Method for predicting mobile communication telephone traffic based on clustered LS-SVM (Least Squares-Support Vector Machine)

Abstract

The invention discloses a method for predicting mobile communication telephone traffic based on clustered LS-SVM (Least Squares-Support Vector Machine), belonging to the field of communication and solving the problem that the LS-SVM has high computational complexity and lowered generalization capability in practical application. The method comprises the following steps of: 1. selecting the historical data of telephone traffic of 4-6 months at the current moment, taking the historical data of telephone traffic as a training sample, pre-processing the historical data, clustering, carrying out LS-SVM modeling, and obtaining C LS-SVM prediction models; 2. pre-processing a newly-input sample, and carrying out reconstruction of phase space and normalization processing; 3. classifying the reconstructed newly-input sample according to C clustered results obtained in the step 1, and determining the type of the sample; and 4. according to the classification result obtained in the step 3, inputting the newly-input sample into the LS-SVM prediction model of the corresponding class, outputting a prediction value, and quickly predicting the input sample to be processed.

Description

technical field [0001] The invention relates to a mobile communication traffic prediction method based on clustering LS-SVM, which belongs to the field of communication. Background technique [0002] The size of the mobile communication traffic reflects the strength of the voice channel being occupied to a certain extent. Mobile communication traffic data forecasting is of great value for mobile network maintenance and mobile communication decision-making. Timely and accurate traffic forecasting can effectively help mobile operators optimize and make decisions, guide network expansion, and improve network performance. running quality. [0003] The methods used in traffic forecasting in the past mainly include moving average, exponential average and other methods, but these methods are only suitable for trend forecasting of traffic volume, and cannot predict violent fluctuations in it, so the prediction accuracy is very low. The traffic volume is a typical time series, so i...

AI Technical Summary

Problems solved by technology

[0027] The purpose of the present invention is to solve the problem of high computational complexity and reduced generalization ability of LS-SVM in practical applications, and provides a mobile communication traffic prediction method based on clustering LS-SVM

Images