陣列訊號處理中訊號指向向量之估測與應用

Estimation and Application of Steering Vector for Array Signal Processing

曾俊卿(Chun-Ching Tzeng)

指導教授 ： 陳逸民

元智大學/工程學院/電機與資訊工程研究所/碩士(1995年)

摘要

一般常見陣列訊號處理的演算法均假設天線的位置．或天線的相位響
應．或增益響應等等已知，MUSIC的方法運用天線位置的相對結構(
Structure)以搜尋訊號到達角度(DOA)．在本篇論文中運用兩種不同的方
法在天線的位置．天線的相位響應．增益響應均未佑的情況下來估算指向
向量(Steering Vector)．而因為在上述情況下運算，所以我們可很方便
地運用到天線位置自我校正．天線相位響應自我校正．天線增益自我校正
．訊號源方位估測等等問題上．而第一種方法運用遞迴(Iterative)投影
的方式分別遞迴地地投影到訊號子空間(Signal Subspace).正交條件空
間(orthogonal ConditionSpace).及天線增益空間，直到收斂到穩定值．
如果初始值夠好，這個演算法將可很快收斂到穩定值．而第二種方法運用
分解訊號子空間產生基底(Basis)，而用這些基底作線性組合產生指向向
量，運用最小誤差解找出最適當的指向向量，這些指向向量所生成的空間
將會與訊號子空間相似(Similar)．而第一種方法受到訊號源需獨立(
Independent)的限制，可由第二種方法來彌補這點不足．

關鍵字

指向向量 ； 陣列訊號處理

並列摘要

Generally, array signal processing always assume sensor
location, or sensorresponse, or sensor gain response,etc. have
be well known. The MUSIC estimatesthe direction of arrival (DOA)
of emitting sources by matching the assumed model to the signal
subspace obtained from the eigenstructure of the data covariance
matrix. In this thesis we assume the sensor location, sensor
phase response and sensor gain response are unknown, and we
apply two kind of algorithm to estimate steering vector. Because
those response and location are unknown, so we can apply to
sensor shape self-calibration, sensor phase response self-
calibration, and sensor gain self-calibration, etc. First
algorithm is a iterativeapproach, iterativelly project to signal
subspace, orthogonal condition subspace and sensor gain space
until the steering vector convergence to a stable value. If the
initial point is sufficiently good, this algorithm will converge
tothe global maximal. Second algorithm decompose signal sub-
space to getting subspace basis and linear combination these
basis to getting steering vector and finding the least mean
square error solution of steering vector.

並列關鍵字

Steering Vector ； Array Signal Processing