Time Series Prediction of Wireless Network Traffic Flow Based on Wavelet Analysis and BP Neural Network

Building an appropriate model is the key issue of network traffic flow prediction [1]. This paper presents a hybrid model based on the combination of wavelet denoising and BP neural network. First, wavelet transform is used to normalize and denoise the collected raw traffic flow. The index for estimating prediction accuracy is Mean Absolute Error (MAE). The original data is supported by a large telecommunications company in China. At present, research has made corresponding progress and the prediction accuracy of this combined model has reached more than 80%.