Efficient algorithm for 2-D DOA estimation based on noise subspace mapping

To reduce the computational complexity of the two dimensional Multiple Signal Classification (2-D MUSIC) algorithm and make it suitable for real-time applications, this paper presents a new computationally efficient method for 2-D direction-of-arrivals (DOA) estimation with arbitrary 2-D array configurations based on noise-subspace mapping. Exploring the idea of spatial angle dividing and non-linear transformation, the orthogonal relationship between the signal-subspaces and noise-subspaces is compressed to a small angular sector, leading to a series of virtual mirrors for each true DOA in a given sector. This allows fast estimation for the virtual DOAs by spectral search over only one sector, which further gives the value of the true DOAs since they are mathematically related. It is shown by theoretical analysis as well as experimental results that the new approach has a much lower computational complexity and an improved resolution as compared to the standard MUSIC.