Source localization in a multipath environment using random or deterministic signals

The paper develops a general theory of array processing in a multipath environment. It works with sources radiating bandlimited Gaussian or deterministic signals in confined spaces such as an underwater acoustic channel. Cramer-Rao bounds on the mean square error of source location parameter estimates are used to evaluate the effects of signal properties, sensor array geometry, source location parameters and propagation conditions on localization accuracy. General expressions are calculated for the minimum mean square error of source location parameters in terms of a generalized post-beamforming signal to noise ratio beta /sub k/ at frequency omega /sub k/ and the vectors of transfer functions from the source to various sensors. A comparison is made between the location accuracy attainable with random and with deterministic signals. Also explored are the effects on location accuracy of array geometry, possible uncertainties in propagation conditions and uncertainties in array orientation.<>