應用LASSO和QR正交化分解來達成低複雜度的Volterra濾波器並應用在長距離被動式光纖系統中

Abstract

科技和技術隨著時間快速的進步著，自二零一四年起到二零一九年，都會傳輸的傳輸速度呈現著快速成長的趨勢。為了因應未來對於高畫質和品質的傳輸需求，長距離被動式光纖已被認為是有潛力的選擇來滿足下一世代的被動光纖標準(NG-PON2). 其中，正交分頻多工系統的長距離被動式光纖網路(OFDM-LR PON)的低計算複雜度，使它成為在實際搭建高容量且覆蓋範圍廣泛網路中，一個有前途的方法。然而，在OFDM中，非線性的系統失真一直是一個需要克服的問題。因此，提出了利用Volterra濾波器來完成系統的非線性補償，但Volterra濾波器的高計算複雜度讓他在實際上難以實現。因此，我們介紹了兩種稀疏化的Volterra濾波器，並實際運用在系統當中。

Nowadays, technology and science are rapidly progressing and there is a trend of fast growth in metro traffic from 2014 to 2019. In order to meet the demand of broadband and high quantity of data transmission, long reach passive optical network(LR-PON) is considered to be the most promising candidate to meet the standard of next-generation passive optical network two(NGPON2). Furthermore, orthogonal frequency-division multiplexing long reach passive optical network(OFDM LR-PON) is regard as a hopeful solution to economically build a high-capacity and wide-coverage access network because of having low computation complexity.However, the nonlinear distortion in the system is a problem need to be solved. As a result, Volterra filtering technique is introduced to compensate the nonlinear distortion. One important factor that limits the applicability of Volterra filter is the computation complexity. Therefore, we provide two different solution to reduce the computation complexity of volterra and form two sparse volterra filter.