利用子符號內插法實現OFDM通道估計

Abstract

在OFDM系統的通道估計上，通常藉由訓練訊號 (Training signal)的輔助來找出通道響應。但考慮到通道的時變性，使用訓練訊號求得的通道響應，並不能代表資料訊號(Data signal)的通道狀況。在快速衰減的通道下，我們提出子符號多項式內插方法配合最強路徑選取(most significant taps)演算法來內插出資料訊號的通道響應。在子符號多項式內插上，將均方誤差分成模型誤差跟雜訊誤差，討論不同系統參數包括都卜勒頻率，訓練率(Training rate)和多項式階數。我們會使用一個已知統計特性的通道模型檢查推導出來的誤差與模擬誤差是否配合。最後拿子訊號內插法比較其他現有的多項式內插方法，直接判斷演算法(Decision direct algorithm)與線性內插方法，子訊號內插法方法會有最好的表現。

In OFDM channel estimation, we usually utilize training signals. In case of a time varying channel environment, the channel responses, estimated during training period, can't represent the channel during data transmission. In such fast fading channel, we propose the sub-symbol polynomial interpolation algorithm that retains most significant taps algorithm to interpolate channel responses in data position. We derives its mean square error (MSE) that includes both model and noise errors. We verify the MSE performances of the derived results and the simulation results by using a time varying channel, whose statistics are known. The proposed sub-symbol polynomial interpolation that is the most effective in the fast fading channel, compared with the existing polynomial interpolation, the decision direct algorithm, and the linear interpolation.