A Low Complexity Hybrid Memory Polynomial Model for Digital Predistortion of RF Power Amplifier

In this paper, a low complexity hybrid memory polynomial (LCHMP) model based on the k-means++ algorithm and the error variation ranking (EVR) pruning method is proposed to reduce the running complexity of the digital predistortion (DPD) model. The DPD models of different complexity are first selected for modeling by classifying the input signals, and then each model is effectively pruned to achieve a reduction in complexity. Comparing the proposed model with existing models, the experimental results show that the LCHMP model has a 67.2% reduction in operational complexity compared to the traditional generalized memory polynomial (GMP) model, while the degradation in linearization performance is negligible.