On the Use of Undersampled Signals for Power Amplifiers Linearization Using Digital Predistortion

This paper investigates the use of under-sampled feedback signals for the synthesis of digital predistortion functions. The use of under-sampled feedback signals typically requires a computationally complex signal reconstruction algorithm. In this work, it was shown that the computational complexity of such process can be significantly reduced by allowing for an imperfect reconstruction of the full-rate signal. This was found to have minimal impact on the linearization capability of the digital predistorter. Experimental validation on a commercial GaN based power amplifier was carried out using a 5G new radio having a bandwidth of 20MHz. The results show that up to 80% reduction in the computational complexity of the signal reconstruction algorithm can be achieved by allowing for a 1dB tolerance in the normalized mean squared error, which in turns results in a few dB degradation in the adjacent channel leakage ratio measured at the output of the linearized amplifier.