Network traffic prediction based on improved support vector machine

Abstract

Network traffic is featured by non-linear time-varying and chaos, and the existing prediction models based on support vector machine (SVM) have low stability and precision. We adopt fuzzy analytic hierarchy process to improve the SVM-based prediction model by first optimizing the parameters $$\sigma$$ and $$C$$ . Then SVM is trained using the optimal parameters, and the prediction model is built to forecast the network traffic. Experiment shows that the proposed algorithm cannot only track the variation trend of network traffic, but also achieve an accurate prediction with very small fluctuation of prediction error. Thus SVM-based model has high precision in predicting network traffic.