標題: 編碼多天線頻域正交多工信號在時變通道下之偵測
On the Detection of Coded MIMO OFDM Signals in Time-Varying Channels
作者: 洪佳君
Chia-Chun Hung
蘇育德
Yu T. Su
電信工程研究所
關鍵字: 頻域正交多工系統;通道估測;多收發天線;時變通道;低密度同位校正碼;OFDM;channel estimation;Multiple Input Multiple Output;time-varying channel;Low Density Parity Check code
公開日期: 2004
摘要: 這篇論文主要是針對使用低密度同位校正(Low Density Parity Check, LDPC) 碼之頻域正交多工 (Orthogonal Frequency Division Multiplexing, OFDM) 系統,從理論上探討其接收端在時變通道下,相關的系統架構以及信號處理流程等重要的設計課題。 我們提出一種利用結合低密度同位校正碼的解碼器以及頻域正交多工系統之通道估測與解調的接收系統架構。透過所謂的渦輪原理(Turbo Principle),我們讓接收器反覆的在解碼器以及通道估測解調器之間交換軟輸出資訊。同時,我們也提出了改良式的模型基礎(model-based)通道估測法,這種新型估測法可以有效的改善先前類似的通道估測法之效能,並保留其追蹤時變通道頻率反應的能力。 上述反覆式接收器架構要從單收發天線(Single Input Single Output, SISO)系統延伸到多收發天線(Multiple Input Multiple Output, MIMO)系統時,會有額外的同天線(空間)干擾(Co-Antenna Interference, CAI)的問題要解決。我們討論了兩個方案並建議一種適合在此環境下使用的反覆式架構。這種架構包含了兩個反覆迴圈(Iterative Loop),內迴圈(Inner Loop)結合了同天線干擾消除與解碼,外迴圈(Outer Loop)之功能則類似單收發天線接收系統,從事反覆的通道估計與解碼工作。透過電腦模擬,可以正確的評估我們所提出架構與方法之效能。種種數值模擬的結果都顯示,我們的接收系統的確可以保證極低的位元錯誤率(Bit Error Rate, BER)以及訊框錯誤率(Frame Error Rate, FER)。
This thesis is mainly concerned about the developments of the receiver structure and the signal processing algorithms for LDPC-coded OFDM systems in a time-varying channel. We suggest a joint channel estimation and decoding receiver structure using an iterative decoding approach that exchanges the information between the low density parity check (LDPC) code decoder output and the orthogonal frequency division multiplexing (OFDM) channel estimator input. Efficient new model-based algorithms for estimating time-varying multi-carrier channel frequency responses are given. We consider both single-input single-output (SISO) and multiple-input multiple-output (MIMO) cases. Computer simulation is performed to estimate the performance of the proposed algorithms. Numerical results reveal that our algorithms do yield excellent bit error rate (BER) and frame error rate (FER) performance.
URI: http://140.113.39.130/cdrfb3/record/nctu/#GT009213532
http://hdl.handle.net/11536/69745
Appears in Collections:Thesis


Files in This Item:

  1. 353201.pdf