Title: Detection of redundant traffic in large-scale communication networks based on logistic regression

Authors: Xin Wen; Liyu Huang; Yin Zheng; Hailin Zhao

Addresses: Guangzhou Power Supply Bureau, Guangdong Power Grid Co., Ltd., Guangzhou, 510730, China ' Guangzhou Power Supply Bureau, Guangdong Power Grid Co., Ltd., Guangzhou, 510730, China ' Guangzhou Power Supply Bureau, Guangdong Power Grid Co., Ltd., Guangzhou, 510730, China ' Guangzhou Ji Neng Information Technology Co., Ltd., Guangzhou, 510670, China

Abstract: In order to improve the traffic precision of network redundant traffic detection methods and reduce the time consumption of traffic classification, this paper proposes a large-scale redundant traffic detection method based on logical regression. Firstly, the logical regression architecture is analysed, and a feature extractor is constructed to extract redundant traffic features. Secondly, the weight matrix of the linear transformation between layers to be trained is obtained. Then, Gini coefficient is selected to determine the dispersion degree of redundant traffic, and redundant traffic classification function is constructed. Redundant traffic detection results are obtained through logical regression algorithm to complete network redundant traffic detection. The results show that the traffic classification time of this method is 53 ms; the precision rate is as high as 99%, which shows that the network redundant traffic detection method in this paper is effective.

Keywords: logical regression; Gini coefficient; loss function; softmax function; redundant flow detection.

DOI: 10.1504/IJRIS.2024.137468

International Journal of Reasoning-based Intelligent Systems, 2024 Vol.16 No.1, pp.8 - 15

Received: 21 Sep 2022
Accepted: 07 Nov 2022
Published online: 19 Mar 2024 *