An Accurate Complexity-Reduced Simplified Volterra Series for Rf Power Amplifiers

Abstract

Abstract-An accurate complexity-reduced simplified Volterra (ACR-SV) series is introduced for RF power amplifiers (PAs). Based on the conventional simplified Volterra (SV) series, it takes memoryless nonlinearity and memory effect into consideration separately, while connected with a nonlinear memory effect (NME) in order to increase accuracy of the model. The proposed ACR-SV model is assessed using a GaN Class-F PA driven by two modulated signals (a WCDMA 1001 signal and a single carrier 16QAM signal with 40 MHz band width). The experimental results in forward modeling and DPD application demonstrate that the proposed ACR-SV model outperforms the memory polynomial (MP) model, the augmented complexity-reduced generalized memory polynomial (ACR-GMP), and the SV model. Compared with the MP model, the ACR-SV model shows a normalized mean square error (NMSE) improvement of 2.61 dB in forward modeling, average adjacent channel power ratio (ACPR) improvement of 3.7/4.2 dB in the DPD application with less 13% number of model coefficients. In comparison with the ACR-GMP model, the ACR-SV model shows NMSE improvement of 1.39 dB, ACPR improvement of 0.7/0.6 dB with comparable number of model coefficients. In contrast with the SV model, the ACR-SV model achieves similar model accuracy, but reduces approximately 53% of coefficients.

A Modified Generalized Memory Polynomial Model for Rf Power Amplifiers

“…Although the generalized memory polynomial (GMP) [6] model including the cross-band modulation terms is proposed for its high accuracy, a high model complexity is unavoidable when the values of memory depth and nonlinearity order are high. This is because all memory depths have the same nonlinearity order as many predistortion models [7][8][9][10].…”

A Modfied Real-Valued Feed-Forward Neural Network Low-Pass Equivalent Behavioral Model for Rf Power Amplfiers

“…An accurate model for the PA, which relates complexvalued envelope signals and is capable of predicting nonlinear dynamic behaviors observed in PAs with reduced computational cost, is essential to designing a DPD that works properly [7][8][9][10][11][12][13]. A real-valued feed-forward artificial neural network (ANN), when excited by the same complex-valued envelope signal applied to the PA input and after proper supervised training, can mimic the PA dynamic nonlinear behaviors and, in this way, generate a complex-valued envelope signal very close to the measured PA output signal [10].…”

Low-Pass Equivalent Behavioral Modeling of Rf Power Amplifiers Using Two Independent Real-Valued Feed-Forward Neural Networks

“…Volterra series [7][8][9] and artificial neural networks (ANNs) [10][11][12][13][14][15] are the most widely reported techniques that can provide an adequate mathematical description for the PA behavior. The selection of a particular technique targets the improvement of the trade-off between increasing modeling accuracy and reducing computational cost.…”

“…In literature, there are various techniques that can provide an adequate mathematical description for the PA behavior. Volterra series [7][8][9] and artificial neural networks (ANNs) [10][11][12][13][14][15] are the most widely reported techniques that can simultaneously describe nonlinear and dynamic behaviors. ANNs have the advantage of requiring a lower number of parameters than the Volterra series and have more general validity than polynomial approximations.…”