Analysis and Prediction of Sky Wave Channel Based on LSTM Network

ZHAO Sifang; MA Qiyuan; LI Tienan

doi:10.3969/j.issn.1672-9730.2019.12.017

Abstract：The short wave communication process is affected by many complicated factors such as time,space,frequency and sunspots,resulting in unstable signal quality. The traditional method of calculation and prediction is difficult to simulate the influ?ence of all communication elements from the "theoretical model" level. The detection frequency selection method has a strong depen?dence on the equipment and platform,and it will cause the pollution of the short wave spectrum. In this paper,a multilayer neural network algorithm is used to analyze and predict the shortwave sky wave communication circuit using historical experience data. The first layer uses the long and short memory neural network model to quickly fit the slow and fast change characteristics of the short wave sky wave circuit with time. The second layer uses the deep neural network model to transform the high dimension stack se?quence into the same standard output sequence as the sample,thus the prediction result of the change law of the short wave sky wave circuit is obtained. The prediction results of the communication circuit are compared with the theoretical calculation method and the measured result. Finally,the comparison results show that the multi-layer neural network algorithm can obtain higher pre?diction accuracy than the theoretical calculation method,and it is suitable for the analysis and prediction of the short wave sky wave circuit.