Concurrent estimation of signal parameters using extended rational function model

A frequency domain based parallel processing scheme for fast estimation of signal parameters from noisy signals is presented. In this scheme, the entire DFT (discrete Fourier transform) based spectrum is divided into several segments. The samples of the transfer function within different segments are modeled independently as rational function models so that they can be processed concurrently in a multiprocessing environment. In this investigation, this new signal parameter estimation scheme was implemented on an Intel iPSC/2 hypercube. Results of this new method are compared with a conventional frequency domain estimation method.<
>