网络流量时延特征数据的识别方法仿真  

Identification Method and Simulation of Network Traffic Delay Characteristic Data

作　　者：周家恺 綦方中[1] ZHOU Jia-kai;QI Fang-zhong(Zhejiang University of Technology,Hangzhou Zhejiang 310023,China)

机构地区：[1]浙江工业大学,浙江杭州310023

出　　处：《计算机仿真》2022年第5期398-401,460,共5页Computer Simulation

摘　　要：当前的网络流量时延特征识别方法未能在特征识别过程提升流量梯度,导致识别出现较大偏差,且方法的响应时间较长。为此提出基于朴素贝叶斯的网络流量时延特征识别方法。利用移动蜂窝网络通信链路技术和无线资源控制机制造成的网络流量时延波动完成建模分析,同时结合往返时延计算结果,获取与数据时延相关的网络流量特征。通过特征描述得到不同网络节点接入互联网技术差异导致的时序分布。将极端梯度提升树模型和朴素贝叶斯相结合,构建分类器,完成网络流量时延特征的识别。仿真结果表明,所提方法能够获取高精度的网络流量时延特征识别结果,同时还能够有效缩短响应时间。Indeed,the traditional identification methods of network traffic delay characteristics have large identification deviation and long response time,being caused by the improvement of traffic gradient.In this regard,we report a network traffic delay identification method based on Naive Bayes.The network traffic delay fluctuation caused by mobile cellular network communication link technology and wireless resource control mechanism was combined to complete the modeling and analysis.Simultaneously,the round-trip delay calculation results were also combined to obtain the network traffic characteristics related to data delay.Based on the feature description,the timing distribution caused by different network nodes’ access to the Internet technology was obtained.Extreme gradient lifting tree model and naive Bayes were used to construct classifiers to identify the characteristics of network traffic delay.The simulation results show that this method has high-precision identification results of network traffic delay characteristics and short response time.

关 键 词：朴素贝叶斯 网络流量 时延特征识别 极端梯度提升树模型 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]