應用於高速有線傳輸系統之非同步取樣式PAM-4訊號眼圖重建及參數估測演算法

Abstract

隨著科技日新月異，4G、5G行動通訊以及高畫質影像的快速發展，在高速光纖傳輸系統當中，傳統NRZ訊號所攜帶的資訊量已不夠人們的需求。因此攜帶資訊量為NRZ訊號兩倍的PAM-4訊號逐漸取代NRZ訊號成為下世代高速有線傳輸的主角。 在高速有線傳輸系統當中會隨著不同的雜訊、色散、衰減等因素影響整體系統的訊號傳輸品質，因此即時的訊號品質監測及分析對於高速有線傳輸系統來說是不可或缺的。由於高速有線傳輸系統當中訊號時脈速率較高，因此以同步方式進行訊號品質分析時需要複雜的高速時脈資料回復電路(CDR)來獲取訊號時脈資訊，再對訊號進行分析。本論文中我們採用非同步的方式在接收端直接對訊號取樣，再將取樣後的訊號進行分析。這麼做可以不必使用CDR而有效降低訊號監測的設計成本。此外，非同步的方式還可以應用在不同速率的多通道系統當中。 本篇論文採用非同步的方式取樣PAM-4訊號，再利用適當的演算法重建PAM-4訊號眼圖。又因為PAM-4訊號眼圖有別於以往NRZ訊號眼圖的單純性，因此本篇論文也提出眼圖分割、眼圖重疊方法以便估計PAM-4訊號個別子眼圖的眼寬、眼高、峰對峰抖動以及符號錯誤率，並且找出針對訊號整體之最佳取樣位置，產出系統即時的訊號品質分析。 本篇論文所設計的PAM-4眼圖參數分析演算法，具快速、簡單的特性且經過修改後，亦可適用於高階的調變訊號進行即時訊號品質監測及分析。

With the rapid development of 4G, 5G mobile communication and High Definition (HD) video, the information volume carried by traditional NRZ signal is not enough for today’s needs. Because the PAM-4 signal can carry twice the information volume of the NRZ signal, the PAM-4 signals have been adopted to replace NRZ signals in the next generation high-speed wired communication system. There are different types of noise, dispersion, attenuation, et cetera. which affect the overall signal transmission quality in the high-speed wired communication system, so real-time signal quality monitoring and analysis for high-speed wired communication system is indispensable. Due to the high data rate in high-speed wired communication system, a complex high-speed CDR (clock data recovery) circuit is needed for recovering the clock from the signal during the signal quality analysis in synchronous sampling. In this thesis, asynchronous sampling is employed on the receiver end which can avoid the use of the CDR to effectively reduce design cost of signal monitoring. A part from that, the asynchronous sampling can also be used in multi-channel systems of various data rates. In this thesis, asynchronous sampling of the PAM-4 signal is adopted, and an novel algorithm is development for reconstructing the PAM-4 signal eye diagrams. Because the PAM-4 signal is different from the NRZ signal, we proposed an eye segmentation and eye pattern overlap method for estimating the parameters of each individual PAM-4 signals’ sub-eye diagrams. The parameters include eyewides, eyeheights, peak-to-peak jitters, and symbol error rates. After extracting the signal parameters, the best sampling position for the overall signal can be determined. The PAM-4 eye-diagram parameter analysis algorithm presented in this thesis can obtain the signal parameters rapidly with a simple implementation. With slight modification, it can be applied for real-time signal quality monitoring and analysis of higher-order modulated signals.