Complexity-Optimized Vector-Switching Digital Predistortion Using Basis-Propagation Selection for RF Power Amplifiers

Paper i proceeding, 2025

A novel digital predistortion (DPD) framework is developed by integrating the vector-switched (VS) modeling approach with the Basis-Propagation Selection (BAPS) model to enable highly efficient and accurate linearization of nonlinear radiofrequency (RF) power amplifiers (PAs). In the proposed VS-BAPS architecture, the input signal space is partitioned into multiple regions via clustering on input amplitude statistics, and within each region, an independent BAPS model is identified using a greedy basis construction algorithm. This structure exploits the adaptive local modeling capability of the VS model while incorporating the complexity-optimized basis generation of the BAPS approach. Experimental evaluations demonstrate that the VS-BAPS model achieves superior linearization performance in terms of normalized mean square error (NMSE) and adjacent channel power ratio (ACPR), while significantly reducing computational complexity compared to conventional generalized memory polynomial (GMP), vector-switched GMP (VS-GMP), and standalone BAPS models.