Support Vector Machines On Mobile Gsm Network Traffic Prediction And Empirical Analysis

With competion becoming more and more intense in telecommunications market,China mobile pay more attention on telephone traffic forecasting. Traffic directly related to the size of the economic benefits of mobile companies. Traffic how much direct impact on network design, network planning and network performance evaluation, wireless mobile telephone traffic forecast will enable the company to better prevent the network congestion, coverage and interference issues, so that the loss of Traffic re - Absorption come back, so to increase revenue for the purpose. Therefore research of telephone traffic forecasting is necessary.The support vector machine is a date mining new technology; it is a new tool that draws support the/ptimized method to solute the machine learning questions. But applied it the telecommunications market to forecast certainly were not actually many. This article has analyzed telephone traffic forecast in the present telecommunications market, in this foundation,it propose using the support vector machine to forecast the telephone traffic. Under the rationale of time series forecast,the article produced a model that base on the time series support vector machine forecast. And it selects the date of telephone traffic of Zhuzhou branch of HuNan mobile from January to August,2008. In the situation of choosing suitable kernel function and its parameters,it carried on the real examination to forecast the telephone traffic regarding the support vector machine.The results showed that the SVM method not only can more accurately forecast the mobile GSM network Traffic trends, forecasts and also has a good effect. Thus it an be seen,carries on the forecast prospect using the SVM to the telecommunications market extremely to favor.