A novel approach for 2D-DOA estimation using cross-shaped arrays

DOA-based beamforming is a smart antenna technique that utilizes the spatial signatures of signals arriving at the base station of a mobile network to estimate the DOA of each signal and its multipath replicas. Then this information is used to form adaptive beams that have main lobes in the intended directions and nulls in the directions of interferes or multipath. According to (Elzooghby, 2005) and the references there in, the method of directing receiving and transmitting beams in the direction of the estimated DOAs has the potential to be successful, even in FDD systems where the uplink and downlink utilize different frequencies. The main part of such systems is the DOA estimation algorithm that can spatially resolve a composite signal consisting of signals coming form different users and their multipath replicas. An enormous number of DOA estimation algorithms are reported in the literature. On developing these algorithms, uniform linear arrays of isotropic elements are always assumed, and the algorithm developed is to find the angle between the propagation vector of the arriving signal and the array line in the plane containing both. Then these algorithms are generalized using planar or circular arrays to be implemented in 3D-space where the azimuth and elevation angles are required for source localization. In this case, the generalized algorithm is named 2D-DOA estimation since it searches not only for a single angle of arrival but it searches for both of the azimuth and elevation angles. The use of 1D-DOA estimation in a cellular system, where sources of signals cannot be assumed coplanar, represents a great source of ambiguity. Thus, it is very important to use 2D-DOA estimation algorithms. However, the main drawback of the 2D-DOA generalization is that the computational complexity increases substantially with the number of elements. This makes the implementation in real-time operation not feasible. In this paper a simple approach for 2D-DOA generalization i- - s presented depending on a cross-shaped array. This array consists of two intersecting ULAs working independently, thus the computational complexity is only double that of a single dimensional array. The ambiguity resulting in a cellular system when using single dimension DOA estimation is discussed in the next section, while the proposed algorithm and its validation are presented in the rest of the paper.