An Online Sideband Suppression Architecture for PA With Strong Nonlinear Distortion

For power amplifier (PA) linearization under strong distortion scenarios, the inverse transfer function of conventional digital predistortion (DPD) might be unstable, and it would be quite difficult for DPD to work properly. In this letter, an online sideband suppression architecture is proposed for PA with strong distortion. In this architecture, an iterative learning control (ILC) algorithm is modified and unrolled into a K-stage cascade structure, which could be readily implemented for online applications. For predistorted signal generation, only the PA's out-of-band (OOB) distortion needs to be used for accurate model fitting, and the estimated model parameters are updated into the K-stage cascade structure. Moreover, in order to improve linearization performance, three parameter updating strategies are explored to track the PA's nonlinear feature and update the nonlinear models within this cascade structure. Finally, experimental results verify that the proposed method could significantly improve the adjacent channel leakage ratio (ACLR) performance by 22 dB when the PA is driven to -24-dBc ACLR.