本論文在多輸入多輸出正交分頻多工(MIMO-OFDM)系統下使用適應性濾波器估計通道特性。無線通訊系統中MIMO-OFDM架構是一個重要的技術,為了提升系統效能,精確估計通道狀態 (Channel State Information, CSI)是必要的。估計通道的方法繁多,例如最小平方 (Least Square, LS)、最小均方誤差(Minimum Mean Square Error, MMSE)與適應性濾波器(Adaptive Filter),其中MMSE雖然有最好的效能,但它必須事先知道通道狀態且運算複雜度高,因此常作為比較的基準。LS在高度干擾或多雜訊的環境下其效能並不理想,因此本論文使用適應性濾波器與領航符碼(Pilot symbol)估計通道特性。但因傳統適應性濾波器會有收斂速度過慢的問題,因此我們進一步提出結合LS與適應性濾波器的方法,包括LS結合最小均方(LS+LMS)與LS結合遞迴式最小均方(LS+RLS)兩種,以提升適應性濾波器的收斂速度。模擬結果顯示我們提出的方法之收斂速度遠優於適應性濾波器,且其均方誤差(Mean Square Error, MSE)與適應性濾波器相當。
In this study the application of adaptive filter for channel estimation in MIMO-OFDM system is investigated. During the years, MIMO-OFDM has been adopted in various wireless communication systems due to its many advantages. However, accurate channel estimation is essential for such systems to achieve good performance. Many channel estimation schemes have been presented, such as least square (LS), minimum mean square error (MMSE), and adaptive filtering. While MMSE has the best performance among them, it needs to know the channel state in advance. In addition, its computational complexity is the highest. Therefore, MMSE is commonly used as a benchmark for performance comparison. On the other hand, performance of LS degrades as channel interference or noise increase. Hence, adaptive filter and pilot symbol are employed in this study for channel estimation. To overcome the problem of slow convergence rate in conventional adaptive filter, two estimation approaches that combine LS and adaptive filters with least mean square (LMS) and recursive least square (RLS) are proposed. Experimental result indicates that the proposed schemes achieve mean square error (MSE) performance similar to the conventional adaptive filter with a faster convergence rate.