Adaptive Tracking Digital Predistortion for RF Power Amplifiers Under Variable Bandwidth and Power Transmission

In this paper, a low-complexity adaptive tracking digital predistortion (DPD) method is proposed for the emerging intelligent communication transmitters operating under variable bandwidth and power conditions. To reduce the number of DPD coefficients, a block-level contribution-driven pruning (BCP) algorithm is employed to dynamically select and activate optimal combinations of basis function blocks (CBFB) according to the bandwidth condition. Furthermore, a novel parameter sensitivity recognition method is employed, along with an adaptive DPD coefficient extraction and tracking scheme, to identify the key basis functions that can reflect the dominant nonlinear features, and adjust their coefficients via fine-tuning factors. This enables the DPD coefficients to adapt to the variations in PA’s nonlinearity caused by dynamic conditions without real-time recalibration. Experiments carried out on a 3.55 GHz Doherty PA demonstrate that the proposed method exhibits significant robust performance across wide bandwidths and varying power levels.