An Out-of-Band Linearization Architecture for PA With Strong Nonlinearity

For the strong nonlinear distortion scenarios, the power amplifiers (PAs) are driven into the strong nonlinear region, and the inverse of the PAs obtained through the conventional full-band (FB) digital predistortion (DPD) might be unstable, which would fail to correct the nonlinear distortion. In this letter, a novel out-of-band (OOB) linearization architecture is proposed for PA with strong nonlinear distortion. In the proposed architecture, an out-of-band iterative learning control (OILC) training algorithm is proposed to minimize the OOB nonlinear distortion, which reduces the compensated peak amplitude to focus on linearizing the OOB components. An out-of-band generalized memory polynomial (OGMP) model is proposed to further optimize the accuracy of the model identification by removing the irrelevant terms, which would be a perfect match with the desired OOB DPD signal generated by the OILC training algorithm. Finally, experimental results verify that the proposed architecture could improve the adjacent channel power leakage ratio (ACLR) performance by 18 dB when the PA is driven to $-$ 28 dBc ACLR.