THREE-LAYER PERCEPTRON VERSUS RADIAL BASIS FUNCTION FOR THE LOW-PASS EQUIVALENT BEHAVIORAL MODELING OF WIRELESS TRANSMITTERS
Keywords:Bandpass systems, digital baseband predistortion, modeling, radial basis function, three layer perceptron, wireless communication systems
This paper addresses the low-pass equivalent behavioral modeling of wireless transmitters using artificial neural networks (ANNs). Radial basis function (RBF) and three-layer perceptron (TLP) are the most widely adopted ANN architectures. In wireless transmitter behavioral modeling literature, it is reported the effective use of either TLP or RBF. The contribution of this work is to present a comparative study between TLP and RBF in terms of computational complexity and modeling accuracy. In here, the basic requirements that the ANN-based models must comply with to properly model the behavior of bandpass systems is first identified. Then, both TLP and RBF are applied to the low-pass equivalent behavioral modeling of three different wireless transmitters: a GaN HEMT class AB, a Si LDMOS class AB and a GaN HEMT Doherty. For the three studied transmitters, RBF models trained by the orthogonal least squares algorithm do not provide acceptable modeling accuracies, while TLP models trained by the back-propagation algorithm do produce highly accurate models. For a fair comparison between TLP and RBF models, all models are then trained by the same nonlinear optimization tool based on the Gauss-Newton algorithm with line search. Indeed, by changing the training algorithm, the modeling accuracy of RBF models is significantly improved for the three studied transmitters. In the one hand, for the GaN HEMT Doherty transmitter in a scenario of similar number of network parameters, it is verified that the TLP provides a significant higher accuracy than the RBF, illustrated by improvements in NMSE and ACEPR metrics by up to 8.4 dB and 8.7 dB, respectively. On the other hand, for the GaN HEMT class AB and Si LDMOS class AB transmitters in a scenario of similar number of network parameters, the TLP and RBF modeling accuracies are very similar, although a slight superior performance is observed for TLP models, quantified by up to 0.7 dB and 1.5 dB improvements in NMSE and ACEPR metrics, respectively.
no.4, pp.824-840, Apr. 2012.
 W. Chen, S. Zhang, Y.-J. Liu, F. M. Ghannouchi, Z. Feng, and Y. Liu, "Efficient Pruning Technique of Memory
Polynomial Models Suitable for PA Behavioral Modeling and Digital Predistortion," IEEE Trans. Microw. Theory
Tech., vol.62, no.10, pp.2290–2299, Oct. 2014.
 G. Xu, T. Liu, Y. Ye, T. Xu, H. Wen, and X. Zhang, "Generalized Two-Box Cascaded Nonlinear Behavioral Model for
Radio Frequency Power Amplifiers With Strong Memory Effects," IEEE Trans. Microw. Theory Tech., vol.62, no.12,
pp.2888–2899, Dec. 2014.
 H. Raab, P. Asbeck, S. Cripps, P. B. Kenington, Z. B. Popovic, N. Pothecary, J. F. Sevic, and N. O. Sokal, "Power
amplifiers and transmitters for RF and microwave," IEEE Trans. Microw. Theory Tech., vol.50, no.3, pp.814–826, Mar.
 S. Cripps, RF Power Amplifiers for Wireless Communications, 2nd edition. Norwood, MA: Artech House, 2006.
 P. B. Kenington, High Linearity RF Amplifier Design. Norwood, MA: Artech House, 2000.
 J. C. Pedro and S. A. Maas, “A comparative overview of microwave and wireless power-amplifier behavioral modeling
approaches,” IEEE Trans. Microw. Theory Tech., vol. 53, no. 4, pp. 1150–1163, Apr. 2005.
 S. Haykin, Neural Networks: a Comprehensive Foundation, 2nd edition. New Jersey: Prentice Hall, 1999.
 T. Liu, S. Boumaiza, and F. M. Ghannouchi, “Dynamic behavioral modeling of 3G power amplifiers using real-valued
time-delay neural networks,” IEEE Trans. Microw. Theory Tech., vol. 52, no. 3, pp. 1025–1033, Mar. 2004.
 E. G. Lima, T. R. Cunha, and J. C. Pedro, “A Physically Meaningful Neural Network Behavioral Model for Wireless
Transmitters Exhibiting PM–AM/PM–PM Distortions”, IEEE Trans. Microw. Theory Tech., vol. 59, no. 12, pp. 3512–
3521, Dec. 2011.
 X.-H. Yuan and Q. Feng, “Behavioral modeling of RF power amplifiers with memory effects using orthonormal
Hermite polynomial basis neural network”, Prog. Electromagn. Res. C, vol. 34, pp. 239–251, 2013.
 K. Fu, C. L. Law, and T. T. Thein, “Novel Neural Network Model of Power Amplifier Plus IQ Imbalances”, Prog.
Electromagn. Res. B, vol. 46, pp. 177–192, 2013.
 L. B. C. Freire, C. França, and E. G. Lima, “Low-pass equivalent behavioral modeling of RF power amplifiers using
two independent real-valued feed-forward neural networks”, Prog. Electromagn. Res.C, vol. 52, pp. 125–133, 2014.
 M. Isaksson, D. Wisell, and D. Ronnow, “Wide-band dynamic modeling of power amplifiers using radial-basis function
neural networks,” IEEE Trans. Microw. Theory Tech., vol. 53, no. 11, pp. 3422–3428, Nov. 2005.
 M. Li, S. He, and X. Li, "Complex radial basis function networks trained by QR-decomposition recursive least square
algorithms applied in behavioral modeling of nonlinear power amplifiers," Int. J. RF Microw. C. E., vol. 19, no. 6, pp.
634–646, Nov. 2009.
 S. Chen, C. F. N. Cowan, and P. M. Grant, “Orthogonal least squares learning algorithm for radial basis function
networks,” IEEE Trans. Neural Netw., vol. 2, no. 2, pp. 302–309, Mar. 1991.
 J. Dennis Jr., “Nonlinear least-squares,” in State of the Art in Numerical Analysis: Conference Proceedings, D. A. H. Jacobs, Ed.
London: Academic Press, 1977, pp. 269–312.
 M. S. Muha, C. J. Clark, A. Moulthrop, and C. P. Silva, “Validation of power amplifier nonlinear block models,” in
IEEE MTT-S Int. Microwave Symp. Dig., Anaheim, CA, Jun. 1999, pp. 759–762.
 M. Isaksson, D. Wisell, and D. Ronnow, “A comparative analysis of behavioral models for RF power amplifiers,” IEEE
Trans. Microw. Theory Tech., vol. 54, no. 1, pp. 348–359, Jan. 2006.
 A. V. Oppenheim and R. W. Schafer, Discrete-time Signal Processing, 3rd edition. Upper Saddle River, NJ: Pearson
Higher Education, 2010.