Sparsity Adaptive Estimation of Memory Polynomial Based Models for Power Amplifier Behavioral Modeling

作者：Li, Mingyu^*; Yang, Zhenxing; Zhang, Zhongming; Li, Ruoyu; Dong, Qing; Nakatake, Shigetoshi

来源：IEEE Microwave and Wireless Components Letters, 2016, 26(5): 370-372.

DOI：10.1109/LMWC.2016.2549024

• Behavioral modeling
• compressive sensing (CS)
• memory effect
• power amplifiers (PAs)
• Volterra series

In this letter, a new efficient model pruning method, called regularized sparsity adaptive matching pursuit (RSAMP), is presented to prune the redundant terms of memory polynomial based models for PA behavioral modeling. Unlike other CS based model pruning techniques, RSAMP method can estimate the sparsity level of the models adaptively with a regularized stagewise backtracking process. Using this method, the dominant sparse terms can be efficiently estimated based on the chosen supports without requiring the sparity level as an input parameter. Experiment results show that the RSAMP algorithm can efficiently construct a sparse behavioral model with very few terms, but almost have the same model performance with the full model.

• 出版日期2016-5
• 单位重庆大学