Modeling Broad-band Power Amplifier Using Hybrid Time-Delay Neural Network

被引：0

作者：

Hui Ming ^{[1
,2
]}

Zhang Xingang ^{[1
,2
]}

Zhang Meng ^{[1
,2
]}

机构：

[1] Nanyang Normal Univ, Coll Phys & Elect Engn, Nanyang 473061, Peoples R China

[2] Nanyang Normal Univ, Intelligent Control Petr Equipment Henan Prov Eng, Nanyang 473061, Peoples R China

来源：

2018 CROSS STRAIT QUAD-REGIONAL RADIO SCIENCE AND WIRELESS TECHNOLOGY CONFERENCE (CSQRWC) | 2018年

关键词：

Power amplifier (PA); Neural network (NN); Linearization; modeling;

D O I：

暂无

中图分类号：

TM [电工技术]; TN [电子技术、通信技术];

学科分类号：

0808 ; 0809 ;

摘要：

A new hybrid time-delay neural network (TDNN) is proposed in this paper for modeling Broad-band power amplifier (BPA). This neural network includes the generalized memory effect of input signals, complex valued input signals and fractional order of complex valued input signal module, thus the modeling accuracy is improved significantly. To demonstrate the merits of this model, a 51-dBm BPA with a 25-MHz bandwidth mixed test signal is used for verification. Compared with real-valued TDNN, the hybrid TDNN is highly effective, leading to an improvement of 5 dB in the NMSE.

引用

页数：2

共 9 条

[1] Adaptive B-spline neural network based nonlinear equalization for high-order QAM systems with nonlinear transmit high power amplifier [J].