Nullspace-Based Finite Rate of Innovation Algorithm to Resolve Multiple Signal Sources.

The finite rate of innovation (FRI) algorithms is a representative technique to find sparse signal sources covered by a sampling kernel. It is widely used in various applications such as angle-of-arrival estimation, angle of departure estimation, multipath delay estimation, and edge detection. However, it has a weakness of performance degradation when an interval of signal sources is small. To improve the resolution, a nullspace-based algorithm is proposed in this paper. The proposed algorithm is based on the orthogonality between signal subspace and noise subspace like the multiple signal classification (MUSIC) algorithm, but evaluates the orthogonality by using overfitted annihilating filters. To reduce the amount of computation, a parameterization method is also introduced instead of finding the peaks of the spectrum.