Digital Linearization of Multiple Power Amplifiers in Phased Arrays for 5G Wireless Communications

A novel digital predistortion (DPD) architecture is proposed to linearize power amplifiers (PAs) in a phased array applicable to the wireless communication systems. First, the composite signal of the array outputs over the air is captured by an extra observation chain deployed in this architecture. Although the observed signal may differ from the composite signal of the array pointing to an arbitrary target, it contains all the nonlinearities of the PAs inside the array. Thus, the nonlinearities of the PAs in the array can be simply characterized from the observed signal by using a single nonlinear model. Based on this nonlinear model, an indirect learning method is then adopted to estimate the reverse nonlinear model coefficients, i.e., the DPD model coefficients, for linearization of the PA array. Moreover, a testbed is designed with a PA array composed of four identical PAs and experiments are performed on a 256-QAM waveform of 100-MHz bandwidth to evaluate the effectiveness of the proposed method. Experimental results have validated that the adjacent channel power ratio is improved from -31 dBc to -48 dBc and the error vector magnitude is improved from 5% to 0.8%.