New algorithms for array processing using higher order statistics

Higher order statistical (HOS) methods for direction of arrival (DOA) estimation in array processing have many advantages over the conventional techniques. The authors introduce a set of algorithms for DOA estimation based on a new class of performance criteria related to the bispectra power of the data. The algorithms are embedded in a generalized framework and applied to narrowband sensor arrays. A precise formulation of the source signal model is given. Conditions are stated on the model and the assumed noise sufficient for a solution to the DOA problem to exist for any set of array outputs in the HOS-domain. The new performance criteria are described in relation to the cumulants of the array. Three new HOS-based algorithms are derived based on this criteria. The computational behavior of each of these is also addressed. Numerical simulation results for the algorithms are presented.<>