Wavelet-Based Stacked Denoising Autoencoders for Cell Phone Base Station User Number Prediction

User number prediction in cell phone base station is a very important problem for cell phone communication system design and base station location selection. Recent years, we have witnessed the encouraging potentials of deep neural networks for real-life applications of various domains. User number prediction, however, is still in its initial stage. In this paper, we propose a wavelet-based stacked denoising autoencoder deep learning framework, named as Wavelet-SDA, which adopted wavelet to decompose the user volume signal as several sub channels, for each channel, an independent SDA model is introduce to achieve accurately signal prediction. In order to exploit the correlations between different base stations, a transfer entropy based knowledge transfer is also adopted by the proposed framework. Extensive experiments on real-life cell phone base station log dataset of Wuxi city demonstrate the strong predictive power of Wavelet-SDA comparison to some state-of-the-art competitors.