|
High-resolution 2-D direction-of-arrival(DOA) estimation is
very popular topics in the last years, many researchers have
been studying these problems. Among these high-resolution
algorithms, the eigenstructure-based methods such as MUSIC,
ESPRIT etc., have been widely used. However, MUSIC needs a
search procedure inherently, so that it is computationally more
burden and has more serious restriction in array manifold. So
then, we make use of ESPRIT, the advantages of which does not
need a search procedure and no more serious restriction in
array manifold. Our purpose is to estimate 2-D DOA based on
ESPRIT-like algorithms, which are MI-ESPRIT and MST-ESPRIT.
Beside, we also make analyse of 2-D DOAs estimation for
broadband coherent sources. In 1985, the CSS algorithm was
proposed by Wang and Kaveh, then in 1988, an innoviated
algorithm- RSS was proposed by Hung and Kaveh. Both those
utilized a transformation matrix (or focusing matrix), which it
transforms various signal component in different frequency bins
into the signal component at a common frequency bin. Then it
utilized the frequency smoothing operation to remove the
coherence between sources, also with data reduced step. We
propose another method so called beamspace transformation, and
then compare the performances among beamspace transformation,
RSS and HSST by computer simulations. Since these
transformation matrices requires preliminary estimate of the
DOAs of sources, and if they are inappropriatedly selected
especially in the multigroup sceniarios, the transformation
matrix will degrade the array performance. So many scholars
wanted to find other focusing forms without preliminary known
DOAs. Alougth those literatures are rich, but it is sad to us
because most of those are limited to a linear array. Now, we
try to use eigenvalue decomposition or singular value
decomposition of the array data, to take the signal-beamspace
to finish the focusing work.
|