ESTUDIO COMPARATIVO DE DIFERENTES REDES NEURONALES PARA PREDISTORSIÓN DIGITAL DE AMPLIFICADORES DE RF (COMPARATIVE STUDY OF DIFFERENT NEURAL NETWORKS FOR DIGITAL PREDISTORTION OF RF AMPLIFIERS.)
Resumen
Resumen
En este trabajo se presenta una comparativa de diversas arquitecturas de redes neuronales artificiales, como método de modelado de comportamiento, para realizar predistorsión digital de amplificadores de potencia. Las arquitecturas de redes neuronales que se compararan son RVFTDNN, FC2HLANN, M2HLANN y NARX, con las cuales se realiza el modelado de un amplificador Doherty RTH21007-10 de RFHIC que trabaja con una señal LTE con frecuencia central de 2.1 GHz, ancho de banda de 5 MHz y PAPR de 7 dB. Se compara el error cuadrático medio normalizado (NMSE) obtenido con cada uno de las arquitecturas y se muestra la reducción de la no linealidad del amplificador, por medio de sus características AM-AM y AM-PM, al usarlos como modelos en la predistorsión digital. Por último, se presenta una tabla comparativa donde se muestra el desempeño de cada arquitectura de red neuronal artificial. Se observa que las redes neuronales recurrentes son capaces de reproducir mejor el comportamiento no lineal del amplificador, sin embargo, se logra una mayor reducción de la distorsión con la red feedforward.
Palabras Clave: Amplificador de Potencia, Modelado De comportamiento, Predistorsión Digital, Red Neuronal Artificial.
Abstract
This paper presents a comparison of several architectures of artificial neural networks, as a method of behavioral modeling, to perform digital predistortion of power amplifiers. The architectures of neural networks that are compared are RVFTDNN, FC2HLANN, M2HLANN and NARX, with which it is performed the modeling of a Doherty amplifier RTH21007-10 of RFHIC that works with an LTE signal with a central frequency of 2.1GHz, a bandwidth of 5MHz and PAPR of 7dB. The normalized mean square error (NMSE) obtained with each of the architectures is compared and the reduction of the non-linearity of the amplifier is shown, by means of its AM-AM and AM-PM characteristics, when used as models in digital predistortion. Finally, a comparative table is presented where the performance of each artificial neural network architecture is shown. It is observed that the recurrent neural networks are able to reproduce better the non-linear behavior of the amplifier, however, a greater reduction of the distortion is achieved with the feedforward network.
Keywords: Artificial Neural Network, Behavioral Modeling, Digital Predistortion, Power Amplifier.)
Texto completo:
103-118 PDFReferencias
Aguilar, L. M. Linealización de Amplificadores de Potencia Mediante un Modelo de Predistorsión Digital Basado en la Red Neuronal NARX, Ph. D Tesis, CINVESTAV. Guadalajara, Jal, México. 2015.
Bogya R. I. and Magana, M. E. Linear radio frequency power amplifier design using nonlinear feedback linearization techniques, IEEE 60th Vehicular Technology Conference, 2004. VTC2004-Fall. 2004, Los Angeles, CA, 2004, pp. 2259-2263 Vol. 3.
Lavrador, P. M. Cunha, T. R. Cabral P. M. and Pedro, J. C. The Linearity-Efficiency Compromise, in IEEE Microwave Magazine, vol. 11, no. 5, pp. 44-58, Aug. 2010.
Luongvinh, D. and Kwon, Y. Behavioral modeling of power amplifiers using fully recurrent neural networks, IEEE MTT-S International Microwave Symposium Digest, 2005, Long Beach, CA, 2005, pp. 1979-1982.
Mkadem, F. Ayed, M. B. Boumaiza, S. Wood J. and Aaen, P. Behavioral modeling and digital predistortion of Power Amplifiers with memory using Two Hidden Layers Artificial Neural Networks, 2010 IEEE MTT-S International Microwave Symposium, Anaheim, CA, 2010, pp. 656-659.
Mkadem F. and Boumaiza, S. Physically Inspired Neural Network Model for RF Power Amplifier Behavioral Modeling and Digital Predistortion, in IEEE Transactions on Microwave Theory and Techniques, vol. 59, no. 4, pp. 913-923, April 2011.
Nezami, M. Fundamentals of power amplifier linearization using digital pre-distortion, in High Frequency Electronics, Vol. 3, no 8, pp. 54-59, September 2004.
Shi, B and Sundstrom, L. Linearization of RF power amplifiers using power feedback, 1999 IEEE 49th Vehicular Technology Conference, Houston, TX, 1999, pp. 1520-1524 vol.2.
Xiang, T. and Wang, G. Doherty power amplifier with feedforward linearization, 2009 Asia Pacific Microwave Conference, Singapore, 2009, pp. 1621-1624.
URL de la licencia: https://creativecommons.org/licenses/by/3.0/deed.es
Pistas Educativas está bajo la Licencia Creative Commons Atribución 3.0 No portada. 
TECNOLÓGICO NACIONAL DE MÉXICO / INSTITUTO TECNOLÓGICO DE CELAYA
Antonio García Cubas Pte #600 esq. Av. Tecnológico, Celaya, Gto. México
Tel. 461 61 17575 Ext 5450 y 5146
pistaseducativas@itcelaya.edu.mx