應用於行動通訊中以空間特徵結構為設計基礎之MIMO-OFDM波束形成技術

Abstract

波束形成技術是一種能夠有效增加系統容量和提升訊號對干擾雜訊比的策略，常見的通道模型如通道狀態資訊(CSI)和瑞雷衰減通道雖然能夠模擬整個通道行為，然而這類型的通道模型並不能完全解釋波束形成技術和空間通道之間的關聯性。在實際的戶外行動通訊中，通道一般會呈現所謂的半相關度的特性，上述的特性造成整個基地台和行動台的通訊環境並非呈現對稱關係，這造成傳送和接收端所使用的波束形成技術會有很大的差異性存在。 本篇論文主要分為兩個研究方向，首先將正交分頻多工的概念導入半相關度通道模型中，提出一種適用於波束形成技術的空間特徵結構通道模型。整個通道的空間特徵結構可分離成數個路徑群組的等效路徑增益、等效傳送角度、組合簽名向量和路徑延遲時間，並且利用探測信號和事先設定的測試權重向量可以準確地估測出這些通道參數值。接著藉由所提出的通道模型，傳送和接收端波束形成器架構將採用獨立分離的設計方式。在基地台傳送端採用統計型式的傳送波束形成技巧，如此可以確保接收端能夠得到穩定的接收功率且避免因回授和路徑延遲所造成回授通道資訊過時的現象。除此之外，所提出對路徑群組路徑增益向量做編碼設計的方式能夠達到低回授量的頻寬需求。來自於行動台必須處理即時性的通道資訊，所以在接收端波束形成器設計上，兩種新的適應性權重向量更新方式將被提出用來提升接收功率和解決同頻干擾的問題。可以證明在時變半相關度通道的情況下，傳送和接收端使用獨立分離的波束形成設計方式，可以提升系統的整體效能。

Beamforming technique is an efficient strategy for enhancing the system capacity and increasing signal to interference plus noise ratio (SINR). Typical channel models, such as channel state information (CSI), i.i.d. Rayleigh fading channel, and etc., well describe the characteristics of wireless communications. Unfortunately, these existing channel models cannot appropriately describe the connections between MIMO beamforming and spatial channels. The semi-correlated channel is a proper model for outdoor mobile communications due to the asymmetric features of geometric structure of signal propagation at both sides of link. This implies that the beamformers at the transmitter and receiver should adopt different design methods. In this thesis, two main topics will be discussed. In the first part, a new spatial eigenstructure channel model combining OFDM concepts with semi-correlated channels will be proposed for the optimal beamforming algorithms. Overall spatial eigenstructure consists of several resolvable clusters modeled by path gain, direction of departure (DOD) of path, composite signature vector (CSV), and path delay time. Besides, an efficient and realizable estimation method will be proposed to obtain these channel parameters of spatial eigenstructure mentioned above by sending training signal and using probing weight vectors. In the second part, we will propose new beamforming algorithms according to spatial eigenstructure channel model. The transmit beamformer at the BS will use statistical channel information to supply the receiver with stable transmitting power. It avoids using the out-of-date channel information caused largely by the feedback delay and processing delay. A novel codebook design applied in cluster path amplitude vector will also be implemented to achieve bandwidth-limited feedback. From the receiver's point of view, the MS must operate based on the real-time channel information. In the design of receiver-end beamformers, two new adaptive algorithms applied in weight vector update will be exploited to enhance the received power and suppress the co-channel interference (CCI). In semi-correlated and time-varying channels, it will be shown that system yields good performances by using different beamforming design methods at the transmitter and receiver.