PERFORMANCE ANALYSIS OF THE CYCLIC MUSIC METHOD OF DIRECTION ESTIMATION FOR CYCLOSTATIONARY SIGNALS

The Cyclic MUSIC algorithm exploits the cyclostationarity exhibited by most man-made communication signals to perform signal-selective estim ation of directions of arrival at an ai ray of sensors. In this paper, the bias and mean-squared error of these direction estimates are deri ved under the assumption that the signals are complex, possibly non-Ga ussian, and cyclostationary. The analytical results are shown to be ex cellent predictors of empirical results obtained from computer simulat ion of the Cyclic MUSIC algorithm, and are also compared with the Cram er-Rao Bound, which in certain cases can be orders of magnitude less t han the Cramer-Rao Bound for the directions of arrival of stationary s ignals.