Low Complexity Concurrent Multi-Band Modeling and Digital Predistortion for Harmonically Driven Wireless Amplifiers

With the cutting-edge 5G/6G signals and corresponding demands of the multi-band and ultra-wideband transmitters, numerous digital predistortion (DPD) models have been proposed for the removal of Power Amplifier’s (PA) distortions in concurrent multi-band and ultra-wideband communication. These demands result in challenges such as 5G/6G communication when frequency bands are harmonic or near-harmonic related. This paper proposes modeling approach and novel DPD model for removing PA’s nonlinearity, intermodulation distortions (IMDs), cross-modulation distortions (CMDs), and harmonic distortions for carrier aggregated harmonic related frequencies. This model is a neural network (NN) based and its input consists of basis corresponding to harmonic distortions, CMDs, and IMDs. The proposed NN DPD model is shallow, robust, and scalable to the growth in the dimension of multi-band transmission at harmonic frequencies, it delivers better removal of distortions with reduced complexity than the state-of-the-art DPDs. Furthermore, the proposed model also compensates for I/Q imbalance in the transmitting path. To test the robustness of the proposed shallow NN DPD, the performance is measured in adjacent channel power ratio (ACPR), normalized mean square error (NMSE) and error vector magnitude (EVM) for different 5G New Radio (NR) relevant signals scenarios.