On frequency-wavenumber estimation by state-space realization

A frequency-wavenumber estimation problem usually requires a computationally expensive numerical search over a 2-D parameter space for the maximum or minimum of a function. A method using state-space models has been proposed. This method decomposes the 2-D problem into two 1-D problems and provides an analytical solution (i.e., without searching) in each dimension separately. It is shown how the separate estimates obtained by the state-space approach can be grouped into pairs of estimates corresponding to each 2-D signal. It is also shown how a 2-D forward-backward data matrix can be used to improve the performance of the state-space approach