Fourth-order cyclic cumulant TLS-ESPRIT algorithm to estimate direction of cyclostationary coherent sources

In this paper, considering that most signals exhibit cyclostationarity and coherent sources caused by multipath propagation are often encountered in a variety of wireless communication systems, we propose a direction-finding algorithm based on cyclic cumulants to extract the desired non-Gaussian cyclostationary coherent sources by designing a fourth-order cyclic cumulant matrix and using TLS-ESPRIT algorithm. To coherent sources, a 'de-correlating' method is exploited in our cyclic cumulant matrix model. Simulation results show that not only does the approach possess signal-selectivity capability but can effectively suppress additive stationary noise and Gaussian noise in environments where the spatial characteristics of noise are unknown, even when the noise shares the same cycle frequency as the signals of interests (SOI's). Moreover, under a lower signal to noise ratio (SNR) and a smaller sampling number, it has ability to detect and resolve more coherent and incoherent sources using fewer sensors.