Research On Radio Frequency Digital Cancellation Technology Based On Deep Learning

As the electromagnetic environment of the battlefield becomes more and more complex,the simultaneous operation mode of the jammer in the radar system has become a focus problem.After the development of recent years,the adaptive filtering technology in the field of voice communication has been applied to the problem of radar jammer processing at the same time.With the in-depth study of the problem and the development of the signal model,the shortcomings of the existing solutions based on adaptive filter have been found one after another.In this paper,the influence of the nonlinear distortion in the transmitter of radar jammer on the interference cancellation is further studied,and an improved interference cancellation scheme based on deep learning and adaptive algorithm is proposed.A large number of data simulation and result analysis are carried out,and the effectiveness of the proposed scheme is proved by comparing the cancellation ratio before and after the improvement.Firstly,this paper analyzes the signal processing flow of adaptive algorithms such as least mean square(LMS)algorithm by deducing the principle of adaptive cancellation and the structure of digital filter.By analyzing and modeling the different structures of nonlinear power amplifier in the transmitter of radar jammer,two behavior models are obtained.The main difference between them is whether they have memory.The LMS algorithm is used to eliminate two kinds of interference signals which pass through linear and nonlinear power amplifiers respectively.Through the comparison of the results,it is found that the cancellation ratio of the signal considering nonlinear power amplifier is obviously decreased,and the existing LMS algorithm is slightly insufficient in processing nonlinear data because of its linear structure.In recent years,the rapid development of deep learning technology has a strong ability of nonlinear data processing.In this paper,deep neural networks(DNN)is used to re fit the reference signal in LMS algorithm,so as to improve the linearity between the reference signal and the real selfinterference signal,and increase the linearity between the reference signal and the estimated signal,so as to improve the cancellation effect.In the background of jammer with digital RF storage technology structure,through the simulation of signals with two modulation modes,different frequencies and signal to interference ratio,a higher cancellation ratio is obtained compared with the traditional algorithm,which verifies the effectiveness of the adaptive cancellation scheme based on DNN.Secondly,from the point of view of the simulation process,DNN has a higher requirement on the prior conditions of memory model power amplifier fitting.According to this phenomenon,an adaptive cancellation scheme based on recurrent neural network(RNN)is proposed.In order to get a better network model than DNN,the long short term memory(LSTM)network is used to fit the data of the nonlinear model with memory.Finally,the effectiveness of the scheme to improve the cancellation effect is verified by simulation.