LTE User behavior prediction Via LSTM

In this letter we want to use LSTM-RNN to analyze and predict the behavior of LTE users in mobile network. Predicting the user statistics and data usage on the whole mobile network is going to be studied. Analyzing and predicting the user charging data requests in LTE network and finally, finding abnormal base stations has been predicted via DNN networks. To accomplish that we considered aggregated LTE users under each base station and considered the behavior statistics as a time series. Ultimately, we forecasted the coverage index (CI) as a calculated metric to identify anomalies and determine the count of abnormal sites with fewer LTE CDRs compared to normal ones using LSTM on generated time series. This prediction can be used for planning and optimization teams to find and solve any abnormal site level parameters. Proposed method can have both the role of corrective actions on base stations to rectify any telecommunication issue on live network and also predictive action for forecasting network condition in site level scale to increase user satisfaction in advance.