Cyclostationarity-based coherent methods for wideband-signal source location

We propose a new approach for source localization of wideband signals impinging on an array of sensors. The proposed approach is signal-selective, i.e., it exploits the cyclostationarity property, exhibited by most communication signals, to discriminate signals of interest from noise and interfering ones. A coherent combination of the spatial contributions at different temporal frequencies is performed, and signal-subspace properties of the resulting focused matrix are exploited to obtain high-resolution source localization. Numerical results show that practical focusing techniques based on the new approach are superior to existing cyclic algorithms, also assuring good performances when the latter fail as, for example, when the signals of interest are fully correlated.