Ab. Gershman, PSEUDO-RANDOMLY GENERATED ESTIMATOR BANKS - A NEW TOOL FOR IMPROVING THE THRESHOLD PERFORMANCE OF DIRECTION FINDING, IEEE transactions on signal processing, 46(5), 1998, pp. 1351-1364
A new powerful tool for improving the threshold performance of directi
on finding is considered. The main idea of our approach is to reduce t
he number of outliers in DOA estimates using recently proposed joint e
stimation strategy (JES), For this purpose, multiple different DOA est
imators are calculated in a parallel manner for the same batch of data
(i.e., for a single data record). Employing these estimators simultan
eously, JES improves the threshold performance because it removes outl
iers and exploits only ''successful'' estimators that are sorted out u
sing hypothesis testing procedure. We consider an efficient modificati
on of JES with application to the pseudo-randomly generated eigenstruc
ture estimator banks based on second-and higher order statistics. Weig
hted MUSIC estimators based on the covariance and contracted quadricov
ariance matrices are chosen as appropriate underlying techniques for t
he second-and fourth-order estimator banks, respectively. Computer sim
ulations with uncorrelated sources verify dramatic improvements of thr
eshold performance as compared with the conventional second-and fourth
-order MUSIC algorithms. Simulations also show that in the second-orde
r case, the threshold performance of our technique is close to that of
the WSF method and stochas-tic/deterministic ML methods, which are kn
own today as the most powerful tin the sense of estimation performance
) and, at the same time, as the most computationally expensive DOA est
imation techniques, The computational cost of our algorithm is much lo
wer than that of the WSF and ML techniques because no multidimensional
optimization is required.