為了應付行動正交分頻多工通訊系統的通道估計問題,系統中常常需要插入許多領航訊號 (pilots) 因而降低頻寬使用效率。為節省所使用的領航訊號,本論文研究在時變通道環境下,利用決策回授機制(Decision-Feedback),提出改良的最小平方法 (LS) 時域通道響應估測器。然而,決策回授系統的缺點為決策錯誤累積(Decision-error propagation),我們提出錯誤檢測器來偵測決策錯誤,並且提出補償方法來避免決策錯誤累積。在實驗模擬的結果中,我們比較所提出不同的檢測器與通道估計補償方法的性能,在時變通道下得到可以接受的位元錯誤率,並且在追蹤的過程中不需要任何領航訊號,大幅減少系統傳輸效率的損失。 To deal with the channel estimation problem, the mobile orthogonal frequency division multiplexing (OFDM) system usually requires pilots to be inserted in transmitted signals in which the bandwidth efficiency is reduced. For the purpose of saving pilots, this paper studies a new least squares (LS) time-domain channel estimator using a decision-feedback loop. However, the drawback of the decision-feedback system is decision-error propagation. We propose error detectors for reducing the error probability and some compensation methods for enhancing channel estimation. In simulation results, the performances of the proposed decision-error detectors and compensation methods are compared. We show that the acceptable bit error rate can be achieved without pilots used for channel tracking and thus, the loss of bandwidth efficiency due to pilots is significantly reduced.