基于EWT和NeuralProphet-MLP的蜂窝网络流量长期预测方法
Long-term Prediction Method for Cellular Network Traffic Based on EWT and NeuralProphet-MLP
蒋东浩 1赵洪华 1王真1
作者信息
- 1. 中国人民解放军陆军工程大学 指挥控制工程学院,江苏 南京 210007
- 折叠
摘要
蜂窝网络流量长期预测对网络扩展和优化具有重要意义,针对长期预测中数据可用性低以及非线性等弊端所带来的诸多挑战,提出一种基于分解的分频预测模型.分别采用NeuralProphet模型和多层感知机对分解出的低频分量和中高频分量进行预测,最后对各分量预测结果进行逆经验小波变换得到最终结果.在真实的蜂窝网络流量数据集上进行验证,结果表明所提方法相较于传统预测模型在准确度上有较大提升,具有较好的应用价值.
Abstract
Long term prediction of cellular network traffic is of great significance for network expansion and optimization.To address the many challenges brought by low data availability and nonlinearity in long-term prediction,a decomposition-based frequency division prediction model is proposed.The NeuroalProphet model and multilayer perceptron are used to predict the decomposed low-frequency components and mid-to-high frequency components,and the final results are obtained by applying inverse empirical wavelet transform to the predicted results of each component.Verification is carried on a real cellular network traffic dataset,the results show that the proposed method has a significant improvement in accuracy compared to traditional prediction models and has good application value.
关键词
蜂窝网络流量预测/经验小波变换/NeuralProphet模型/多层感知机Key words
cellular network traffic prediction/empirical wavelet transform/NeuralProphet model/multilayer perceptron引用本文复制引用
出版年
2024
国家科技期刊平台
NSTL
万方数据