用记忆型BP神经网络实现HPA预失真的算法研究

doi:10.3969/j.issn.1000-436x.2014.01.003

摘要/Abstract

摘要：

在分析宽频带CMMB直放站高功率功放(HPA）特性的基础上，提出了一种可分离处理功放记忆效应和非线性的延时神经网络（FIR-NLNNN）模型。该模型以实数延时神经网络（RVTDNN）为基础，用Levenberg-Marquardt（LM）优化算法确定神经网络系数，在模型中新增参数 w⁰，给出了 LM 算法的修改公式。接着在预失真神经网络系统中引入Bayesian机理消除LM算法的过拟合现象，构建CMMB数字直放站的间接学习预失真器，拟合HPA的非线性和记忆效应。结果表明：RVTDNN和FIR-NLNNN 2种预失真器均能显著提高系统性能，降低邻信道功率比30 dB左右。在保持均方误差（MSE）小于10^?6的情况下，FIR-NLNNN结构的网络参数比RVTDNN结构减少了近50%，迭代过程中的乘法和加法次数约降低75%。

关键词: 高功率功放, 预失真器, 神经网络, 记忆效应, LM算法, Bayesian算法

Abstract:

Based on the characteristic analysis of the high power amplifier (HPA) in wide-band CMMB repeater stations,a novel neural network was proposed which can respectively process the memory effect and the nonlinear of power amplifier.The novel model based on real-valued time-delay neural networks(RVTDNN) uses the Levenberg-Marquardt (LM) optimization to iteratively update the coefficients of the neural network.Due to the new parameters w⁰in the novel NN model,the modified formulas of LM algorithm were provided.Next，in order to eliminate the over-fitting of LM algorithm,the Bayesian regularization algorithm was applied to the predistortion system.Additionally,the predistorter of CMMB repeater stations based on the indirect learning method was constructed to simulate the nonlinearity and memory effect of HPA.Simulation results show that both the NN models can improve system performance and reduce ACEPR (adjacent channel error power ratio ) by about 30 dB.Moreover,with the mean square error less than 10^?6,the coefficient of network for FIR-NLNNN is about half of that for RVTDNN.Similarly,the times of multiplication and addition in the iterative process of FIR-NLNNN are about 25% of that for RVTDNN.

Key words: HPA, predistortion, neural network, memory effect, LM algorithm, Bayesian algorithm

黄春晖,温永杰. 用记忆型BP神经网络实现HPA预失真的算法研究[J]. 通信学报, 2014, 35(1): 16-23.

Chun-hui HUANG,Yong-jie WEN. Algorithm study of digital HPA predistortion using one novel memory type BP neural network[J]. Journal on Communications, 2014, 35(1): 16-23.

