Nonlinear RLS algorithm for amplitude estimation in class A noise

An adaptive nonlinear recursive least square (RLS) algorithm for amplitude estimation in class A noise is presented. For Gaussian input signal and class A noise, its mean and mean-square behaviours are studied. It is shown that the linear RLS and nonlinear RLS algorithm with the clipper function are stable in the mean and mean square. For non-Gaussian input, amplitude estimation in CDMA communication is presented. Simulation results show that the nonlinear RLS can provide good performance close to the Cramer-Rao bound and outperform the nonlinear LMS and the conventional RLS in impulse noise.