MUSIC algorithm based on eigenvalue clustering

The traditional MUSIC algorithm needs to know the number of target signal sources in advance, and further determine dimensions subspace noise subspace, finally search for spectral peaks. In engineering, it is impossible predict be measured. To solve above-mentioned problem, an improved without estimating proposed. present algorithm, all eigenvectors covariance matrix are regarded as estimation, but existence will make result unreliable. order estimation more accurate, a new weighting method results simulation show that can accurately estimate direction when unknown, has greater practicability than algorithm. addition, better robustness.