A low-complexity method for signal subspace fitting is proposed based on the multi-stage Wiener filter(MSWF) can span a constringent signal subspace

which is completely equivalent to the Krylov subspace.Secondly

a new criterion function for signal subspace fitting is derived

and then a novel method for direction of arrival (DOA) estimation is developed.Analysis indicates that the proposed method can work very well even in the case where the rank of the MSWF is much lower than the number of signals.Since finding the constringent signal subspace merely involves the forward recursion of the MSWF

does not need to estimate the array covariance matrix or compute its eigenvectors

the proposed method is computationally efficient.Finally

the effectiveness of the proposed approach is verified by numerical results.