面向IP网络扩容应用的复杂网络流量预测方法

doi:10.11959/j.issn.1000-0801.2023175

摘要/Abstract

摘要：

IP 网络扩容是通信运营商保持网络平稳运行的一种常见维护方式，核心在于预测未来一段时间的网络流量走势。IP 网络流量非常复杂，具有局部不确定性、突发性、异质性等，给预测带来困难。提出了一种针对复杂网络流量的预测方法，它采用编码-解码结构，即在编码层增加全局特征、在解码层增加全局特征和局部特征解析的方式解决局部不确定性；通过增加先验知识缓解突发性；模型采用样本均衡、归一化等方式尽量提取数据的共性，避开数据的异质性。模型整体的参数较少，具有较强的泛化性能；同时采用人工特征和自动特征结合方式保证了浅层网络的准确率。实验结果表明，所提出的方法具有准确率高、泛化性能强的特性。目前该方法已经在工程中大规模应用。

关键词: 网络流量预测, IP网络扩容, 复杂网络流量

Abstract:

IP network expansion is a common maintenance method for communication operators to keep the network running smoothly.The core is to predict the network traffic trend in the future period.IP network traffic is very complex, with local uncertainty, burstiness, heterogeneity, etc., which brings difficulties to prediction.A prediction method for complex network traffic was proposed.It adopted an encode-decode structure, adding global features to the encoding layer, and combining global features and local features in the decoding layer to solve local uncertainties.The model used sample balance, normalization and other methods to extract the commonality of the data as much as possible to avoid the heterogeneity of the data.And emergencies were alleviated by increasing prior knowledge.The overall model had fewer parameters and had strong generalization performance, at the same time, the combination of artificial features and automatic features ensured the accuracy of the shallow network.The experimental results show that the proposed method has the characteristics of high accuracy and strong generalization performance.At present, this method has been applied on a large scale in engineering.

Key words: network traffic prediction, IP network expansion, complex network traffic

中图分类号:

TP393

高勇, 陆钱春, 李锋. 面向IP网络扩容应用的复杂网络流量预测方法[J]. 电信科学, 2023, 39(9): 21-31.

Yong GAO, Qianchun LU, Feng LI. A complex network traffic prediction method for IP network expansion applications[J]. Telecommunications Science, 2023, 39(9): 21-31.

图/表 9

图1

图2

图3

图4

图5

表1

表2

图6

表3

参考文献 16

[1]	王帅, 李毅, 陈茂林 ,等. 高校不限流量套餐下 LTE 网络扩容[J]. 电信科学, 2020,36(1): 101-108.
	WANG S , LI Y , CHEN M L ,et al. LTE network expansion under unlimited traffic packages in colleges and universities[J]. Telecommunications Science, 2020,36(1): 101-108.
[2]	黄圣红, 李洲 . 一种基于用户体验的4G网络扩容策略研究[J]. 电信工程技术与标准化, 2019,32(4): 42-45.
	HUANG S H , LI Z . Research on TD-LTE capacity expansion strategy based on user experience[J]. Telecom Engineering Technics and Standardization, 2019,32(4): 42-45.
[3]	ZARE M H , MASNADI-SHIRAZI M A . ARIMA model for network traffic prediction and anomaly detection[C]// Proceedings of 2008 International Symposium on Information Technology. Piscataway:IEEE Press, 2008: 1-6.
[4]	TAYLOR S J , LETHAM B . Forecasting at scale[J]. The American Statistician, 2018,72(1): 37-45.
[5]	CHEN T Q , GUESTRIN C . XGBoost:a scalable tree boosting system[C]// Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining. New York:ACM Press, 2016: 785-794.
[6]	MITCHELL R , FRANK E . Accelerating the XGBoost algorithm using GPU computing[J]. PeerJ Computer Science, 2017,3:e127.
[7]	BANERJEE T , SINHA S , CHOUDHURY P . Long term and short term forecasting of horticultural produce based on the LSTM network model[J]. Applied Intelligence, 2022,52(8): 9117-9147.
[8]	LAI G K , CHANG W C , YANG Y M ,et al. Modeling long- and short-term temporal patterns with deep neural networks[C]// Proceedings of SIGIR’18:the 41st International ACM SIGIR Conference on Research ＆ Development in Information Retrieval. New York:ACM Press, 2018: 95-104.
[9]	ZHOU H Y , ZHANG S H , PENG J Q ,et al. Informer:beyond efficient transformer for long sequence time-series forecasting[J]. Proceedings of the AAAI Conference on Artificial Intelligence, 2021,35(12): 11106-11115.
[10]	BI H J , LU L L , MENG Y Z . Hierarchical attention network for multivariate time series long-term forecasting[J]. Applied Intelligence, 2023,53(5): 5060-5071.
[11]	JIANG W W , LUO J Y , HE M ,et al. Graph neural network for traffic forecasting:the research progress[J]. ISPRS International Journal of Geo-Information, 2023,12(3): 100.
[12]	LI Z L , YU J , ZHANG G W . Dynamic spatio-temporal graph network with adaptive propagation mechanism for multivariate time series forecasting[J]. Expert Systems with Applications, 2023,216:119374.
[13]	熊伟 . 基于突变理论及协同学的网络流量异常检测方法研究[D]. 武汉:华中科技大学, 2011.
	XIONG W . Research on anomaly detection method of network traffic based on catastrophe theory and synergetics[D]. Wuhan:Huazhong University of Science and Technology, 2011.
[14]	马佳艳, 王萍, 夏伟 ,等. 复杂网络下的网络流量预测和预警研究[J]. 计算机与现代化, 2018(1): 102-106.
	MA J Y , WANG P , XIA W ,et al. Research on network traffic prediction and early warning in complex networks[J]. Computer and Modernization, 2018(1): 102-106.
[15]	WANG S P , NIE L S , LI G J ,et al. A multitask learning-based network traffic prediction approach for SDN-enabled industrial Internet of Things[J]. IEEE Transactions on Industrial Informatics, 2022,18(11): 7475-7483.
[16]	BERGSTRA J , YAMINS D , COX D D . Making a science of model search:hyperparameter optimization in hundreds of dimensions for vision architectures[C]// Proceedings of the 30th International Conference on International Conference on Machine Learning - Volume 28. New York:ACM Press, 2013: 115-123.

数据集名	开始采集时间	结束采集时间	端口数量/个	缺失数据占比
数据集A	2019年11月	2020年11月	14 824	0.1%
数据集B	2017年2月	2019年12月	35 464	10%

方法	prophet	XGBoost	XGBoost+hyperopt	LSTNet	Informer	本文方法
准确率	60.64%	68.78%	68.49%	58.56%	55.51%	76.34%

训练情况	测试集为数据集A的准确率	测试集为数据集B的准确率
训练数据集A	76.34%	67.58%
训练数据集B	77.08%	68.96%
训练数据集A+修正值	77.40%	68.64%