Optimal linear filtering for power estimation of slowly-varying complex-valued signals

A novel method for designing the FIR filter in the Hammerstein model for power estimation of complex-valued signals is proposed. This enables a computationally effective optimal power estimation with a prescribed delay in minimum mean squared error sense, a great reduction of the estimation bias which is inherent in power estimation. Illustrations are given by assuming a constant input signal add a Rayleigh distributed input fading signal, respectively, both signals are complex-valued and corrupted by Gaussian noise.