A spectral decomposition method for the linear least-squares adaptive filters

A spectral decomposition method is presented for solving the least-squares problem of adaptive filters. Compared with the recommended singular value decomposition method which is defined by a set of singular-vectors and a corresponding set of nonzero singular values, the presented one is much simpler in the sense that it can be uniquely defined by the set of nonzero singular values, which means that the proposed approach has a higher level of computational accuracy than the recommended one because of the fewer intermediate parameters in this procedure.