Non-linear Frequency Estimation Based Upon Convolution Neural Network

Abstract

A frequency estimator is to find the frequency of the received signal distorted by noise, and it has been found in systems such as radar, sonar, etc. In this research, the linear frequency modulation signal(LFM)of the radar system, the detection of the moving target and its frequency estimation were highlighted. The pre-trained convolutional neural network algorithm was used, which proved highly accurate in detecting the target and estimating its frequency, despite the lower level of the signal to the noise ratio(SNR) where it was dealt with. On the basis of the signals received as a two-dimensional image and taking into account the Doppler effect, as the target moves at different speeds and accelerations, the signals have been classified into single-frequency signals and LFM signals