適用於多輸入多輸出系統之低複雜度與改善性能之方法

Abstract

多輸入多輸出天線系統（Multiple Input and Multiple Output，簡稱MIMO） 因為擁有更佳的傳輸效率及訊號的品質，所以被廣泛的應用在無線通訊的系統中。通道解碼使得整個系統在低訊雜比的環境中可以有較佳的表現。因此在目前所使用的許多規格中如IEEE802.11n、IEEE802.16e和3GPP LTE，會將多輸入多輸出天線系統在接收端訊號偵測與通道解碼做連接。其中因為規格的需求，通道編碼是利用低密度奇偶校驗碼（Low Density Parity Check，簡稱LDPC Codes）或渦輪解碼（Turbo Decoding），為了得到較好的表現，在MIMO的訊號偵測中需要產生軟輸出。然而，在產生軟輸出的過程中會使得整體解碼複雜度的提升，因此，本篇論文提出可以降低複雜度的演算法，並提出可以更加改善傳統MIMO訊號偵測錯誤率的兩種技巧。 首先，在本篇論文中提出雙方向演算法（Bi-direction），利用兩次的QR分解去做成本的計算，與QOC的方法相比，運算量最多可以減少30 %。接著透過非線性量化的方法（Nonlinear Quantization），把MIMO訊號偵測的軟輸出做分佈上的改變，使得錯誤率最大可以達到2dB的改善。最後，運用二階段（Two-stage）的技巧，在第一階段維持原來LDPC的解碼，在第二階段將一些不可靠的位元做數值的翻轉，在MIMO與LDPC連接的系統中，錯誤率最大可以達到0.3dB的改善。

This thesis proposes three approaches to apply for the channel-coded MIMO system. MIMO communication system has been widely applied in many wireless communications for better transmission efficiency and signal quality. Channel coding or forward error correction allows the system work better with the additional coding gain in lower SNR environment. Channel-coded MIMO system which refers to the MIMO detection with channel coding scheme is defined in many communication standard such as IEEE802.11n, IEEE802.16e and 3GPP LTE. Soft-output of MIMO system is required to a concatenate channel y-check (LDPC). However, complexity increases intensively as the hard-outputs transfer to soft-outputs of MIMO detection. A bi-direction method to reduce the complexity of soft-output MIMO decoding is proposed in this thesis. Sphere decoding of MIMO system encounters the empty-set issue as generating LLRs. Bi-direction method solves the empty-set issue and reduce the computations at most 30% compared to the original method QOC. In addition, nonlinear quantization (NLQ) is applied to the list sphere decoding and obtains a great performance improvement at most 2dB with little computation complexity. Furthermore, a two-stage algorithm applied on the channel-coded MIMO system is presented to improve the error performance at most 0.3dB.