Sparse Estimation Technique for Digital Pre-distortion of Impedance-Mismatched Power Amplifiers

Abstract

This paper proposes the application of the Wiener–Hammerstein with feedback (WHFB) model as a digital pre-distortion (DPD) behavioural model for power amplifiers (PA) under load impedance mismatch, since more traditional models suffer performance degradation in this condition. Moreover, this paper proposes the use of the least absolute shrinkage and selection operator (LASSO) approach for the sparse, parsimonious estimation of the WHFB model. Using this technique, the number of coefficients is significantly reduced, thus reducing the DPD running complexity. Additionally, block-oriented LASSO extensions, such as group-LASSO and sparse-group LASSO, are proposed for model dimensioning, i.e. for setting parameters values. Finally, a simplified approximate technique is proposed, in which the most relevant blocks in the model are selected prior to running LASSO, resulting in lower estimation cost. Experimental results demonstrate the ability of the proposed techniques to adequately linearize PAs subject to load impedance mismatch.