Dice Coefficient Matching-Based Sparsity Adaptive Matching Pursuit Algorithm for the Digital Predistortion Model Pruning

In this paper, an efficient Dice coefficient matching-based sparsity adaptive matching pursuit (DSAMP) algorithm is introduced to prune the redundant terms of power amplifier (PA) digital predistortion (DPD) models. The atoms matching criterion based on Dice coefficient is proposed for the DSAMP algorithm, which can use the function of its important component vector for quickly locating residual signal main component, and replace traditional rule of inner product similarity measure. Compared with the other sparsity pruning algorithm, DSAMP algorithm can get faster run speed. A Class_F PA excited by 10MHZ long-term evolution LTE signals is used as a proof of the algorithm capabilities. Experimental results show that the DSAMP algorithm can prune the coefficient of the PA DPD models dramatically, while the sparse DPD model almost possesses the same performance with the full model.