2005文献Joint time-scale and TDOA estimation--analysis and fast approximation.pdfVIP

IEEE TRANSACTIONS ON SIGNAL PROCESSING, VOL. 53, NO. 8, AUGUST 2005 2625 Joint Time-Scale and TDOA Estimation: Analysis and Fast Approximation Y. T. Chan, Senior Member, IEEE, and K. C. Ho, Senior Member, IEEE Abstract— Relative motion (rm) between a signal source and a scale, time scaling, or time scaled, depending on the context. receiver causes a time scaling of the signal arriving at the receiver. The two receivers’ output become jointly nonstationary when When estimating the time-difference-of-arrival (TDOA) of a signal there is TS, with the consequence of a bias in the TDOA estimate at two receivers, time scaling, when not properly accounted for, from cross correlation. This bias dominates the estimator error can introduce a bias that could dominate the estimation errors. Segmentization processing cannot reduce this bias. Following the and increases with time, as Section II shows. derivation of the formulae for bias and mean-square errors of As a remedy, [7] partitioned each of the record lengths into TDOA estimation under rm, this paper moves on to the joint esti- segments and computed TDOA from the segments. Assuming mation of TDOA and time scale. It proposes an iterative search for that the TDOA is a known function of time (for example, linear the maximization of the cross-ambiguity function (CAF), which is also the maximum likelihood function for additive Gaussian or quadratic in time), a least-squares fit through the TDOAs bandlimited white noise disturbance. In addition, a quadratic from the segments will give the time dependen