A Computationally Efficient Predistortion and Segment Thresholding for Distributed PA Arrays

Multipath MIMO systems have emerged, providing faster and more reliable wireless communications. These systems require distributed arrays of nonlinear power amplifiers, which in turn require novel digital predistortion solutions. In this paper the authors propose a novel technique called, variation in the slope, to reduce the computational overhead of multipath segmented DPD for distributed arrays of power amplifiers. Additionally the approach to calculate the threshold for the segments requires only the second derivative calculation of gain versus input power and thus avoids computationally expensive clustering techniques. This approach was validated using a 100MHz 5G NR signal and a hybrid beam former module, resulting in an achieved -50dB ACLR on average. This value meets the spectral mask requirement for 5G FR2 standards. This method requires fewer training samples and DPD parameters compared to stateof-the-art techniques for multipath systems.