Novel fast MUSIC algorithm for spectral estimation with high subspace dimension

Multiple signal classification (MUSIC) algorithm has been employed for many applications of frequency estimation, emitter localization, direction-of-arrival (DOA) estimation, etc. However, when the MUSIC algorithm is applied, a lot of computational resource is required to carry out the eigen-decomposition in order to extract the subspace information. In this paper, we would like to present a novel fast MUSIC algorithm. Since the data storage devices become less and less costly, spectral estimation of large dimensions appears crucial in modern telecommunication and signal processing applications. Our proposed computationally-efficient MUSIC algorithm, which can be facilitated in real time, would be very useful in the future. Our scheme is based on the fast eigen-decomposition method, and the computational complexity of our new technique is O (ρM²) (ρ ≪ M) compared to O (M³) of the conventional MUSIC algorithm when the size of the autocorrelation matrix of the received signal is M × M.