一種應用於准循環低密度奇偶檢查碼解碼之再編碼預處理技術及改進解碼矩陣架構

Abstract

由於低密度對偶檢查碼(LDPC)的編碼效能十分逼近夏儂極限 (Shannon limit) 以及編碼程序上有低複雜度的特性，所以在近年受到廣泛的討論。在這些討論之中，有關於解碼演算法的部分，主要是以改良解碼演算法，或是解碼的硬體架構來取得權衡效能與複雜度。效能包含了解碼的位元錯誤率，解碼的吞吐量，或是解碼所耗之能量等等的部分去做討論。鮮少有和解碼的延遲時間有關之研究。但為了提升低密度對偶檢查碼的解碼效能，長度較長的碼長是必須的。加上在低密度對偶檢查碼解碼步驟中，需要先得到完整的碼字後才能開始解碼的步驟，所以至少會延遲等待接收碼字的時間。所以隨著通訊要求的提升，若要兼具通訊的速度和品質，解碼的延遲時間就會顯的越來越重要。 本篇論文題出一個創新的想法從利用接收碼字的時間來改善解碼的延遲時間。在接收碼字的過程中，我們以本篇論文所提出之預處理技術對已收到的碼字訊息做修改的動作，使得碼字在解碼時能減少解碼的遞迴數，已達到改善解碼的延遲時間的目的。

Since their performances are very close to Shannon limit, low-density parity-check (LDPC) codes have attracted a lot of attention since recent years. Although there are many articles discussing LDPC codes, the researches about the LDPC decoding almost focus on modifying decoding algorithms or hardware implementations, in order to achieve lower bit error rates, reduce hardware complexities, increase decoder throughputs, or reduce decoder power consumption. And there are very few researches about the latency of LDPC decoding. But if we want better decoding performances, long codeword lengths are needed. Since conventionally one needs to receive the whole codeword before decode LDPC code, a long code length means a long waiting time to start decoding procedures. In this thesis, we propose a novel preprocessing scheme on the received codework, for reducing the decoding latency. While receiving the codeword, we can simultaneously apply the preprocessing technique that modifies the received bit messages. And after received whole codeword, we decode the modified messages which can reduce the decoding iterations to achieve the goal of reducing the decoding latency.