On the elimination of extraneous roots

In this paper we develop a technique to eliminate extraneous roots from reduced rank, linear predictive frequency and direction of arrival (DOA) estimation algorithms. These singular value decomposition (SVD) based algorithms produce a noise cleaned linear prediction vector and then root this vector to obtain a subset of roots, whose angles contain the desired frequency or DOA information. The roots closest to the unit circle are deemed to be the "signal roots". All the rest of the roots are "extraneous". The extraneous roots are expensive to calculate and complicate the extraction of the signal roots.