本論文主要研究數位電視通道估測技術,我們利用市面上販售的數位電視機上盒,擷取目前現有的數位電視之中頻訊號,並透過撰寫程式進行解碼。數位電視在通道估測階段前,必須先經過同步的階段,由於是接收目前台灣五家電視台所發射的數位電視訊號,再加上我們所使用的同步方法是使用較簡單好實現的,所以同步結果不能很準確,而這導致在通道估測時會遇到相位旋轉等問題,本論文也針對這問題進行探討,最後得到結論後並提出一套有效方法來處理。至於通道估測部分,實作了一些常被使用且效果不錯的二維線性內插法方法,探討這些方法運用在真實訊號情況下效果為何,並且對於解不好的檔案,去觀察其通道現象,並再使用自迴歸、決策回饋、轉換域處理方法來改善通道估測,最後運用自迴歸和回饋決策的結合來設計一個改善方法,在較低運算複雜度上,使通道估測有顯著改善。
This thesis mainly studies Channel Estimation technique for DVB-T. We use existing DVB-T set-top box to get its intermediate frequency (IF) signal, which is digitized and stored in a PC. After getting the signal we write a program to decode. Before the channel estimation stage, have to finish synchronous stage. It is because we get real signal which broadcasts from five television stations in Taiwan, and synchronous method we used is easy algorithm, so synchronous can't be very accurate. According to this reason, it makes problems of phase rotate. This thesis also to probe the causes of the phase rotate, and finally we propose a solution to handle it. At the channel estimation stage, we implement some 2D interpolation method which has good effect. For some file after decoding performance is not good, we observe their channel environment. For making estimation accurate, we use some methods which are Auto-Regressive, Decision Feedback, Transform-domain processing, etc. Finally we combine Decision Feedback with Auto-Regressive to design an improvement method. This method can improve performance effective under lower operate.