Mixture of Experts Approach for Piecewise Modeling and Linearization of RF Power Amplifiers

Piecewise behavioral models are commonly adopted for modeling and linearization of RF power amplifiers (PAs) that exhibit strong amplitude-dependent nonlinear distortion characteristics, as global polynomial approximations tend to underperform in such scenarios. In this article, we consider a new piecewise model PAs based on the mixture experts (ME) approach, which builds probabilistic allows different submodels cooperate—as opposed operating an independent fashion is case existing reference methods. We first introduce ME framework theory while also extend it can be applied complex baseband signals nonlinearities. Then, show how overcoming some intrinsic shortcomings exhibit, translates into improved accuracy performance. Furthermore, extension approach tree-structured regression model, referred hierarchical introduced shown provide further performance improvements over basic approach. The proposed solutions validated with extensive measurements, covering both PA direct digital predistortion (DPD)-based linearization, gallium nitride (GaN) load-modulated balanced PA, GaN Doherty class AB high electron mobility transistor being compared against several state-of-the-art results demonstrate framework-based outperform state art.