Study On Digital Predistortion Technology Of Power Amplifier

With the advent of the information age,the demand for high-speed information transmission is increasing day by day.Nowadays,the research direction of improving the information transmission rate mainly includes improving the spectrum utilization and increasing the signal transmission bandwidth.Power amplifier is an important part of communication system.It is not strictly linear between input and output signals due to its inherent nonlinear characteristics.When the input signal has a large transmission bandwidth or peak to average power ratio,the impact of this nonlinear distortion behavior is particularly obvious,which leads to the decline of the system's communication ability and seriously affects the user's communication experience.Therefore,the research on power amplifier linearization is an important and practical communication topic.As one of the research directions of power amplifier linearization technology,digital predistortion method has attracted more and more researchers' attention because of its good stability and practical value.It is a linearization technology with better application prospects.The neural network model has strong fitting ability and flexibility.The digital predistortion method based on the model is one of the current research hotspots.In this paper,the popular predistortion model of memory neural network is studied.Aiming at the problem of too many parameters in the neural network model,a new network structure is proposed,which simplifies the number of model parameters on the premise of ensuring performance.High bandwidth also brings about the problem of high sampling rate.When the sampling rate does not meet the demand of digital predistortion system,because the collected signal loses the high-order frequency component of the original signal,there is a big error between the parameters of the predistortion model and the actual situation,which affects the performance of the predistortion system.Therefore,this paper also studies the predistortion problem in the case of undersampling.Discusses the random demodulation method and the spectrum extrapolation method respectively,and puts forward the solution to the ideal filter problem in the spectrum extrapolation method,and proves the effectiveness of the proposed solution through simulation.In addition,this paper analyzes and improves the unstable performance of the spectrum extrapolation method in the case of the gradually reduced sampling bandwidth and it is demonstrated by simulation.