared to alternative methods.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

X-CLNN: A Hybrid Neural Network-Based Approach Combining CNN and RNN for Communication Services Classification

  • 期刊论文

  • 发表日期：

    2025年6月16日

查看全文
  • Wenbo Xiao,
  • Chenchen Xu,
  • Min Peng,
  • Caihong Kai,
  • Tian Chen

摘要

Communication Service Classification (CSC) plays an important role in cyber security and network optimization, particularly in anomaly detection and enhancing service quality. CSC categorizes communication services into various categories based on users’ shared behaviors across different applications. The CSC differs from traditional network traffic classification (NTC) in that it encompasses a broader range of applications, and it can understand the types of services frequently used by users based on these classifications allows for more effective network management. Therefore, the dataset we use differs from publicly available datasets online. We utilized Wireshark software to capture traffic for seven categories of communication services within a campus network environment, thereby constructing our own communication service dataset. We extracted the packet level characteristics and timestamp information for constructing our own communication service dataset. For that, we propose an X-CLNN (extreme gradient boosting classification algorithm based on Convolutional Neural Networks (CNN) and Long Short-Term Memory (LSTM) feature extraction). This algorithm leverages CNN and LSTM to extract spatiotemporal features of network traffic and then uses an extreme gradient boosting tree classifier to categorize the extracted features. Experimental results show that the X-CLNN model achieved an accuracy of 94.8% on our dataset, demonstrating superior classification performance compared to alternative methods.