請用此 Handle URI 來引用此文件:
http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/56446
標題: | 支援快速時變通道訊符偵測的正交分頻多工接收模組 An OFDM Receiver Module for Symbol Detection in Fast Time-Varying Channels |
作者: | Ting-li Liu 劉頂立 |
指導教授: | 郭斯彥(Sy-Yen Kuo) |
關鍵字: | 高速運載車輛,行動正交分頻多工,快速時變通道,接受器窗函數,子載波間干擾, high-speed vehicle,mobile OFDM,fast time-varying channel,receiver window,intercarrier interference (ICI), |
出版年 : | 2014 |
學位: | 博士 |
摘要: | 鑑於高速運載車輛搭乘大眾對寬頻通訊的需求,我們研發支援正交分頻多工(OFDM)行動通訊設備在快速時變通道中執行訊符偵測(symbol detection)的接收器模組。因此,本研究必須達到三個目標:高都卜勒效應通道中具低錯誤率、低計算量以符合實務,並與現行系統具高相容性。
我們分析高速移動正交分頻多工系統頻域通道矩陣(channel matrix)的內涵。為了降低快速時變通道帶來的載波間干擾(ICI),我們利用帶狀(banded)通道矩陣建立系統模型,及估算帶狀近似產生的誤差,做為通道估測(channel estimation)與等化(equalization)程序中最小均方差(MMSE)估測元件的輸入。 為了有效降低載波間干擾,我們在傳送器的反離散傅立葉轉換(IDFT)單元後插入時域餘弦窗函數。依據我們的分析,帶狀近似誤差降低為載波間干擾的–19到–49分貝。我們進一步分析在接受器的離散傅立葉轉換(DFT)單元前插入時域窗函數產生的效應。與最佳值近似,我們推導得到2個固定的窗函數,分別以–23分貝及–52分貝固定增益減少載波間干擾。 基於在接受器插入窗函數所產生極高的載波間干擾降低效益,系統的通道矩陣精確地近似一個嚴謹帶狀矩陣。利用帶狀近似誤差的估算值,我們提高訊符偵測程序中最小均方差估測器的效能。在系統傳送梳式(comb-type)非聚集型(non- clustered)指標訊符(pilots)協助通道估測的環境中,我們提出渦輪(turbo)最小均方差訊符偵測,利用資料訊符的先備知識(prior knowledge)重複迭代基底膨脹模型(BEM)通道估測與最小均方差等化。模擬結果顯示,我們研發的接受器模組在模擬快速時變通道中用合理計算量執行訊符偵測程序,並達到相當低錯誤率的目標。 Seeing the needs of the broadband communication for the people on high-speed vehicles, we develop an auxiliary module which assists orthogonal frequency-division multiplexing (OFDM) mobile devices to detect symbols through fast time-varying channels. Hence, this research targets low bit error rates (BER) in highly Doppler- affected channels, applicably low computational consumption and high compatibility with currently deployed systems. We analyze insights of the frequency-domain channel matrix for the highly mobile OFDM systems. To mitigate the intercarrier interference (ICI) introduced with fast time-varying channels, we model the systems by a banded channel matrix, and evaluate the banded approximation errors for the inputs to the minimum mean square error (MMSE) estimators in both channel estimation and equalization. To mitigate ICI efficiently, we insert the time-domain cosine windows immediately after the inverse discrete Fourier transform (IDFT) unit in the transmitters. Upon our analysis, the banded approximation errors are decreased to –19∼–49 dB of the ICI power. Moreover, we consider the receiver with a time-domain window inserted before the discrete Fourier transform (DFT) unit. In controlled deviations from the optimum, we derive two invariant windows which remove the ICI power by the constant gains of –23 dB and –52 dB, respectively. Thanks to extremely high ICI-mitigation gains benefited from insertion of the receiver windows, the channel matrix is precisely approximated to a strictly banded matrix. With aids of the evaluations of the banded approximation errors, we enhance the performance of the MMSE estimators in symbol detection. In the environment where the comb-type non-clustered pilots are transmitted to assist channel estimation, we propose the turbo MMSE detection that iterates the basis expansion model (BEM) channel estimation and the MMSE equalization each other with the prior knowledge of the data symbols. Simulations show that the receiver module performs the symbol detection through fast time-varying channels with the results in considerably low BER at the expense of reasonable computations. |
URI: | http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/56446 |
全文授權: | 有償授權 |
顯示於系所單位: | 電機工程學系 |
文件中的檔案:
檔案 | 大小 | 格式 | |
---|---|---|---|
ntu-103-1.pdf 目前未授權公開取用 | 2.45 MB | Adobe PDF |
系統中的文件,除了特別指名其著作權條款之外,均受到著作權保護,並且保留所有的權利。