A novel model for digital predistortion based on a gravitational search algorithm for linearization of transmitters in LIE networks

摘要

A high-performance model with a gravitational search algorithm (GSA)-based generalized parallel two-box (GPTB) structure is suggested for digital predistortion in modern transmitters exhibiting memory effects, where the GSA is applied to identify the minimum dimension for the GPTB model. An indirect learning structure in conjunction with the GSA method is employed to identify the coefficients of the GSA-based GPTB model. The GPTB-GSA method is verified using simulations of a transmitter excited by quadrature amplitude modulation (QAM) signals in ADS software and simulations of the GSA in MATLAB software. The MATLAB results demonstrate the ability of the GSA to determine the dimension of the GPTB model efficiently. Also, the adjacent channel power ratio (ACPR) measure is decreased by about 16 dB according to the simulation. The proposed model and algorithm can reduce the number of coefficients by approximately 25% in comparison with the memory polynomial model.