摘要
通信网络的业务量指标对指导网络运维工作非常重要。如果能够提前预知业务量指标的变化,则有助于网络中资源调度和节能减排工作的成功实施。当前,对大规模小区的海量业务量指标实现快速、精准预测是一个不小的挑战。本文研究了一种智能化大规模小区业务量指标预测的方法。首先,通过构造鲁棒的特征,对小区的高低负荷情况进行描述;之后,运用聚类算法,对小区进行基于业务量情况的场景划分;最后,对于不同业务场景下的小区,通过集成学习算法进行预测模型的建模与训练。经实验验证,该模型预测性能良好,成功实现了对大规模小区业务量指标的精准预测。
In the communication network,cell business indicators have an great importance of network optimization work.For instance,cell business indicators prediction could help to realize the work of resource scheduling and energy conservation successfully.It is a great challenge of making prediction quickly and accurately for the large scale cells.This paper proposed an intelligentized method that could be elegantly applied to the problem of large scale cells business indicators prediction.First,robust feature was constructed to precisely describe condition of cell’s business.Then,cluster algorithm was hired to split large scale cells into diff erent business scenes.Finally,for each scene,prediction model is built by GBDT algorithm.The experiments exhibit that our proposal has the great performance and successfully realizes the predicting for the large scale cells.
作者
葛澍
王西点
郭若沛
GE Shu;WANG Xi-dian;GUO Ruo-pei(China Mobile Group Co.,Ltd.,Beijing 100032,China;China Mobile Group Design Institute Co.,Ltd.,Beijing 100080,China)
出处
《电信工程技术与标准化》
2023年第5期1-7,共7页
Telecom Engineering Technics and Standardization
关键词
通信网络
网络运维
业务量预测
communication network
network optimization
cell business prediction
