Reduced-order root-MUSIC based on Schur spectral factorization

Abstract

The root multiple signal classification (Root-MUSIC) has recently drawn a considerable attention, by using a polynomial rooting instead of spectral searching to reduce the complexity. The Root-MUSIC is computationally efficient in conjunction with a uniform linear array (ULA) composed of M sensors. Compared with traditional multiple signal classification (MUISC) algorithm, Root-MUSIC is more advantaged but also has a redundancy by solving a (2M-2) order polynomial. The polynomial is intensely complex when large number of sensors is used, and consequently, tremendous computations are required. A reduced-order Root-MUSIC based on the Schur spectral factorization is presented in this paper, which only need to calculate a (M-1) order polynomial. Simulations are conducted to support the validity of the algorithm. The results show that reduced-order Root-MUSIC has a similar root mean square error (RMSE) performance as Root-MUSIC with less computation.