在不匹配環境下於雙邊基地互質天線多重輸入多重輸出雷達系統之強健性波束成型演算法

Abstract

本篇論文主要探討在不匹配非理想情況下於雙邊基地互質天線多重輸入多重輸出雷達系統之強健演算法之設計，探討之常見的非理想環境包括: 天線間的未知耦合(unknown mutual coupling, UMC)、天線元件增益相位誤差(gain-phase error, GPE)等等。當這些多重誤差存在於環境中時，會使的系統效能產生嚴重的下降，因此許多強健演算法被提出，但這些演算法多是適用於均勻線性陣列天線(Uniform Linear Array, ULA)，如果我們直接將這些演算法套用在互質陣列天線(Coprime Array, CPA)上，常常無法得到很好的強健效果。因此，如何在互質陣列天線與多重誤差存在的情況下，提出能保持良好效能的強健演算法是我們所探討的問題。 本篇論文中提出的演算法，當我們接收到資料向量時，首先會針對想要的訊號(desired signal)與其他的干擾(interference)做離開方向(direction of departure, DOD)與到達方向(direction of arrival, DOA)的角度估計，而傳統上二維MUSIC (Multiple Signal Classification)演算法，其計算複雜量過於龐大，因此分離資料向量為傳送端與接收端兩個部分，分別使用一維MUSIC估計DOD和DOA，以降低角度估計的計算複雜度。 接下來，已知估計的角度後，我們將提出方法估計UMC和GPE的係數，並為了得到進一步更精確的估計結果，有參考過去實驗室曾提出過的解最佳化問題，以及提出迭代估計的兩種方法。有了較為精確估計的結果後，我們可以重建去除掉想要的訊號後的干擾加雜訊自相關矩陣，進行波束成型以提升效能。

This thesis mainly explores the design of the robust algorithm of the bistatic multiple-input multiple-output radar system with coprime array in the mismatch scenarios. The mismatch scenarios include gain-phase error (GPE) and unknown mutual coupling (UMC). When the multiple errors exist in the environment, the system performance will seriously decrease. Therefore, many robust algorithms have been presented. However, most of the algorithms are applicable to uniform linear array (ULA). Accordingly, if we directly apply these algorithms to the coprime array, we can seldom get good and robust effects. As a result, how to maintain good efficiency of algorithms in coprime array with multiple errors is what we are going to explore. In our presented algorithm, first, we want to estimate the direction angle of departure (DOD) and arrival (DOA) based on the desired signal and other interferences after receiving data vector. However, if we use the two dimension MUSIC (Multiple Signal Classification) algorithm to estimate angles, it will lead to high computational complexity. In the purpose of solving this problem, we use the separate model to split the data vector into transmit part and receive part. Thus, we can use one dimension MUSIC algorithm to estimate DOD and DOA, respectively, to reduce the complexity. Next, we present the method to estimate GPE and UMC elements. Then, in order to get more precise estimations, we use two different ways to get it. One refers the method by solving optimization problem which our lab member has been presented, and another is by iteratively estimating parameters. With more precise estimations, we can reconstruct the interference-plus-noise covariance matrix, and perform the beamforming to robust the system performance.