This fundamental result, although often skipped in spectral analysis books, is a reason why the input signal can be distributed into signal subspace eigenvectors spanning ( for real valued signals) and noise subspace eigenvectors spanning . It is based on signal embedding theory and can also be explained by the topological theory of manifolds.

Comparison to other methods

MUSIC outperforms simple methods such as picking peaks of DFT spectra in the presence of noise, when the number of components

1.Previous
3.Next