Digital Predistortion of Millimeter-Wave Arrays Using Near-Field Based Transmitter Observation Receivers

Abstract

This article proposes a novel method to train the digital predistortion (DPD) function in a radio frequency (RF) beamforming array using feedback signals acquired from a near-field-based transmitter observation receiver (NF-TOR). Accordingly, the proposed DPD training method relies only on information/signals available at the transmitter side and does not require the deployment of a far-field (FF)-based TOR. The method starts by first identifying the coefficients of the propagation channels between the radiating antennas and the NF probes. These NF channel coefficients are then used to estimate the forward model of each power amplifier in the array and to train the DPD function. Proof-of-concept validation experiments were conduthis article, the complex basebandcted on a 16-element 29-GHz RF beamforming array with four embedded NF probes. When driven by 400-MHz orthogonal frequency-division multiplexing signals with subcarriers modulated using 256-QAM, the DPD trained using the NF-TOR allowed for an improvement in the adjacent channel power ratio and error vector magnitude (EVM) from −29.66 to −40.49 dB and 8.29% to 2.40%, respectively, in the main beam direction. Furthermore, the application of the NF-TOR trained DPD allowed the effective isotropic radiated power to increase from 28.3 to 33.3 dBm while maintaining an EVM below 3.5%. These results, obtained using the proposed NF-TOR training method, are similar to those obtained with a DPD function trained using an FF TOR method.