Robust direction-of-arrival estimation in contaminated noise
Abstract
The problem of estimating directions-of-arrival (DOA) of radiating sources from measurements provided by a passive array of sensors is frequently encountered in radar, sonar, radio astronomy and seismology. In this study various robust methods for the DOA estimation problem are developed, where the term robustness refers to insensitivity against small deviation in the underlying Gaussian noise assumption. The first method utilizes an eigenvector method and robust reconstruction of the correlation matrix by time series modeling of the array data. Secondly, a decentralized processing scheme is considered for geographically distributed array sites. The method provides reliable estimates even when a few of the subarray sites are malfunctioning. The above two techniques are useful for narrow band and incoherent sources. The third robust method, which utilizes Radon Transform, is capable of handling both the narrow band and wide band sources as well as the incoherent or coherent sources. The technique is also useful in situations of very low SNR and colored noise with unknown correlation structure. The fourth method is an efficient narrow band robust maximum likelihood DOA estimation algorithm which is capable of handling coherent signals as well as the single snapshot cases. Furthermore, relationships between eigenvector methods and a ML DOA estimation, where the source signals are treated as sample functions of Gaussian random processes, are investigated.
Degree
Ph.D.
Advisors
Kashyap, Purdue University.
Subject Area
Electrical engineering
Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server.