| With the gradual popularization of 5G communication,mobile communication has become a hotspot.As an important device in wireless communication systems,power amplifiers(PAs),due to their inherent non-linear and memory characteristics,have significant non-linear distortion after the signal has been amplified,so amplifier linearization has become a key technology in the field of communication,among these linearization techniques,digital pre-distortion is one of the most suitable methods for modern communication systems.Based on this,this thesis focuses on exploring digital pre-distortion techniques based on master curve analysis in terms of improving modelling accuracy and iterative convergence speed.The main work of this thesis is as follows.In digital pre-distortion techniques,generalized memory polynomial models(GMP)are widely used in amplifier modelling.In this thesis,we propose a modelling method that combines master curve analysis with GMP to solve the problem of insufficient modelling accuracy due to excessive redundant terms in the kernel of the GMP model.The results of the MATLAB simulation show that although the proposed method takes about 6s longer than the GMP model in terms of time consumption,there is a significant improvement in modelling accuracy: the normalized mean square error(NMSE)improves by 2d B when a class F amplifier is selected,and by 6d B when a Doherty class amplifier with worse linearity is chosen.The proposed algorithm has a better balance between modelling accuracy and modelling time.In the pre-distortion technique based on iterative learning architecture,the traditional iterative learning algorithm often uses the P-type learning law.In this thesis,the PD-type adaptive iterative learning method is proposed to solve the problem that the only adjustable parameter in the traditional P-type iterative learning law causes the convergence speed to be not fast enough.The result shows that the proposed method has about the same convergence accuracy as the traditional iterative algorithm but has a better convergence speed through MATLAB simulation analysis.After obtaining the optimum input according to the adaptive iterative learning algorithm,a pre-distortion device is built using the combination of the master curve analysis and GMP model proposed in this thesis,and then the pre-distortion device is cascaded with the amplifier to form a complete pre-distortion structure.Firstly,through simulation analysis,the results show that the adjacent channel power ratio(ACPR)of the Doherty class amplifier output signal has improved by 19.89 d B.Then the conventional algorithms are compared and show that better linearization can be achieved using the proposed structure.Finally,the performance of the proposed structure is verified by a digital pre-distortion platform,and the experimental results show that the ACPR of the Doherty class amplifier output signal is improved by 17.99 d B.Therefore,the proposed structure has good correction effect on amplifier non-linearity. |
