全象度分散式巨量天線系統:通道估計與錯誤率分析

Abstract

本論文主要在研究分散式巨量天線系統的通道估計與錯誤率(BER)效能分析。這樣的系統由裝有大型天線陣列的基地台(BS)來服務多個單天線用戶(UE)，由於天線數目很多，相關的通道估計通常需要大量的運算與複雜的信號處理。我們延伸前人利用了通道在轉換域(transform domain)的稀疏性來發展二維和三維的通道估計演算法。不同於前人的做法，我們使用二維壓縮得通道模式，並同時考慮了單一叢集(cluster)和多個cluster叢集的通道。除了LS演算法，我們也考慮了LMMSE演算法以及以Chebyshev為基底的演算法。我們分析了通道估計演算法的均方誤差(mean square error)及使用所估計的通道向量去解調信號的逼零(zero-forcing)接收機之錯誤率效能。利用我們所推導的MSE與BER的分析式，我們分別提出了兩種決定通道有效維度(dominant rank)的演算法，並比較分析其效益。 我們也透過分析，推導出接收信號之平均入射角(Angle of Arrival, AoA)及入射角擴散(Angle Spread, AS)等參數與通道能量頻譜參數的關係式。因為我們通道估計演算法需要空間相關矩陣的資訊，因此我們也提出空間相關矩陣的估計法。另外，由於通道估計的效能與所使用的基底有關，我們又分析並比較了DCT及DFT基底的優劣。

We investigate transform domain based estimators for both two-dimensional (2D) and three(full)-dimensional (3D) massive MIMO channels with single or multiple narrowband arriving clusters. The estimators take advantage of the channel's transform domain sparsity but require side information about the energy spectrum of the channel to determine the dominant channel rank and the associated subspace. Closed-form expressions which relate the dominant signal subspace support to the mean angle of arrival (AoA) and the corresponding angle spread (AS), which used to be estimated by iterative algorithms, are obtained. We analyze the mean squared error (MSE) of the estimators and the bit error rate (BER) performance of the corresponding (adaptive) zero-forcing and MMSE receiver in the presence of channel estimation error. We compare the MSE- and BER-minimizing channel estimators and study the associated rank distribution and performance. Optimal power allocation between pilot and data symbols is investigated and the resulting performance analyzed. Various numerical results are provided to examine behaviors of our channel estimators and related receivers and to validate the superiority of the proposed schemes. We show that the resulting receivers perform Rake-like combining in the spatial domain that automatically capture most of the energy of the desired signal while rejecting noise and interference from other directions.