A lattice-based memory polynomial model for nonlinear MIMO transmitter behavioral modeling using fixed point arithmetic

Recently, there has been an accelerating trend towards the deployment of multiple-input multiple-output (MIMO) architectures for communication systems. While these architectures greatly enhance the channel capacity and the data rates delivered to users, they suffer from issues such as the presence of crosstalk and leakage between the signal paths. The presence of such effects greatly increases the number of coefficients required by behavioral models describing MIMO transmitters, which in turn exacerbates the ill-conditioning problem encountered in the coefficient-extraction process. This places a demand for a higher number of bits to be used when implementing digital predistorters (DPD) based on these models. To address this problem, a low-complexity lattice filter is proposed in this work to reduce the conditioning of the behavioral models used. Through experimental validation using a 2×2 MIMO transmitter setup, it was found that the proposed method greatly reduces the condition number of the model used and enables satisfactory modeling using only 16 bits.