大規模多天線系統之 低複雜度天線選擇演算法與入射角度估測演算法

Abstract

本篇論文將針對分散式大規模多天線系統與集中式大規模多天線系統，分別提出一個低複雜度天線選擇演算法與低複雜度入射角估測演算法。 在分散式大規模多天線系統中，在論文[1]中有針對此環境的天線選擇演算法，並且衡量準則為最大傳送速率之總和，因此複雜度相當的高。另外，由於使用最大傳送速率之總和當做衡量準則，所以會導致有使用者無法被服務。本篇論文想針對此兩個缺點，提出一個低複雜度天線選擇演算法，並且兼顧服務品質，在複雜度分析可以證明本篇論文所提出方法有較低複雜度，另外在模擬分析，最大傳送速率之總和差距不大，本篇論文所提出演算法還可以確保每個使用者的可達信息率至少都大於零。 在集中式大規模多天線系統中，在基地台端天線布置在立體的二維陣列上，因此為了未來設計3D波束成型，所以需要知道詳細入射角的資訊。傳統上估計入射角方法使用綜合搜索方式，導致計算複雜度相當龐大，尤其在此假設環境當中，所以本篇論文想提出一個低複雜度兩段式入射角估測演算法，此演算法將利用多重訊號分類方法，此演算法可以先估計出仰角，再估計出水平角。在後面複雜度分析可以證明此演算法有較低複雜度的優點，由模擬結果，效能差距也不是很大。

In the thesis, we propose a low-complexity antenna selection algorithm for distributed large-scale MIMO (DLS) systems and low-complexity DoA estimation algorithm for full-dimension massive MIMO systems. In DLS MIMO systems, Liu and Lau [1] proposed an antenna selection scheme for DLS MIMO systems. The criterion of the algorithm is the sum-rate maximization. Therefore the required algorithmic complexity could be large. It also causes the situation that some users are unserved. We want to propose antenna selection algorithm to overcome these two shortcomings. The flop-count analyses are provided for the low-complexity advantage. It allows each user to be served by at least one antenna. In full-dimension (FD) massive MIMO systems, the 2-dimensional (2D) antenna array is placed at the base station. In order to implement the next generation technique three-dimensional beamforming, accurate estimation of the directions-of-arrival (DoA) of the signal sources is necessary. Existing DoA estimation methods using 2D arrays typically resorted to joint search over the 2D angular domain; however, the required algorithmic complexity could be prohibitively large, especially in the massive MIMO scenario. We propose a two-phase low-complexity DoA estimation scheme, in which two MUSIC algorithms are performed for estimating, respectively, the elevation and azimuth angles of the impinging sources. Detailed flop count analyses are provided for justifying the low-complexity advantage.