[1]张鑫雯,邹瑞,宋成,等.基于多尺度特征提取的网络入侵检测方法[J].计算机技术与发展,2025,(06):87-93.[doi:10.20165/j.cnki.ISSN1673-629X.2025.0053]
　ZHANG Xin-wen,ZOU Rui,SONG Cheng,et al.A Network Intrusion Detection Method Based on Multi-scale Feature Extraction[J].,2025,(06):87-93.[doi:10.20165/j.cnki.ISSN1673-629X.2025.0053]
点击复制

基于多尺度特征提取的网络入侵检测方法()

《计算机技术与发展》[ISSN:1006-6977/CN:61-1281/TN]

卷:
期数:
2025年06期
页码:
87-93
栏目:
网络空间安全
出版日期:
2025-06-10

文章信息/Info

Title:
A Network Intrusion Detection Method Based on Multi-scale Feature Extraction
文章编号:
1673-629X(2025)06-0087-07
作者:
张鑫雯; 邹瑞; 宋成; 李言
新疆工程学院 工程技能实训学院,新疆 乌鲁木齐 830000
Author(s):
ZHANG Xin-wen; ZOU Rui; SONG Cheng; LI Yan
School of Engineering Skills Training,Xinjiang Institute of Engineering,Urumqi 830000,China
关键词:
网络入侵; 多尺度特征; 时间位置向量; 趋势向量; 混合注意力
Keywords:
network intrusion; multi-scale features; time position vector; trend vector; mixed attention
分类号:
TP391
DOI:
10.20165/j.cnki.ISSN1673-629X.2025.0053
摘要:
网络流量异常通常预示着潜在的网络入侵,但由于流量数据的高维度和时间波动性,捕捉其多尺度特征具有较大挑战。 为解决这一问题,该文提出了一种基于多尺度特征提取的 MSFE-DDR 模型。 该模型采用分解与重建的架构,首先通过预处理生成时间位置向量、趋势向量和特征向量。 其中,时间位置向量用于编码流量数据中的多尺度时间信息,使模型能够识别不同时间尺度上的波动特征,从而更好地捕捉流量的时序变化。 趋势向量则通过加权平均提取流量的稳定变化趋势。 接着,分解模块结合混合注意力和多头注意力机制,增强了特征表示的能力,以便更精准地捕捉复杂的时间动态。 为适应流量在工作日、周末及节假日等不同时间段的变化,设计了趋势偏移模块,通过在设定的窗口范围内计算流量数据的趋势分量,并将其从特征向量中减去,能够动态调整趋势向量的变化方向,更加灵活地捕捉流量的长期变化趋势。最后,重建模块将这三种向量进行融合,并通过多尺度记忆残差网络将不同卷积核的卷积结果进行融合,从而提取数据的多尺度特征。 实验结果表明,MSFE-DDR 模型在公开数据集 UNSW-NB15 和 NSL-KDD 上取得了显著的性能提升,证明了其在网络流量异常检测中的有效性和优越性。
Abstract:
Network traffic anomalies often indicate potential network intrusions. However,due to the high dimensionality and temporal volatility of traffic data,capturing its multi-scale features presents a significant challenge. To address this issue,we propose a MSFE-DDR model based on multi-scale feature extraction. The model adopts a decomposition-and-reconstruction architecture,starting with preprocessing to generate time position vectors,trend vectors,and feature vectors. The time position vector encodes multi-scale temporal information within the traffic data,enabling the model to recognize fluctuations across different time scales and better capture temporal variations in the traffic. The trend vector,on the other hand,extracts the stable variation trends in the traffic through weighted averaging.Subsequently, the decomposition module combines hybrid attention and multi - head attention mechanisms to enhance feature representation, enabling more precise capture of complex temporal dynamics. To adapt to traffic variations across different time periods, such as weekdays,weekends,and holidays,a trend offset module is designed. This module calculates the trend component of the traffic data within a specified window and subtracts it from the feature vector,dynamically adjusting the direction of the trend vector’s variation.This allows the model to more flexibly capture long-term traffic variation trends. Finally,the reconstruction module integrates these three vectors and fuses the convolution results of different kernels through a multi-scale residual memory network,thereby extracting the multi-scale features of the data. Experimental results demonstrate that the proposed MSFE-DDR model achieves significant performance im-provements on the publicly available UNSW-NB15 and NSL-KDD datasets,proving its effectiveness and superiority in network traffic a-nomaly detection.

