| Summary: | 碩士 === 元智工學院 === 電機與資訊工程研究所 === 84 === In array signal processing, on-line DOA (direction of arrival)
estimation and tracking is very important research field. The
conventional algorithms need eigendecomposition to estimate the
signal subspace, and need parameter-search tofind the sources'
DOA. These cost a lot of computation, especially for large
array. About these problems, there are two parts of the
solutions; one is conquer the computation needed for
eigendecomposition, the other is overcome the computation of
parameter-search. For first part, the proposed methods estimate
the signal subspace using only linear operations, and can hence
be implemented in a very efficient manner. Another new approach
for tracking the singal subspace recursively has been proposed.
It obtain signal subspace directly by recesived signals, so it
need not to compute the correlation matrix firstly. About second
part, in this dissertation, the steering vector decomposition
algorithm using polynomial-rooting approach is proposed for the
sources' DOA estimation. This method can be used in one
dimension or two dimension DOA estimationproblems without any
constraints in array geometry. Besides, some adaptivesignal
processing methods have been employed to track sources' DOA by
signal subspace. Furthermore, an adaptive method which track DOA
from recesived signals, and it isn't in need of the signal
subspace. Computer simulation results are given for
illustration.
|
|---|
