An Improved Digital Predistortion Mechanism via Joint Baseband and Radio Frequency Optimization

Predistortion techniques can be categorized into digital predistortion (DPD) and analog predistortion (APD), which are always implemented separately in their own domains. The baseband-domain mean square error (MSE) is the design target of DPD, and the RF-domain inter-modulation distortions (IMDs) are applied for APD. In this letter, we propose a hybrid MSE-IMDs based DPD framework that optimizes the above two criteria simultaneously. Specifically, to make the joint optimization possible, we define a new IMDs metric which can capture the DPD model coefficients. The classic recursive least square (RLS)-based DPD scheme is improved to balance the MSE and IMDs metrics without increasing the complexity.