High-resolution spectrum estimation for non-coherent source location using the multi-frequency virtual correlation matrix

We explain how multiple frequencies can be used to extend the capabilities of the spectrum estimation method for high-resolution location of multiple non-coherent point sources. The techniques covered in this paper are MUltiple SIgnal Classification (MUSIC), Autoregressive (AR) spectrum estimation and Capon minimum variance spectrum estimation. Our approach uses coarray equivalence to construct a ldquovirtual correlation matrixrdquo with larger size than the correlation matrix that is directly obtainable from the physical array. The larger dimension of the virtual correlation matrix allows for more point sources to be resolved when the spectrum estimation methods are applied. This method will prove to be useful in situations where the number of antenna elements must be limited, but there are many sources.