Multisensor multitarget time delay vector estimation

This paper presents the results of an investigation of the optimum and suboptimum signal processors for passive time delay vector estimation in a multisensor, multitarget environment. Target-sensor geometry is assumed stationary over the observation interval and the signal spectra are assumed known. Using the maximum likelihood estimation (MLE) approach, the optimum time delay vector estimator is derived and its performance analyzed in terms of the Cramer-Rao lower bound (CRLB). Analytical closed-form expressions are obtained for the case of two targets with two sensors and one target with M sensors. The bias and variance characteristics of a conventional time delay processor in a multitarget environment are discussed; also presented are performance comparisons between the optimum and the conventional generalized cross-correlation (GCC) processor. In addition, a novel, suboptimum multitarget multisensor postcorrelation processor (MMPCP) is proposed. Its performance has been analyzed and verified through extensive simulation. The study of the optimum processor realization shows that it requires a coupled, multichannel processor. The required implementation is considerably more complex than the conventional GCC approach. The examination of the post-GCC multitarget processing capability shows that, in general, the MMPCP shows a marked improvement over the GCC estimator and that the performance degradation is acceptable with respect to the optimum processor.