| This dissertation examines the problem of multichannel angle-of-arrival estimation through a decomposition of the received data into uncorrelated components. Several new angle-of-arrival estimators based on the recently introduced MUSIC algorithm are presented and compared to the MUSIC and maximum entropy methods.;One major problem in angle-of-arrival estimation has been determining the number of emitters. This problem is solved by applying a likelihood ratio test developed by Anderson {1963} and Lawley {1956}.;Finally, the application to single-channel frequency estimation of a fast, noncomplex, version of MUSIC is discussed. This algorithm compares to maximum entropy in the number of multiplications required but produces unbiased spectra having greater resolution.;An analytical comparison of the MUSIC estimator to the maximum entropy, maximum likelihood, and beamforming techniques is described. This comparison shows the latter three techniques to have spectral equations similar to that of MUSIC but with extra noise induced terms. |
