MIMO-OFDM系統之偵測與低密度同位檢查碼解碼

Abstract

本篇論文主要為研究多輸入多輸出（MIMO）正交分頻多工（OFDM）系統上之訊號偵測與低密度同位檢查碼之解碼。在低密度同位檢查碼的解碼上，我們使用了三種解碼演算法，包括Normalized belief propagation（BP）based演算法、Normalized a-posteriori probability（APP）based演算法、和Layer normalized BP based演算法。這三種不同的解碼演算法，皆比低密度同位檢查碼標準的解碼演算法─ Sum-product演算法有更高的可實現性，同時在運算複雜度與效能的取捨上，都有不錯的表現。而在MIMO訊號的偵測上，我們使用了最小均方誤差（MMSE）與最大事後機率（MAP）偵測。其中，我們將MMSE偵測與軟性反對映結合以獲得軟性輸出（soft outputs），而MAP偵測則可直接由所接收到的訊號來求得軟性輸出。另外，為了降低MAP偵測本身龐大的計算量，我們在MAP偵測前預先使用了list sphere decoding演算法，以降低在MIMO訊號偵測上的運算複雜度。最後，我們在IEEE 802.11n與IEEE 802.16e系統上模擬，以驗證低密度同位檢查碼對系統效能上的改善，特別是使用MAP偵測MIMO訊號時，使用低密度同位檢查碼對系統效能會有相當大的改進。

In this thesis, we study signal detection and decoding of low-density parity-check (LDPC) codes for multi-input-multi-output (MIMO) orthogonal-frequency- division-multiplexing (OFDM) systems. Three types of LDPC codes decoders are investigated, the Normalized belief-propagation (BP) based algorithm, the Normalized a-posteriori probability (APP) based algorithm, and the Layered normalized BP based algorithms. These decoding algorithms are much simpler to implement than the standard LDPC codes decoding algorithm, namely the sum-product algorithm, and can achieve good tradeoff between decoding complexity and performance. For MIMO signal detection, we consider the minimum-mean-squared error (MMSE) and a maximum a-posteriori probability (MAP) detector. The MMSE detector is combined with a soft-bit demapper to obtain soft outputs, while the MAP detector is designed to have soft outputs directly. To redue the high computational inherent in the MAP detector, we apply an efficiency algorithm called the list sphere decoding. Simulations with IEEE 802.11n and IEEE 802.16e systems show that the LDPC codes decoder can effectively improve the system performance, particularly when it combined with the MAP detector.