BLIND CLOSED FORM PARAMETERS ESTIMATION FOR HYBRID SOURCES

A novel bind algorithm is proposed for ranges and direction-of-arrivals (DOAs) estimation of the hybrid narrow-band sources, which contain both the near field sources and the far field ones. To separate them, a two-step algorithm is exploited. In the first step, the far-field sources are identified by exploiting the MUSIC method. After eliminating the far field source components in the auto correlation matrix and adding the conjugate formulation, the dimensions of covariance matrix for near field sources are increased. Therefore, compared with the conventional methods, more independent near filed sources, near 4p/3 ones can be identified with a uniformly linear array (ULA) of 2p + 1 elements. Furthermore, in the near-field source estimation of our method, the peak search and pairing operations, needed in most algorithms for near field sources, can be omitted completely. Simulation results show that our ESPRIT-based near-field method provides the improved performance over conventional ones because of the dimensions increment of the covariance matrix.