Missing Data Prediction Using LSTM for Radio Frequency Transfer System

Missing data affects stability analysis of Radio Frequency (RF) transfer system. In this paper, a missing data prediction method using Long Short-Term Memory (LSTM) network is proposed for the short-term failure in RF transfer system. The performance is verified by experimenting on the measurement data of 2.4 GHz RF transfer system using 3000 km single optical fiber bi-directionally. The results demonstrate that our proposed method can achieve higher prediction accuracy compared to back propagation neural network (BPNN) and Support Vector Regression (SVR). When the prediction time is 240 s, the Root Mean Squared Error (RMSE) between the predicted value and the actual value is less than 0.001. This provides an effective and feasible method to impute missing data for RF transfer system.