Font Size: 

A powerful tool for improving the threshold performance of direction finding is considered. Our technique referred to as PseudoRandom Joint Estimation Strategy (PR-JES) allows to reduce the number of outliers in Direction Of Arrival (DOA) estimates using spatial spectrum resampling. The essence of PR-JES is to use the eigenstructure-based estimator bank generated in pseudorandom manner for given sample covariance or data matrices. Appropriately combining the results of "parallel" underlying estimators, PR-JES removes the outliers and improves direction finding performance in the threshold domain. The applications of estimator bank approach to the second- and higher-order eigenstructure techniques are developed and efficient beamspace root implementations for a Uniform Liner Array (ULA) are considered. Friedlander array interpolation approach is exploited to extend these implementaions to arbitrary arrays. Simulation results show that our technique dramatically outperforms the MUSIC estimator and achieves the performance similar to and sometimes better than that of stochastic ML technique.