Develops algorithms based on higher-order moments and cumulants to estimate the parameters of a linear or harmonic process which is corrupted by multiplicative noise. The observations may be further corrupted by additive colored Gaussian noise. It is shown that parameter estimation is fairly easy when the multiplicative noise is i.i.d. and has nonzero mean. The colored Gaussian and zero-mean i.i.d. multiplicative noise cases are also considered.<
Published in:
Spectrum Estimation and Modeling, 1990., Fifth ASSP Workshop on
Date of Conference: 10-12 Oct. 1990