Analysis of derivative MUSIC with two correlated or uncorrelated sources and its extension to a planar array

This thesis presents a novel spatial spectrum estimation technique, ∂-MUSIC, for discriminating between two closely spaced sources which are highly correlated. The ∂-MUSIC algorithm is tested, modified, and compared to the MUSIC algorithm using a point source simulation. Various power levels, samples sizes and angle separations are used on a linear and a planar array for correlated and uncorrelated sources. The algorithm is found to be relatively insensitive to correlation and can separate targets to one-half of the angular separation threshold of ∂-MUSIC. The ∂-MUSIC algorithm is tested using a simulation that generated terrain scattered interference representative of a propagation scenario involving multiple paths. The simulation shows that ∂-MUSIC is able to resolve the direct path and image at less than one-fourth of a beam width, with a ten degree angle to the surface, whereas MUSIC finds a single angle which is biased toward the image