一种短信业务分布预测算法分析

随着移动通信短信业务的迅速发展,现有的短信业务分布预测算法已经无法满足需求。提出了一种基于聚类分析和模拟退火的短信业务分布预测算法。根据电子地图、基站语音和短信业务统计数据,提取影响短信业务分布的相关属性,采用聚类分析方法中的自组织映射(Self Organizing Map,SOM)方法和k-means算法,将网络中基站进行分类,然后采用模拟退火算法和最小二乘法计算得到不同类型基站中的不同地物的短信日均业务密度值,从而达到对短信进行业务分布预测的目的。通过实际基站短信业务统计数据的验证,该算法大幅提高了短信业务分布预测的准确度。

An Algorithm of SMS Distribution Prediction

With rapid development of SMS in mobile communication field,the existing SMS distribution prediction algorithm cannot meet the requirements.This paper presents a new SMS distribution prediction algorithm based on Cluster Analysis and Simulated Annealing.Firstly,the algorithm abstracts the related properties affecting SMS distribution according to digital map statistical data and base station voice and SMS service statistical data;then,classifies the base stations in network,using cluster analysis methods,SOM and k-means;next,calculates the SMS density of different base station classes and different ground object types using Simulated Annealing algorithm and least square algorithm;thus realizes the average daily SMS distribution prediction.Through the validation by actual base station SMS statistical data,it proves that this algorithm can improve the accuracy of SMS distribution prediction remarkably.