MULTIPLE-TARGET DIRECTION FINDING (ARRAY SIGNAL PROCESSING, RADAR, SONAR, DETECTION, ESTIMATION)

This thesis addresses the problem of direction finding with a sensor array in a multiple-target environment. Determination of the number of targets and estimation of their directions of arrival are the main concerns in this study. The basic approach that is taken is parametric model identification with emphasis on the use of the method of signal-subspace processing.;A simple method for wideband active direction finding and measurement of distances is also presented. With the use of the waveform information of the transmitted signal, a two-sensor wideband active system can be constructed for location of more than two targets. A signal-subspace processing algorithm is given and its performance examined. Unlike the traditional two-sensor systems based on the time-delay estimation or the triangular geometrical calculation using distance measurements from each sensor, the new system can handle correlated targets and does not have the inherent ambiguity problem with proper design.;Performance of narrowband signal-subspace processing is examined in detail. Analytical expressions for evaluation of detection and estimation performances are derived for the case of closely spaced two targets in the spatially white noise. For signal-subspace processing in wideband passive direction finding, a coherent algorithm is developed and its performance compared with available incoherent signal-subspace processing algorithms. Both detection and estimation performances are shown, by theoretical analysis as well as simulations, to be significantly better than those offered by incoherent ones, especially when the signal-to-noise ratios are low and/or sources are correlated.