Maximum likelihood two-sensor detection and TDOA estimation for cyclostationary signals

Summary form only given. It has been shown that the maximum-likelihood two-sensor detector for a random signal in white Gaussian noise contains terms corresponding to both generalized cross correlation and cyclic methods of detection and time-difference-of-arrival (TDOA) estimation. The relative importance of each of these terms has been studied, and their robustness against nonwhite and nonGaussian noise and interference backgrounds has been investigated. Computer simulations have been carried out for several types of corruptive environments. Empirical receiver operating characteristics have been used to measure detection performance of the various methods, and the quality of the TDOA estimates has been measured in terms of bias and variance. The Cramer-Rao lower bound (CRLB) for the cyclostationary model has been evaluated and compared with that for the stationary model