Novel Fast Subspace Decomposition Using Lanczos Recursion

In this paper a new form of the initial vector is presented, and it is proved that in the context of space-time white noise the Krylov subspace composed of the initial vector and the covariance matrix of the observed signal is equivalent to the signal subspace, therefore the fast estimation of signal subspace can be performed only by computing the basis of the Krylov subspace with Lanczos recursions. By numerical simulation, it is clear that the method presented in this paper can perform the fast subspace decomposition efficiently and effectively.