Identification of non-skewed MA processes using a fourth-order cumulant-based approach
Diego Pablo Ruiz, D. Blanco, Maria Carmen Carrion · 2003
The purpose of this communication is the formulation of three new linear algorithms for blind non-minimum phase MA system identification. These methods have been derived starting from new equations involving system coefficients and q-slices of the k-th order cumulant sequence of the output MA system. In particular, these new algorithms use only fourth-order cumulants, and thus, are especially useful when the driving system noise has a symmetric probability density function and a possible additive Gaussian noisy process contaminates the system output.