Residual Neural Networks for Digital Predistortion
Paper in proceeding, 2020

Tracking the nonlinear behavior of an RF power amplifier (PA) is challenging. To tackle this problem, we build a connection between residual learning and the PA nonlinearity, and propose a novel residual neural network structure, referred to as the residual real-valued time-delay neural network (R2TDNN). Instead of learning the whole behavior of the PA, the R2TDNN focuses on learning its nonlinear behavior by adding identity shortcut connections between the input and output layer. In particular, we apply the R2TDNN to digital predistortion and measure experimental results on a real PA. Compared with neural networks recently proposed by Liu et at. and Wang et at., the R2TDNN achieves the best linearization performance in terms of normalized mean square error and adjacent channel power ratio with less or similar computational complexity. Furthermore, the R2TDNN exhibits significantly faster training speed and lower training error.

Behavioral modeling

Power amplifiers

Radio Frequency amplifiers

Author

Yibo Wu

Ericsson

Chalmers, Electrical Engineering, Communication, Antennas and Optical Networks

Ulf Gustavsson

Ericsson

Alexandre Graell I Amat

Chalmers, Electrical Engineering, Communication, Antennas and Optical Networks

Henk Wymeersch

Chalmers, Electrical Engineering, Communication, Antennas and Optical Networks

2020 IEEE Global Communications Conference, GLOBECOM 2020 - Proceedings

9322327
9781728182988 (ISBN)

2020 IEEE Global Communications Conference, GLOBECOM 2020
Virtual, Taipei, Taiwan,

Deep RF

Swedish Foundation for Strategic Research (SSF) (DnrID19-0021), 2020-01-01 -- 2024-12-31.

Ericsson, 2020-01-01 -- 2024-12-31.

Subject Categories

Telecommunications

Communication Systems

Signal Processing

DOI

10.1109/GLOBECOM42002.2020.9322327

More information

Latest update

12/29/2021