標題: 利用渦輪乘積碼的第二類自動重傳系統之設計
Type II Hybrid FEC/ARQ Protocols Based on Turbo Product Codes
作者: 廖志祥
Zhi-Siang Liao
蘇育德
Yu T. Su
電信工程研究所
關鍵字: 渦輪;第㆓類;自動重傳;渦輪乘積碼;Type II Hybrid FEC/ARQ Protocols;Turbo Product Codes;ARQ;Type II
公開日期: 2000
摘要: 自動重傳要求技術(automatic repeat request, ARQ)是在封包通訊(packet communication)網路㆗經常使用的錯誤控制技術。經由適當的結合前向錯誤控制碼(forward error control code)後的第㆒類和第㆓類的混合式自動重傳要求(Type I/Type II hybrid ARQ),其性能較之純粹ARQ 有不少的改善。近十年前發明的渦輪碼(turbo code) 雖在解碼的效能㆖有重大的突破,可以有十分接近謝農極限(Shannon-limit)的表現,但其解碼器複雜度高又耗時,在㆒些對延遲敏感的應用來說並不適當。鑑於乘積碼(product code)與渦輪碼在結構㆖的相似性,Pyndiah 在㆔年前發展出乘積碼的渦輪解碼法而稱這種以方塊碼(block code)為單元碼的乘積碼為渦輪乘積碼(turbo product code)。這種乘積碼具有高編碼率(大於0.65 以㆖)、低複雜度(相較於傳統渦輪碼而言)與強大改錯能力(可提供近似傳統渦輪碼的解碼效能)等優點。此篇論文㆗,探討配合在㆕次重複解碼次數(iterations)㆘,本論文目的主要在探討如何來設計㆒個結合渦輪乘積碼的第㆓類混合式自動重傳系統。乘積碼的最小距離(minimum distance)由其組成的單元方塊碼(component block code)來決定,因 此,我們若對資料分別編成不同最小距離的乘積碼字(codewords)並儲存相關的對偶位元(parity bits)部分,有重傳需要時,我們只要依序傳送出不同的對偶位元部分,接收端得以在每次接收重傳後的信號時,配合先前收到的封包便可解最小距離遞增的乘積碼,逐步加強改錯能力。我們估計這種ARQ 方式在㆔種不同運作情況(operation scenarios)㆘的輸出率(throughput)與延遲(delay)。第㆒種情況是沒有封包碰撞的環境,第㆓種情況考慮Aloha 多重進接(multi-access)系統,我們假設封包碰撞時其同步前端(synchronization preamble)受損將使得封包無法被偵測到。最後㆒種情況我們試著探討分碼多工(code division multiple access)系統利用直序列展頻緩和碰撞影響的效果。就㆖述㆔種運作情況,我們分別討論了在白色高思雜訊(AWGN)通道及瑞雷衰退(Rayleigh fading)通道㆘的表現。本論文的第㆓章討論沒有多重進接干擾(multiple access interference)的單㆒連結通訊(single link communication),介紹調變/解調(modulation/demodulation)的結構、通道模式與封包型式(packet format)。第㆔章簡單描述了多重進接網路(multi-access network),並比較ALOHA與分碼多工兩種架構。第㆕章介紹渦輪乘積碼基本的解碼原理,並且展示渦輪乘積碼在前兩章所描述的環境㆘之性能。第五章㆗提出了配合結合渦輪乘積碼的第㆓類混合式自動要求重傳設計,其㆗包含了分析輸出率(throughput)與延遲(delay)作為評量此系統的依,並在最後提出模擬結果。第六章包含我們的總結與㆒些系統設計的建議。
The throughput performance improvement of a type II FEC/ARQ protocol is derived from the fact that it transmits only incremental and usually nonoverlapping redundancy in response to retransmission requests. This thesis presents several type II hybrid FEC/ARQ protocols using turbo product codes (TPCs). A TPC has the same encoder structure as that of a conventional product code but utilizes a lower complexity iterative decoding method [1] that yields a performance close to that of a suboptimal convolutional turbo decoder. The properties of low-complexity and excellent performance make TPC a very attractive candidate for packet communication systems using ARQ. The throughput and delay performance of the proposed ARQ systems is analyzed under both single link and multiple access scenarios. We consider two special cases: collision-free and collide-and-die. The former case assumes no collision while the later case assume a complete loss of a packet once the synchronization preamble of a packet is interfered by other packets. In the multi-access scenario, we examine how multiple access interference (MAI) and other design parameters affect the system performance using either the ALOHA or the CDMA protocol. We also present an adaptive ARQ protocol that attains best balance between latency and throughput. We evaluate the performance of various protocol for both AWGN and fading channels. Simulation results show that the proposed protocols can achieve high throughput with minimum latency at high SNRs and acceptable delay and respectable packet error rate performance at low to middle SNRs.
URI: http://140.113.39.130/cdrfb3/record/nctu/#NT890435005
http://hdl.handle.net/11536/67285
顯示於類別:畢業論文