Research On Theory And Method Of Wide-band Adaptive Digital Predistortion In Mobile Communication System

Mobile communication systems use non-constant envelope linear modulationmethod, such as QAM (Quadrature Amplitude Modulation), OFDM (OrthogonalFrequency Division Multiplexing) and so on, which can more effectively improve thespectrum efficiency compared with mobile communication systems using constantenvelope modulation signal. However, when a non-constant envelope goes through thepower amplifier, power amplifier output signal will produce intermodulation distortionand spectral spreading because of its inherent non-linear, which will lead to a declinein the quality of the transmission system. Therefore, we must make the appropriatecompensation for its non-linear feature.For the purpose of obtaining a higher power output efficiency in a wirelesscommunication system with non-constant envelope digital modulation, we need theamplifier works in the vicinity of the saturation point，which has the most seriousproblems of nonlinear. So the question of power efficiency and linearity of thecommunication system has become a major contradiction. Fortunately, linearizationtechniques play a very crucial role in solving conflicts between power efficiency andlinearity.Among the many linearization techniques, digital pre-distortion technology standsout as the most popular linearization technology due to its good stability, suiting forbroadband communications, high precision and so on. Because of broadband amplifierserious memory effect, which makes it difficult to get good linearization results bytraditional non-memory model and predistortion technique. So, memory broadbandamplifier behavioral modeling and digital pre-distortion technology have become animportant way to reduce the spread spectrum and adjacent channel interference, thusimproving the efficiency of the power amplifier.The main contents of this paper include:(1) Predistorter model options;(2) twopre-distorter model parameter identification algorithms, RLS (Recursive Least-Square)algorithm and Gauss-Newton algorithm;(3) digital predistortion joint verificationsystem design based on ADS (Advanced Design System) and MATLAB. Poweramplifier behavioral model is mainly divided into no memory behavioral models andmemory behavioral models. The paper comparatively studies several commonbehavioral models and then puts forward a generalized multidimensional memory polynomial (GMDMP) model. Simulation results prove that the GMDMP model hashigher accuracy than other models, among which normalized mean square error getsimproved nearly17dB compared with no memory polynomial model. Additionally, itcan better approximate the actual amplifier characteristics. Therefore, GMDMP modelcould be used to construct predistorter.To further verify the degree of improvement with power amplifier nonlinearity,owing to predistorter of GMDMP model. Digital predistortion joint verification systembased on ADS and MATLAB is designed in this paper. System amplifier circuit iscomposed respectively by MRF7S21170H and MRF6S21140H transistor.3G systemWCDMA (Wideband Code Division Multiple Access) signal with10MHz bandwidthand4G system LTE (Long Term Evolution) signal with60MHz bandwidth are chosento test the system. The result shows that the ACPR (Adjacent Channel Power Ratio) ofthe PA output signal compensated by predistorter is respectively improved18.585dBand11.702dB compared with the PA (Power Amplifier) original output signal andnonlinear memory effects of PA dynamic characteristic curve are also improved, whichverifies the digital pre-distortion system created in this paper can effectively improvesolving power amplifier nonlinear problems.