ENHANCED RESOLUTION BASED ON MINIMUM-VARIANCE ESTIMATION AND EXPONENTIAL DATA MODELING

In various signal processing applications, it is desired to appropriat ely modify a given data set so that the modified data set possesses pr escribed properties. The modification of the given data set serves as a preprocessing step of 'cleaning up' the data before estimating the V alues of the signal parameters. In this paper, evaluation and improvem ent of a signal enhancement algorithm, originally proposed by Tufts, K umaresan and Kirsteins and recently generalized by Cadzow, are present ed. In essence, the newly proposed algorithm first arranges the data i n a very rectangular (instead of a square) Hanker structured matrix in order to make the corresponding signal-only data matrix orthogonal to the noise, then computes a minimum variance (instead of a least squar es) estimate of the signal-only data matrix and finally restores the H ankel structure of the computed minimum variance estimate. An extensiv e set of simulations is given demonstrating a significant improvement in resolution performance over Cadzow's method at a comparable paramet er accuracy. Moreover, arranging the data in a very rectangular matrix reduces drastically the required computation time. In particular, the newly proposed signal enhancement algorithm can be successfully appli ed to the quantitative One-domain analysis of Nuclear Magnetic Resonan ce (NMR) data.