English  |  正體中文  |  简体中文  |  Items with full text/Total items : 43312/67235
Visitors : 2078981      Online Users : 31
RC Version 5.0 © Powered By DSPACE, MIT. Enhanced by NTU/NCHU Library IR team.
National Chung Hsing University Institutional Repository - NCHUIR > 工學院 > 電機工程學系所 > 依資料類型分類 > 碩博士論文 >  16-QAM 無線光纖傳輸系統之多碼率低密度同位元查核碼解碼器實作與後處理器研究

Please use this identifier to cite or link to this item: http://nchuir.lib.nchu.edu.tw/handle/309270000/116015

標題: 16-QAM 無線光纖傳輸系統之多碼率低密度同位元查核碼解碼器實作與後處理器研究
Implementation and Study of Post Processors for Multi-Rate LDPC Decoders in 16-QAM Radio-Over-Fiber Systems
作者: 蕭家州
Hsiao, Chia-Chou
Contributors: 張振豪;許騰尹
Chen-Hao Chang;Terng-Yin Hsu
林泓均
Hong-Chin Lin
中興大學
關鍵字: 多碼率;低密度同位元查核碼解碼器;無線光纖傳輸;錯誤基數;里德索羅門碼;後處理
16-QAM;Low Density Parity Check;multi-rate;Radio Over Fiber;Layered Min Sum Algorithm;error floor;Reed-Solomon;Check Node Tracing Method;Reduced-Back-Tracking Method;Absorbing set;trapping set;post-processing
日期: 2012
Issue Date: 2012-09-11 10:27:45 (UTC+8)
Publisher: 電機工程學系所
摘要: 本論文提出應用於16-QAM之無線光載射頻通道(Radio over Fiber, ROF)之多碼率LDPC解碼器 VLSI設計與實作。本主題有幾個主要的重點 : (1).建構三個分割轉移LDPC Code (Partition and Shift LDPC, PS-LDPC),其查核矩陣分別為(480,2400)、(800,2400)以及(480,4800),碼率分別是4/5、2/3以及9/10,其Girth分別為8、6、6,三者都具有優秀的更正能力。(2).利用化簡後的Layered Min Sum Algorithm搭配LLR數值量化擴增的方式,在硬體複雜度不高的同時,亦能達到良好的效能。(3).LDPC解碼器架構採用雙路徑部分平行式架構,在切管線使頻率上升的同時,亦不造成電路閒置的情形,使傳輸率倍增。 (4).三個查核矩陣盡量共用硬體,包括計算單元與暫存器的硬體共用技巧,來減少所需耗費的面積。使用聯華電子UMC 90nm CMOS技術實作後,晶片核心面積為8.19 ,在電壓供應0.9伏特時,最高工作頻率為110MHz。固定解碼次數8次下,此三個查核矩陣(480,2400)、(800,2400)、(480,4800)所對應的傳輸率分別為5.5Gbps、3.3Gbps、2.75Gbps,其功率消耗分別為332.7mW、339.5mW、368.6mW。

其次,本論文針對LDPC解碼器常發生之錯誤基數(Error Floor)進行研究與探討。此議題可在LDPC解碼器之後加RS解碼器解決,我們利用UMC 90nm 實作4/5碼率之LDPC加RS解碼器晶片,在0.9V供應電壓下,經過實際量測後RS解碼器最高頻率為202M Hz,LDPC解碼器最高頻率為67.33MHz。傳輸率可以達到3.58Gbps,核心電路功耗為63.12mW。本論文亦提出了第二種有效降低Error Floor的演算法,此演算法共分2個階段,都是在LDPC完成解碼之iteration後才實施,當作「後處理」的動作。第一階段為Check Node Tracing Method,利用解碼完後判斷Unsatisfied 查核點的位置,往前追朔並找出重疊的變數點,把大部分錯誤更正回來;第二階段為Reduced-Back-Tracking Method,主要是用來解決少數剩餘的錯誤,而這些錯誤大部分是Absorbing Set的結構。我們從16 QAM通道解調端所接收的LLR值,利用4-bit量化的模擬,在解碼之後採用了本論文所提出的後處理方式,的確會比沒使用後處理的情況更能有效的降低Error Floor的發生,由於減少運算所需的位元數,預期可減小硬體複雜度,同時保有優良的解碼效能。
In this thesis, VLSI implementation of a multi-rate low-density parity-check (LDPC) decoder used in 16-QAM Radio over Fiber (ROF) channel is presented with four important features. The first is constructing three Partition and Shift LDPC (PS-LDPC) Codes with the parity check matrices (H) of (480, 2400), (800, 2400) and (480, 4800)corresponding to the coding rates of 4/5, 2/3, 9/10, and girths of 8, 6, 6, respectively. Secondly, good performance was achieved by a modified Layered Min Sum algorithm using quantization expansion of log likelihood ratio for less hardware complexity. The third is the proposed dual path partial parallel architecture using pipeline can increase the operating frequency, and double the throughput with the little circuit overhead. The last feature is the three LDPC codes share the hardware as much as possible, including the computing units and registers to reduce the chip area. Using the UMC 90nm COMS technology, the maximum frequency reaches 110MHz with the core area of 8.19mm2 at supply voltage of 0.9V. With 8 iterations per decoding process, the throughputs of PS-LDPC codes (480, 2400), (800, 2400) and (480, 4800) are 5.5Gbps, 3.3Gbps, 2.75 Gbps with the power consumptions of 332.7mW, 339.5mW and 368.6mW, respectively.

The error floor is the other common issue in LDPC decoders. This problem can be solved by adding the Reed-Solomon (RS) decoder after the LDPC decoder. It was fabricated using the UMC 90 nm CMOS process for the coding rate of 4/5. The measured maximum frequency of RS decoder is 202MHz, and the measured maximum frequency of LDPC decoder is 67.33MHz at supply voltage of 0.9V. The throughput is 3.58 Gbps and the power consumption is 63.12mW. The other method to lower the error floor employs the post-processing technique. It consists of two steps after the normal LDPC decoding procedures. The first step is called Check Node Tracing Method. By checking the locations of unsatisfied check nodes, the overlapped variable nodes may be traced, and their values are flipped to the correct values. The second step is called Reduced-Back-Tracking Method. The main objective is to correct the remaining errors. Most of them are induced by the absorbing sets. The simulation results show that the error floor is significantly improved for 4-bit Log likelihood ratios in 16-QAM demodulation. It is convinced that it can reduce hardware complexity due to less quantization bits, while keeping good decoding performance.
Appears in Collections:[依資料類型分類] 碩博士論文

Files in This Item:

File SizeFormat
index.html0KbHTML576View/Open


 


學術資源

著作權聲明

本網站為收錄中興大學學術著作及學術產出,已積極向著作權人取得全文授權,並盡力防止侵害著作權人之權益。如仍發現本網站之數位內容有侵害著作權人權益情事者,請權利人通知本網站維護人員,將盡速為您處理。

本網站之數位內容為國立中興大學所收錄之機構典藏,無償提供學術研究與公眾教育等公益性使用。

聯絡網站維護人員:wyhuang@nchu.edu.tw,04-22840290 # 412。

DSpace Software Copyright © 2002-2004  MIT &  Hewlett-Packard  /   Enhanced by   NTU/NCHU Library IR team Copyright ©   - Feedback