The Research On Digital Predistortion Technology Wideband Power Amplifier

As the number of data users continues to increase,the demand for wireless communication rates is increasing.The wireless communication system is also developing in the direction of high speed and broadband,but its rapid development has also brought unprecedented energy consumption.In wireless communication networks,the energy consumption of mobile communication base stations accounts for more than 50%of the energy consumption of communication networks.Microwave power amplifiers,as the main body of energy consumption for the main equipment of wireless communication base stations,ensuring linear and high-efficiency operation is of great significance to the system's energy saving and consumption reduction.In order to take into account the efficiency and linearity of the power amplifier at the same time,linearization technology needs to be used to compensate the non-linear distortion generated by the power amplifier when it operates at high efficiency.Digital predistortion has become the mainstream technology of linearization of current power amplifiers due to its simple structure,flexible implementation,and excellent performance.The main research contents and innovations of this article include the following aspects:1.Aiming at the problem of high ADC cost on the feedback path of wideband digital predistortion technology,this paper proposes a novel indirect digital predistortion structure that reduces the number of ADCs in the feedback path.This structure only needs to collect a single component(I component or Q component)of the output signal of the PA,so only one ADC is needed on the feedback path,which reduces the number of ADCs on the feedback path and reduces hardware costs.At the same time,because this structure only collects the I component or Q component of the output signal of the power amplifier to calculate the predistorter coefficients,it can suppress the I/Q imbalance existing in the system to a certain extent.Test results show that when there is no I/Q imbalance in the system,the proposed structure has the same linearization effect as the traditional indirect learning structure.When there is an I/Q imbalance in the system,compared with the traditional indirect learning structure,the performance of the proposed method's Adjacent Channel Power Ratio(ACPR)is improved by about 4?6dB,which indicates that it has a certain suppression effect on the I/Q imbalance present in the system.2.For wideband signals,the amount of coefficients of the digital predistortion model increases significantly with the increase of the nonlinear order and memory depth,but there are a lot of redundant terms.These redundant terms not only increase the complexity of the model,but also increase the calculation error and affect the accuracy of the model modeling.Aiming at this problem,a new power amplifier behavior model pruning method is proposed.This method applies Regularized Subspace Pursuit algorithm to digital predistortion to cut redundant terms in the behavior model of power amplifier modeling.The RSP algorithm introduces the regularization process into the Subspace Pursuit algorithm.It combines regularization and retrospective thinking to realize the secondary screening of atoms and improves the reliability of the selected atoms in each iteration.So it has better model cutting performance.The test results show that when the number of coefficients is set equal,compared with the OMP and SP algorithms,the model after cropping using the RSP algorithm has better ACPR values and normalized mean square error(NMSE)Value,the proposed method has better digital predistortion model clipping ability than using OMP algorithm and SP algorithm.Compared with the full GMP model,after using the proposed algorithm to prune the model,not only the model coefficient amount is reduced by 83.7%,but the performance of NMSE and ACPR is also improved by about 3.4dB and 3dB,respectively.