為提高數位資料儲存系統之記錄密度和資料傳輸速率，現在之數位資料儲存系統常用比較先進之訊號處理技術，利用等化器將讀回信號等化成部份反應信號以降低雜訊接著再利用最佳估算偵測法則即維特比法則(Viterbi Algorithm) 將信號還原為數位資料，此種訊號處理技術稱為部份反應/最佳估算(Partial Response and Maximum Likelihood Detection, PRML)通道技術。本計畫將不再探討維特比偵測法則而是把重點放在研究及探討反覆偵測法則在資料儲存系統部分反應通道後之偵測性能，我們將分析及探討兩種反覆偵測法則(BCJR 演算法則及視窗-事後機率演算法則)之性能。資料儲存系統部分反應通道之柵狀圖可經由轉換之後而以分解圖(factor graph)來描述此通道，BCJR 演算法則及視窗-事後機率演算法則乃利用部分反應通道的分解圖以訊息流通演算法則為基礎做通道偵測。在接收訊息序列很長的時候，使用BCJR 演算來處理通道偵測，會有很大的時間延遲。因此為了要解決時間延遲的問題，同樣使用訊息流通演算法，但我們加入了視窗(window)的概念，稱之為視窗-事後機率演算法(window-APP algorithm)。利用視窗-事後機率演算法，在進行通道偵測時可以做到平行處理，這大大的降低了因為使用BCJR 演算法所造成的時間延遲。整個計劃的研究內容主要是訊息流通演算法則之理論研究，並探討分析BCJR 演算法則及視窗-事後機率演算法則之性能並以電腦模擬偵測系統之效能。除此之外也將考慮把調變碼加入部分反應通道部分反應通道中並且構成一個新的柵狀圖 (trellis diagram)以及分解圖(factor graph)，本計畫將研究及探討BCJR 演算法則及視窗-事後機率演算法則在此新型之通道中之性能。 The advanced signal processing technique referred to as the partial response and maximal likelihood (PRML) channel is usually employed in the digital recording systems to increase the recording density and/or data transmission rate. In the PRML channel, the recording channel is equalized as a partial response channel and the Viterbi algorithm (VA) is used for the detection of equalized signals. In this proposal instead of the Viterbi detection based upon the message-passing algorithm we will investigate two iterative detection algorithms (BCJR algorithm and Window APP algorithm) for the detection of the PRML recording channel. The performance of these two iterative detections will be analyzed and simulated based upon the factor graph of the PRML channel. In addition, the factor graph of the combined modulation codes and PRML channel will be studied, and based on this new factor graph the performance of the BCJR and window APP algorithm on the new channel will be also investigated. Their performance will be analyzed and compared. 研究期間：9308 ~ 9407
