多天線輸入輸出正交分頻多工系統下對抗快速衰減通道之載波頻率偏移與通道協同估測

Abstract

本論文研製了一個可對抗時變之頻率選擇性快速衰減通道的載波頻率偏移及通道追蹤估測方法。3GPP長期演進技術下行鏈路無線通訊標準被用作為設計時的參考系統，2對2的多重輸入多重輸出正交分頻多工系統被納入考慮。此系統工作在2GHz頻段，使用10MHz的頻寬。 模擬平台利用了兩種通道模型，其中之一為ITU-VehA，而另一為ITU-PedB。另時變模型是引入了瑞雷衰減模型來達成。 本論文所提出的載波頻率的估測法於頻域進行，藉由尋找成本函數的最小值作為估測依據。為了達到實作需求，分為三階段達成，此三階段係根據載波頻率頻移量大小，對系統所造成之影響不同之特性，而採用不同的方式處理。 此載波頻域偏移估測演算法於2對2多輸入輸出系統下，可在頻率選擇性通道中進行，並在訊雜比大於1的條件下容忍載波頻率偏移達百萬分之二十五。 本論文提出的通道追蹤估測包含三個步驟。先針對通道變化程度進行初步估測，利用頻空區塊碼經解碼後，經由本論文所提出的通道追蹤估測演算法做處理，以估計出包含雜訊的時變通道，再由一頻域上的濾波器濾去大部分的雜訊，以降低演算法對訊噪比的需求。 此通道追蹤估測演算法在2對2的多輸入多輸出系統下，於ITU-VehA通道模型中可在時速120km/hr、訊噪比24dB的條件下將位元錯誤率降至千分之一，在ITU-PedB通道模型中可在時速120km/hr、訊噪比29dB的條件下將位元錯誤率降至千分之一，且記憶體需求較傳統作法節省百分之九十。 此研究並將此兩估測演算法作了協同式設計，精確小數載波頻率偏移及通道估測初始化作了一個結合，且通道追蹤結果亦將殘餘載波頻率偏移效應估測回饋並消除，達到更精準之震盪器頻域同步。

This thesis presents a cooperative carrier frequency offset (CFO) and decision-feedback tracking channel estimation scheme for MIMO-OFDM systems over time-variant frequency-selective fast-fading channels. LTE Downlink (DL) standard is targeted. It operates at 2GHz band with 10MHz bandwidth. 2 2 transceiver allocation with corresponded Space-Frequency Block Coding (SFBC) is the main consideration. Two ITU channel models are applied to the simulation platform. One is the ITU-VehA model, and the other is the ITU-PedB model. The time-variant effect is achieved by the Rayleigh fading model. The CFO estimation is processed in frequency domain, and evaluated by minimizing a cost function. For achieving the implementation, the estimation is decomposed into three steps. These three stages based on the effect of different scales of carrier frequency offset, using different ways to estimate. The proposed CFO estimation can tolerate the CFO up to 25 ppm under the frequency selective channel at 1dB SNR for 2 2 MIMO-OFDM systems. The decision-feedback tracking channel estimation contains three parts. First, the degree of channel varying will be estimated, and then the proposed channel estimation scheme is applied to the SFBC-decoded symbols in order to estimate the noisy time-variant channels. The last, a frequency domain filter is applied to each estimated noisy time-variant channel frequency response to decrease the Signal-to-Noise Ratio (SNR) requirements that this work needed. The proposed algorithm achieves bit error rate under 120km/hr ITU-VehA model at 24dB SNR and under ITU-PedB model at 29dB for 2 2 MIMO-OFDM systems. And the memory buffer of the proposed algorithm is reduced to 10% , comparing with the conventional method. Also this work joints these two estimation method. The estimation of fine fractional CFO is achieved by the degree of channel varying, which is in the channel estimation stage. And the residual CFO estimation and cancellation will be attained by the feedback of channel estimation result.