Based on the asymptotic statistic characteristics of cyclic correlation matrices, an analysis of the performance of a class of cyclic weighted subspace fitting of directions estimation for cyclostationary signals is made. The analytic expressions of asymptotic estimation variance are derived, from which many interesting conclusions are drawn, e.g., estimation variance's relationship with the arrival angles of interference - the closer to the signal arrival angle, the smaller the estimation variance; and its relationship with cyclic correlation strength - the stronger, the better the estimation. Finally, the agreement of the results expected by the theory with those of computer simulation is shown.
展开▼