Abstract
This paper proposes a wireless network traffic prediction model based on Bayesian Gaussian tensor decomposition and recurrent neural network with rectified linear unit (BGCP-RNN-ReLU model), which can effectively predict the changes in the upstream and downstream network traffic in a short period of time in the future. The research is divided into two parts: (i) The missing observations are imputed by an algorithm based on Bayesian Gaussian tensor decomposition. (ii) The recurrent neural network is used to forecast the true observations only rather than both true and estimated observations. The results show that, compared with other combined models of missing data imputation and neural networks, the BGCP-RNN-ReLU model proposed in this paper has the smallest prediction error for both the upstream and downstream traffic. The new model achieves better forecasting precision, and thus can help to regulate the load of communication station to reduce resource consumption.HighlightsThe problem of forecasting wireless network traffic with missing values is divided in two stages tohandle.A newly proposed method can more efficiently impute missing values in wireless network traffic data.Simple recurrent neural network obtains better prediction performance than other complex networks.
Highlights
At present, mobile communication technology has entered the 5G from the 4G
This paper proposes a prediction model Bayesian Gaussian CP decomposition (BGCP)-recurrent neural network (RNN) for time series with missing values based on BGCP and RNNs
In experiments based on real wireless network traffic data, compared with other time series prediction models that deal with missing values, BGCP-RNN has a smaller number of parameters, and achieves the best results on both the up-down-traffic short-term prediction tasks, which means our model is significantly more accurate and faster than others
Summary
Mobile communication technology has entered the 5G from the 4G. While the rapid development of the mobile Internet has brought many conveniences to people, it has brought great traffic pressure to communication base stations. This paper proposes a prediction model BGCP-RNN for time series with missing values based on BGCP and RNNs. Compared with previous papers which use tensor decomposition and neural networks to predict network traffic, the biggest difference of this paper is that a new tensor decomposition approach is applied to missing value processing, which effectively solves the problem that the general neural network cannot directly process the data with missing values. In experiments based on real wireless network traffic data, compared with other time series prediction models that deal with missing values, BGCP-RNN has a smaller number of parameters, and achieves the best results on both the up-down-traffic short-term prediction tasks, which means our model is significantly more accurate and faster than others.
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
