2009文献TDOA estimation for cyclostationary sources--New correlations-based bounds and estimators.pdfVIP

TDOA ESTIMATION FOR CYCLOSTATIONARY SOURCES: NEW CORRELATIONS-BASED BOUNDS AND ESTIMATORS Moshe Teplitsky and Arie Yeredor School of Electrical Engineering, Tel-Aviv University P.O.Box 39040, Tel-Aviv 69978, ISRAEL E-mail: moshe.teplitsky@, arie@eng.tau.ac.il ABSTRACT gives rise to periodically-varying statistical characteristics, We consider the problem of Time Difference of Arrival and are therefore termed cyclostationary. The periodicity fre- (TDOA) estimation for cyclostationary signals in additive quency is termed cyclic frequency. The Fourier coefficient (at white Gaussian noise. Classical approaches to the problem the cyclic frequencies) of the time-varying correlation func- either ignore the cyclostationarity and use ordinary cross- tion (at each lag) are termed cyclic correlations. correlations, or exploit the cyclostationarity by using cyclic A family of TDOA estimation algorithms for cyclosta- cross-correlations, or combine these approaches into a mul- tionary sources was introduced by Gardner in [2], [3] and ticycle approach. Despite contradicting claims in the litera- used cyclic (rather than “stationary”) cross-correlations. The ture regarding the performance-ranking of these approaches, principal motivation behind the use of such algorithms is there has been almost no analytical comparative perfor- their inherent immunity against interference and noise sig- mance study. We propose to regard the estimated (ordinary nals which are either not cyclostationary, or do not share or c