Multisignal spectral estimation using an iterative relative entropy method

A generalization of a method of spectral estimation by M.A. Tzaness, Ch. Moussas, and N.S. Tzannes (1984) is presented. The algorithm extends the method to the multiple signals case while still using an iterative procedure of the relative entropy method for estimating the spectral densities of the multisignal input data. Several current methods of relative entropy spectral estimation which involve solving N nonlinear equations simultaneously are also presented. The performance of the presented algorithm is compared to the maximum-entropy and other relative entropy methods, in terms of precision and computational efficiency. Classical examples are repeated and compared between the different methods.<
>