A novel DLA-Based Digital Predistortion Technique for Power Amplifier

Direct learning architecture(DLA) is a structure employed by digital predistortion techniques for Radio Freqency Power amplifier(RF PA), and is popular because of it high realtime capability and robustness. However, it suffers from the computational complexity and performance constraints caused by the extra PA modeling. In this paper, a novel DLA-based digital predistortion technique, which uses instantaneous complex gain(ICG) model to replace PA model in traditional DLA in order to avoid PA modeling, was proposed. This technique improves the adaptive algorithm for coefficients extraction and can greatly reduce the computational complexity with good linearization performance. The simulation results show that the computational complexity can be reduced by 50.03%.