Date of Award
Doctor of Philosophy
Electrical and Computer Engineering
The principal topic of this dissertation is the application of array signal processing to angle-of-arrival (AOA) estimation of multiple plane waves. Assuming that a passive linear array of uniformly spaced sensors is used to measure the radiation field, a digital signal processing system is proposed which determines the number of narrowband sources and their respective bearings;The proposed direction-finding system integrates several algorithms in a cohesive arrangement to exploit their mutually similar structures. The Modified forward-backward linear prediction (MFBLP) spectral analysis method of Tufts and Kumaresan is used to obtain high spatial resolution, and the eigenanalysis which is central to its operation provides an excellent point of entry for a procedure to estimate the number of plane waves detected by the array. This procedure utilizes the AIC or MDL information theoretic criteria in an ensemble manner to generate a reliable estimate of the rank of the signal subspace of the deterministic correlation matrix of the array snapshot;Two algorithms are presented for determining the spatial frequencies of the incoming plane waves, based on the MFBLP method: one uses an iterative version of Newton's method to locate the spectral peaks, and the other uses an iterative method to locate the equivalent complex poles;Statistical processing of the bearing estimates from a number of array snapshots is then used to maintain accuracy and precision in noisy array environment; two different estimators are proposed for this ensemble averaging, and their performance is characterized when applied to a single-source scenario. This utilizes both analytical and Monte Carlo computer simulation. The distributions are characterized in both wavenumber and bearing domains. Expressions for the bearing CRLB, expected standard deviation, and bearing estimate confidence interval are developed, believed to be the first known formulations of overall DF system performance as a function of array spacing, number of elements, true source bearing, and sensor signal-to-noise ratio. The statistical precision is shown to be related to the effective aperture of the antenna array, revealing the degradation in performance as the angle-of-arrival approaches endfire.
Digital Repository @ Iowa State University, http://lib.dr.iastate.edu/
John Finney Aurand
Aurand, John Finney, "An antenna array processing system for multiple source bearing estimation " (1987). Retrospective Theses and Dissertations. 8508.