A novel algorithm for 2-dimensional frequency estimation

Abstract

In this paper, an algorithm for 2-D frequency estimation is proposed. This algorithm consists of two parts. First, two sets of 1-D distinct frequencies are estimated separately. Then, the estimates of the 2-D frequencies are obtained by pairing the components in the two sets of estimated 1-D frequencies. The correct pairing is accomplished based on the fact that a vector in the signal subspace is orthogonal to the noise subspace of a 2-D signal. The capability of determining the number of signals is inherited in the proposed algorithm. Computer simulation results are provided to demonstrate the efficacy of the algorithm.<>