Behavioral modeling of LMBA with different back-off state using PSO optimized XGBoost method

In this paper, an optimization technique based on the particle swarm optimization (PSO) algorithm is applied to the eXtreme gradient boosting (XGBoost) method for load modulated balanced amplifiers (LMBAs) modeling, taking into consideration both strong nonlinearity and memory effects. An overview of the basic principles of the proposed modeling technique is provided, as well as a detailed description of how the model is extracted. To improve the performance of the XGBoost model, the hyperparameters are optimized using the PSO algorithm. An in-house designed LMBA was used to perform experimental validation, which demonstrated that the new PSO-XGBoost model provided very efficient and extremely accurate predictions, especially in the case of strong nonlinearities. When compared to traditional Volterra models, canonical piecewise-linear based models, and standard XGBoost models, the proposed PSO-XGBoost model provides improved performance with reasonable complexity.