此篇論文所介紹的是精準估測訊號來源方位角的方法，現今較常使用估測方位角度的演算法大概分為兩種，一種為ESPRIT演算法[9]及MUSIC演算法[17]，而我們將精準定義為誤差角度於1.5度間，為使估測角度與原訊號來源角度誤差更小，將原估測方位角，再加上運算複雜度相對較高的以Matching Pursuit(匹配追蹤)為基礎的兩階段演算法[6]，加以精算得出較精確的結果。 論文中也比較了ESPRIT、MUSIC及Matching Pursuit三種演算法，設定於各項不同環境參數中：天線陣列數、快照數據資料等，估測訊號方位的精準度及效益，並依此三種演算法特性進行精確方位估測的搭配及提升精準度。In this thesis, a high resolution angle of arrival etimation algorithm is developed. There are two popular methods used to estimate the arrival angle of a source, one is the ESPRIT [9] algorithm and the other is the MUSIC [17] algorithm. If the estimation errors are smaller than 1.5 degrees, it is defined as the precision estimation. In order to improve the estimation error, the initial estimates of the ESPRIT or the MUSIC algorithm can be further refined by the Matching Pursuit technique which has a higher computational complexity. The two-stage algorithm is thus designed and the complexity is reduced with comparable performance compared to the pure Matching Pursuit technique. The thesis compares the ESPRIT, the MUSIC, and the Matching Pursuit algorithms in different scenarios such as the number of antenna elements and the number of the snapshots …,etc. The performance of the estimates by the three methods is displayed through simulation results. By the combination of these algorithms, how to achieve the precision estimation is discussed by the trade off between performance and complexity.
