Highly Accurate Two-Dimensional Direction Finding Using Cumulant

This paper proposes a new two-dimensional direction finding algorithm for estimating multiple non-Gaussian signals using cumulants. This new algorithm: 1) employs an array geometry consisting of a nine-element non-uniformly spaced cross-shaped sub-array plus another arbitrarily-spaced sub-array; 2) defines five cumulant matrices, from which the signal direction cosines can be estimated by using parallel factor fitting, and hence, requiring no open-form two-dimensional searching and parameter pairing; 3) can achieve array aperture extension by jointly extending array physical aperture and constructing virtual sensors, without adding more sensors; and 4) can resolve 2(L - 4) signals with an array of L sensors in total. Simulation results are presented to verify the efficacy of the proposed algorithm.