图/表 9

图1

图2

图3

表1

图4

图5

图6

图7

图8

参考文献 25

[23]	钱业青 . 一种高效的用于 RF 功率放大器线性化的自适应预失真结构[J]. 通信学报, 2006,27(5): 36-39. QIAN Y Q . High-efficient adaptive predistortion structure for RF power amplifier linearization[J]. Journal on Communications, 2006,27(5): 36-39.
[24]	HAGAN M T , MENHAJ M B . Training feed-forward networks with the marquardt algorithm[J]. IEEE Transactions on Neural Networks, 1994,5(6): 989-991.
[1]	CAVERS J K . Amplifier linearization using a digital predistorter with fast adaptation and low memory requirements[J]. IEEE Transactions on Vehicular Technology, 1990,39(4): 374-381.
[2]	VARAJRAM P , MOHAMMSDY S , HAMIDON M N . Complex gain predistortion in WCDMA power amplifiers with memory effects[J]. The International Arab Journal of Information Technology, 2010,7(3): 333-339.
[3]	MIRRI D , IUCULANO G , FILICORI F . A modified volterra series approach for nonlinear dynamic systems modeling[J]. IEEE Transactions on Circuits and Systems,Fundamental Theory and Applications, 2002,49(8): 1118-1127.
[4]	ZHU A , BRAZIL T J . Behavioral modeling of RF power amplifiers based on pruned voterra eries[J]. IEEE Microwave and Wireless Components Letters, 2004,14(12): 563-565.
[5]	SAFARI N , ROSTE T , FEDORENKO P . An approximation of volterra series using delay envelopes,applied to digital predistortion of RF power amplifiers with memory effects[J]. IEEE Microwave and Wireless Components Letters, 2008,18(2): 115-117.
[6]	QIU W , ZHONG Z , REN G C . The research of adaptive digital predistortion based on SISO-neural network[A]. International Conference on Communications and Mobile Computing[C]. China, 2009. 451-454.
[7]	ZAYNNI R , BOUALLENGUE R , ROVIRAS D . An adaptive neural network predistorter for nonstationary HPA in OFDM systems[A]. The 15th European Signal Processing Conference[C]. Poznan,Poland, 2007. 1353-1354.
[8]	DING Y M , JIAO D P , WANG J K . A review of adaptive predistortion techniques of HPA[A]. Asia Pacific Conference on Postgraduate Research in Microelectronics and Electronics[C]. Shanghai,China, 2009. 341-343.
[9]	ZAYNNI R , BOUALLENGUE R . Predistortion for the compensation of HPA nonlinearity with neural networks:application to satellite communications[J]. International Journal of Computer Science and Network Security, 2007,7(3): 98-101.
[10]	BOUMAIZA S , MKADEM F . Wideband RF power amplifier predistortion using real-valued time-delay neural networks[A]. Proceeding of the 39th European Microwave Conference[C]. Rome,Italy, 2009. 1149-1151.
[11]	瞿建峰, 周健义, 洪伟 . 有记忆的功放实数延时模糊神经网络模型[J]. 微波学报, 2009,25(5): 41-43. QU J F , ZHOU J Y , HONG W . Real-valued time-delay neuro-fuzzy model for power amplifier with memory effects[J]. Journal of Microwaves, 2009,25(5): 41-43.
[12]	SALEH A . Frequency-independent and frequency-dependent nonlinear models of TWT amplifiers[J]. IEEE Transactions on Communications, 1981,29(11): 1117-1120.
[13]	张玉梅, 南敬昌 . 基于 Saleh 函数的功放行为模型研究[J]. 微电子学与计算机, 2010,27(12): 121-126. ZHANG Y M , NAN J C . Study of behavior models based on the Saleh function for power amplifier[J]. Microelectronics and Computer, 2010,27(12): 121-126.
[14]	JARDIN P , BAUDOIN G . Filter look up table method for power amplifiers linearization[J]. IEEE Transaction on Vehicular Technology, 2007,56(3): 1996-1999.
[15]	HAGAN M T , DEMUTH H B , BEALE M . Neural Network Design[M]. Beijing: China Machine PressPress, 2002. 197-235.
[16]	BOSCH W , GATTI G . Measurement and simulation of memory effects in predistortion linearizers[J]. IEEE Transactions on Microwave Theory and Techniques, 1989,37(12): 1885-1888.
[17]	CUI H , AI B , ZHAO X M . Design of a BPNN predistorter for nonlinear HPA with memory[J]. Journal of Information and Computational Science, 2010,7(4): 863-868.
[18]	ZAYNNI R , BOUALLENGUE R , ROVIRAS D . Adaptive predistortions based on neural networks associated with levenberg marquardt algorithm for satellite down links[J]. EURASIP Journal on Wireless Communications and Networking, 2008,2008:132729.
[19]	RAWAT M , RAWAT K . Adaptive digital predistortion of wireless power amplifiers/transmitters uing dynamic real-vlued focused timedelay line neural networks[J]. IEEE Transactions on Microwave Theory and Techniques, 2010,58(1): 95-100.
[20]	CUI H , ZHAO X M . A novel NN-predistorter learning method for nonlinear HPA[J]. Wireless Pres Commun, 2012,63(2): 469-482.
[21]	MKADEM F , BOUMAIZA S . Physically inspired neural network model for RF power amplifier behavioral modeling and digital predistortion[J]. IEEE Transactions on Microwave Theory and Techniques, 2011,59(4): 913-922.
[22]	RAWAT M , GHANNOUCHI F M . A mutual distortion and impairment compensator for wideband direct-conversion transmitters using neural networks[J]. IEEE Transactions on Broadcasting, 2012,58(2): 168-176.
[25]	FORESEE D F , HAGAN M T . Gauss-newton approximation to bayesian learning[A]. International Conference on Neural Networks[C]. Houston,TX, 1997. 1930-1932.

网络结构	乘法次数	加法次数	执行时间/s
RVTD NN	5.748 5×10¹⁰	5.687 7×10¹⁰	459.427 5
FIR-NLN NN	1.339 3×10¹⁰	1.313 5×10¹⁰	292.582 7