相似文献/References:

[1]王涛 张凤斌 李鑫鑫.基于人工免疫的记忆检测器研究[J].计算机技术与发展,2008,(06):148.
　WANG Tao,ZHANG Feng-bin,LI Xin-xin.Memory Detector Research Based on Artificial Immune Theory[J].,2008,(06):148.
[2]张铮 赵荣彩 单征.基于最大隶属度的网络安全模型[J].计算机技术与发展,2007,(06):141.
　ZHANG Zheng,ZHAO Rong-cai,SHAN Zheng.A Network Security Model Based on Maximum Degree of Membership[J].,2007,(06):141.
[3]朱 鹏,陈 虎*,李 科,等.一种轻量级的多尺度特征人脸检测方法[J].计算机技术与发展,2020,30(04):1.[doi:10. 3969 / j. issn. 1673-629X. 2020. 04. 001]
　ZHU Peng,CHEN Hu*,LI Ke,et al.A Face Detection Method with Lightweight and Multi-scale Feature[J].,2020,30(06):1.[doi:10. 3969 / j. issn. 1673-629X. 2020. 04. 001]
[4]殷 齐,丁 飞,朱 跃,等.基于 CNN 与多尺度特征融合的城市交通流预测模型[J].计算机技术与发展,2022,32(10):175.[doi:10. 3969 / j. issn. 1673-629X. 2022. 10. 029]
　YIN Qi,DING Fei,ZHU Yue,et al.An Urban Traffic Flow Prediction Model Based on CNN and Multi-scale Feature Fusion[J].,2022,32(06):175.[doi:10. 3969 / j. issn. 1673-629X. 2022. 10. 029]
[5]蔡俊敏,孙 涵.基于注意力机制和多尺度特征的伪装目标检测[J].计算机技术与发展,2023,33(08):131.[doi:10. 3969 / j. issn. 1673-629X. 2023. 08. 019]
　CAI Jun-min,SUN Han.Camouflaged Object Detection Based on Attention Mechanism andMulti-scale Features[J].,2023,33(06):131.[doi:10. 3969 / j. issn. 1673-629X. 2023. 08. 019]
[6]郭 锐,熊风光*,谢剑斌,等.基于改进残差池化层的纹理识别[J].计算机技术与发展,2023,33(09):37.[doi:10. 3969 / j. issn. 1673-629X. 2023. 09. 006]
　GUO Rui,XIONG Feng-guang*,XIE Jian-bin,et al.Texture Recognition Algorithm Based on Improved Deep Residual Pooling Layer[J].,2023,33(06):37.[doi:10. 3969 / j. issn. 1673-629X. 2023. 09. 006]
[7]王新武,陈春雨.基于 U 形多尺度注意力方法的真实图像去噪[J].计算机技术与发展,2024,34(04):48.[doi:10. 3969 / j. issn. 1673-629X. 2024. 04. 008]
　WANG Xin-wu,CHEN Chun-yu.Real-world Image Denoising Based on U-shaped Multi-scale Attention Method[J].,2024,34(06):48.[doi:10. 3969 / j. issn. 1673-629X. 2024. 04. 008]
[8]尹春勇,沈子宁.基于交互式特征与多尺度特征的文本相似度研究[J].计算机技术与发展,2024,34(08):86.[doi:10.20165/j.cnki.ISSN1673-629X.2024.0140]
　YIN Chun-yong,SHEN Zi-ning.Research on Text Similarity Based on Interactive Features and Multi-scale Features[J].,2024,34(06):86.[doi:10.20165/j.cnki.ISSN1673-629X.2024.0140]
[9]张国有,崔永强.基于双分支注意力机制的图像自动标注研究[J].计算机技术与发展,2024,34(09):167.[doi:10.20165/j.cnki.ISSN1673-629X.2024.0172]
　ZHANG Guo-you,CUI Yong-qiang.Research on Automatic Image Annotation Based on Dual-branch Attention Mechanism[J].,2024,34(06):167.[doi:10.20165/j.cnki.ISSN1673-629X.2024.0172]
[10]吴铁钰,杨光,邹丽.RSG-YOLO:用于检测道路坑洼的高效神经网络[J].计算机技术与发展,2025,(02):199.[doi:10.20165/j.cnki.ISSN1673-629X.2024.0330]
　WU Tie-yu,YANG Guang,ZOU Li.RSG-YOLO:An Efficient Neural Network for Road Pothole Detection[J].,2025,(06):199.[doi:10.20165/j.cnki.ISSN1673-629X.2024.0330]

更新日期/Last Update: 2025-06-10