Abstract
This work reports expressions for different parameters constituting the main support for convergence of the signed regressor least mean mixed-norm (LMMN) algorithm for complex-valued data. The steady-state mean-square error, the optimum step-size, and the corresponding minimum value of the tracking mean-square error are all derived. Simulation results are conducted to corroborate the theoretical findings. Also, the convergence bahaviour of the signed regressor LMMN algorithm and that of the LMMN algorithm are compared.