Abstract
光學與磁記憶系統乃是資料儲存系統的兩大主流，對高速的讀寫應用而言，磁記憶系統仍是至今較佳的選擇。有許多的技術可提高其記錄密度及品質：尖峰檢測器是使用最廣的檢測技術；在低密度及普通的密度之下，可使用波形苗條器及寫入電流整形器可使波形變的較為纖細以減少相互干擾所造成的效應，而當密度提高時，這些方法將增加高頻的雜訊而降低性能，而部份響應訊號處理則可用來控制波形之間的互相影響且避免增加高頻雜訊的份量。另外，有一種稱為寫入等化器的高通濾波器也可在不增加高頻雜訊份量的情況之下，達到補償磁記憶通道頻率響應的目的。除了等化的技術之外，編碼技術也可用來提高磁記憶系統的品質，持續長度限制碼就是一種最長用來減少波形間相互干擾的編碼技術。曾有人提到在勞侖茲線性通道模形及可加式白色高斯雜訊之下，使用持續長度限制碼將會降低MLSE檢測器的表現。在這篇論文□，我們以抹模擬的方式分別評估持續長度限制碼配合尖峰檢測器及MLSE檢測器的表現。由結果可看到持續長度限制碼有編碼損失的現像，而另一方面，它對尖峰檢測器則是有好處的。這篇論文的結構如下：第二章描述磁記憶通道模型，第三章提到針對磁記憶通道的一些等化技術，第四章提到持續長度限制碼的一些性質，第五章則是模擬的結果，最後，第六章是結論。Optical and magnetic recording systems are two main streams ofdata storage systems. For high data rate read/writeapplications, magentic recording seems to be a better choice upto now. A lot of techniques have been derived to increase thestorage density and improve the quality. Peak detection is themost widely used detecting technique. Pulse sliming and writecurrent shaping are used to "slim" the pulse in order to reducethe effect of ISI (intersymbol interference) at low andmoderate density. As the linear density is increased, thesetechniques enhance too much noise . Partial response signalingis used to control ISI and avoid this situation. Viterbidetector has been used to improve the performance. Equalizer atthe writing side, the write equalizer, is a high pass filter tocompensate the frequency response of magnetic channels withoutnoise enhancement. Beside equalization techniques, codingtechniques are also investigated. Rll (run-length limited)codes are widely used to reduce the effect of ISI thus improvethe performance. K.A.S Immink mentioned that RLL codes resultin coding loss and degrade the performance of MLSE (maximumlikelihood sequence estimation) detector over the Lorentzianlinear channel with AWGN (additive white Gaussian noise). Inthis thesis, performance of RLL coddes with peak detection andMLSE detection has been evaluated by simulaion. The resultsconfirm the coding loss phenomina and indicate the advantagesof RLL codes in peak detection. This thesis is organized asfollows. The magnetic recording channel models are described inChapter 2. In Chapter3, some equalization techniques aimingdirectly at the magnetic recording channels are described. InChapter 4, some properties of RLL sequences are narrated. InChapter 5, performance of RLL codes with peak detection andMLSE detection is evaluated by simulation respectively.Finally, a conclusion is given in Chapter 6.