Toward Interpretable Digital Predistortion: A Frequency-Domain Behavioral Approach
Abstract
This article presents a frequency domain behavioral modeling approach. It introduces the foundations of the Volterra series, the frequency domain parameterization of Volterra kernels, and a comparison between the seRies expansiOn of the Volterra Approximation (ROVA) model and memory polynomials. The proposed modeling framework centers on two interpretable hyperparameters: the order of the Volterra kernels, which defines the degree of nonlinearity, and the complexity of the local polynomial approximation used to capture the frequency dependence of the kernels around the carrier frequency. Together, these elements enable a compact, efficient, and bandwidth limited modeling technique.
https://ieeexplore.ieee.org/document/11130603