This paper presents a generalized parametric estimator for the directions of arrival (DOA) of wide-band signals. This estimator is derived by extenting the geometrical explanation of the ML estimator of narrow-band signals to the focussed correlation matrix [1 The consistency of the estimator for estimating DOA has been proved. This estimator can be considered as a coherent signal processing method by which the computation complexity can be reduced approximately by the number of the frequency bins. We have also shown that under certain condition the proposed estimator is equivalent to the ML estimator derived by applying the likelihood principle on the Fourier coefficients of each frequency bin. Such an equivalence implies that the MLE has some inherent advantages from the perspective of improving performance and that the focussing techniques are not necessary for the ML estimator. The simulation results that the new parametric method possesses higher resolution than the eigendecomposition methods especially at a small sampling size. However the price paid for the better performance is the increase on computation complexity comparing to the eigendecomposition methods.

Date Published: 1 November 1990
PDF: 11 pages
Proc. SPIE 1348, Advanced Signal Processing Algorithms, Architectures, and Implementations, (1 November 1990); doi: 10.1117/12.23488

Published in SPIE Proceedings Vol. 1348:
Advanced Signal Processing Algorithms, Architectures, and Implementations
Franklin T. Luk, Editor(s)