[1]马甜甜,洪 征,陈 乾.基于改进 AE-CM 模型的未知应用层协议识别[J].计算机技术与发展,2024,34(03):118-124.[doi:10. 3969 / j. issn. 1673-629X. 2024. 03. 018]
　MA Tian-tian,HONG Zheng,CHEN Qian.Unknown Application Layer Protocol Recognition Method Based on Improved AE-CM[J].,2024,34(03):118-124.[doi:10. 3969 / j. issn. 1673-629X. 2024. 03. 018]

点击复制

基于改进 AE-CM 模型的未知应用层协议识别()

分享到：

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

卷:

34

期数:

2024年03期

页码:

118-124

栏目:

网络空间安全

出版日期:

2024-03-10

文章信息/Info

Title:

Unknown Application Layer Protocol Recognition Method Based on Improved AE-CM

文章编号:

1673-629X(2024)03-0118-07

作者:

马甜甜; 洪 征; 陈 乾

陆军工程大学 指挥控制工程学院,江苏 南京 210014

Author(s):

MA Tian-tian; HONG Zheng; CHEN Qian

School of Command and Control Engineering,Army Engineering University of PLA,Nanjing 210014,China

关键词:

网络流量; 未知协议识别; 深度自编码器; 高斯混合聚类; 嵌入层; 邻居特征加权

Keywords:

network traffic; unknown protocol recognition; deep autoencoder; GMM clustering; embedding; neighbor feature weighting

分类号:

TP393

DOI:

10. 3969 / j. issn. 1673-629X. 2024. 03. 018

摘要:

现有的未知协议识别方法存在提取的特征不够充分、聚类分配不准确等问题,影响了协议识别结果的准确性。 AE-CM( deep autoencoder with embedding clustering module)解决了当前深度聚类模型异步优化的问题,提高了聚类分配的精度。 该文提出的 DAEC-NM 协议识别模型以 AE-CM 为基础,通过加入高维卷积、时序卷积网络以及调整多层感知机结构的方法,改进了 AE-CM 的特征提取部分。 为了更全面地获取协议信息,DAEC-NM 通过邻居分支采集邻居样本,并分析邻居样本间的局部关联特征,从而增强原样本特征中重要特征对聚类分配的指导能力。 最后,采用了注意力机制来分析特征的重要性,以此为聚类模块设置有效的初始权重,解决了聚类模块在模型更新过程中权重特征更新较慢的问题。 实验结果表明,DAEC-NM 能够有效提高未知协议识别的准确性。

Abstract:

Existing unknown protocol recognition methods suffer from insufficient feature extraction ability and inaccurate clustering assignments,which affect the accuracy of recognition results. AE-CM( deep autoencoder with embedding clustering module) addresses theissue of asynchronous optimization in deep clustering models and improves the accuracy?

of clustering assignments. The proposed DAEC-NM is based on the AE-CM. The feature extraction part of the AE-CM is improved by introducing high-dimensional convolution,　temporal convolution network, and adjusting the structure of multi - layer perceptron. To obtain more comprehensive protocolinformation,DAEC-NM collects neighbor samples through the neighbor model and analyzes the local correlation features to ensure theaccuracy of clustering results. Finally, we use an attention mechanism to capture the importance of features, and set effective initialweights for the clustering module to resolve the slow update problem in the clustering module. Experimental results show the DAEC-NMcan effectively improve the accuracy of unknown protocol recognition.

常用功能