正交分頻多工和正交分頻多重進接系統中雙選擇性通道的估測及等化

Abstract

正交分頻多工（Orthogonal frequency-division multiplexing, OFDM）和正交分頻多重進接（Orthogonal frequency division multiple access, OFDMA）系統被廣泛地應用在無線通訊系統中，因為它們可以有效地對抗頻率選擇性通道。然而通道有時也具有時間選擇性，並可能導致正交分頻多工和正交分頻多重進接系統的效能嚴重下降。因此，有許多研究著重於正交分頻多工系統在雙選擇性通道下的問題。最近幾年，P. H. Chen提出一等效離散時間雙選擇性通道模型。基於分階延遲線（Tapped-delay-line, TDL），此通道模型以單一的都卜勒頻率偏移（Doppler frequency offset, DFO）參數來描述每條雙選擇性通道中的時間選擇性。 本篇論文著重在正交分頻多工和正交分頻多重進接系統中雙選擇性通道的估測及等化。我們利用上述分階延遲線模型來模擬雙選擇性通道。在正交分頻多工系統中，我們提出一基於頻率分析（Frequency analysis-based）的都卜勒頻率偏移估測方式，其估測值係以解析解（Closed-form）輸出。我們也提出了基於最小均方誤差（Minimum mean square error, MMSE）的波束成形器（Beamformer）和通道等化器來改善偵測精準度。此外，我們還提出了一通道篩選方式，透過刪除低訊雜比的通道來減少還原訊號時需要的運算量。模擬結果顯示，使用最小均方誤差的接收器比起過去使用擬逆（Pseudoinverse）的等化器有更好的均方誤差和錯誤率表現。 另外，我們將論文延伸至上行（Uplink）正交分頻多重進接系統。對於都卜勒頻率偏移的估測，我們提出一基於導頻（Pilot-based）的都卜勒頻率偏移估測器並且將P. H. Chen提出的虛擬載波（Virtual carrier-based）都卜勒頻率偏移估測器擴展到上行正交分頻多重進接系統中使用。由於這是一多用戶系統，我們在設計接收器的時候會考量到用戶判別。對於通道估測，我們也提出了基於導頻的通道估測器。除此之外，基於最小均方誤差通道等化器和通道選擇方式也可以用來改善正交分頻多重進接的接收器效能。實驗結果顯示，我們提出的方法有良好的效果。

Orthogonal frequency-division multiplexing (OFDM) and its multiple-access version, orthogonal frequency-division multiple access (OFDMA), have been widely used in wireless communication systems since they have the advantage of being robust to frequency-selective channels. However, the channels can sometimes become time-selective, and this will cause significant performance degradation to OFDM or OFDMA systems. Therefore, the research on the frequency and time-selective channels (also called doubly selective channels) for OFDM-based systems has attracted much attention. Recently, an equivalent discrete-time (DT) channel model for doubly selective channels has been derived by P. H. Chen. Based on the tapped-delay-line (TDL), the model efficiently depicts the time-variation of a doubly selective channel by a Doppler shift offset (DFO) parameter. In the thesis, we focus on the estimation and equalization of doubly selective channels in OFDM and OFDMA systems. We adopt the above TDL channel model to model the doubly selective channels. For OFDM systems, we derive a frequency analysis-based DFO estimator, which gives the DFO estimate in a closed-form expression. We also propose a minimum mean square error (MMSE) beamformer and MMSE channel equalizer to improve detection accuracy. Moreover, we propose a path-selection method that reduces the receiver complexity by discarding the low-SNR paths. Simulation results show that our proposed MMSE methods outperform the existing pseudoinverse method in mean square error and bit error rate performance. In addition, we also extend our study to the uplink OFDMA systems. For DFO estimation, we propose a pilot-based estimator and extend Chen's virtual carrier-based (VC-based) method to the case of uplink OFDMA systems. Since the system is for multiuser, our receiver is designed with user identification in consideration. For channel estimation, we derive a pilot-based channel estimator. Additionally, we can also apply the proposed MMSE channel equalizer and path-selection method to OFDMA systems to further improve the performance. Simulation results show that the proposed methods give good performance.