L-shaped array-based 2-D DOA estimation using parallel factor analysis

For two-dimensional (2-D) directions-of-arrival (DOA) estimation problem, the L-shaped array seems to have higher accuracy than other structured arrays (see [3] for details), and has received much attention. Many algorithms firstly estimate two electric angles separately using the two orthogonal subarrays of the L-shaped array, and then obtain elevation and azimuth angles from these two correctly matched electric angles. However, the failure in pairing will cause severe performance degradation. To avoid the performance degradation resulted from wrong pairing, a parallel factor (PARAFAC) analysis model-based algorithm is proposed in this paper to estimate 2-D DOA in the L-shaped array geometry without pairing parameters. The key points of this paper are: i) Dividing the whole L-shaped array into several subarrays, and constructing several matrices using the second-order statistics of some properly chosen array outputs to alleviate the noise effect and thus improve estimation accuracy; and ii) Forming a PARAFAC analysis model to avoid the performance degradation resulted from wrong pairing. Simulation results are presented to validate the performance of the proposed method.