Estimating direction of arrival with one-dimensional spectral analysis

DOA (direction of arrival) estimation plays a key role in radar and sonar signal processing. Compared with one-dimensional spectral analysis, DOA is more difficult because spatial information is mixed up with temporal information. Temporal coherence between sources usually lowers the performance of high-resolution DOA estimators. To overcome difficulties caused by coherence, some cost in computation or array aperture is often involved. In this paper, a method is presented to reduce the two-dimensional DOA estimation to a problem of one-dimensional frequency analysis. Results of a 15-transducer air sonar experiment demonstrate that the presented approach can resolve both incoherent and coherent signal sources, with resolution and precision equivalent to maximum likelihood and weighted subspace fitting methods