LDPC(Low-Density Parity-Check)碼為下個世代的先進通訊標準所採用的錯誤更正碼，其優異的錯誤更正能力可以逼近Shannon的理論值，配合Sum-Product演算法以訊息傳遞的方式來解碼，可以快速得到傳送端所發出的訊息。但Sum-Product演算法之解碼複雜度偏高為主要缺點，故本篇論文將結合低複雜度解碼演算法與改善迴圈效應演算法，在不犧牲解碼效能的條件下，降低Sum-Product演算法之解碼複雜度。 結合型解碼演算法利用設定門限值(Threshold)的方式，排除部分位元節點以減少參與解碼的位元節點個數，接著在每次疊代運算完成後，判斷位元節點之值是否異號而產生振盪(Oscillation)的現象，並將振盪之位元節點加以修正，改善迴圈效應，以補償因降低解碼複雜度而衰減的效能。 經由模擬與分析的結果可知，低複雜度解碼演算法所降低的複雜度遠大於改善迴圈效應演算法所增加的複雜度，故結合型解碼演算法之複雜度確實低於Sum-Product演算法，且若設定理想的門限值，即可達到與Sum-Product演算法相同之解碼效能。 LDPC code is an error-correcting code used by the advanced communication standard of the next generation. Its error correction ability may approach the Shannon limit. Decoding by the Sum-Product algorithm with the method of message passing, we can decode the received samples at high speed. The decoding complexity of this algorithm, however, is its major disadvantage. In this thesis, we combine the low-complexity decoding algorithm and the improved cycle-effect algorithm to reduce the complexity of Sum-Product algorithm without degrading performance. Our combined decoding algorithm ignores some bit nodes by setting a threshold to decrease the number of decoding bit nodes. At the start, we first find those nodes whose values are oscillating and then try to modify them so that the cycle effect is reduced and the performance is improved. Our results show that the low-complexity decoding algorithm has much lower complexity than that of the cycle-effect one. So the complexity of the combined decoding method is lower than that of the Sum-Product one. If a threshold is properly set, the performance of the proposed algorithm will be close to that of the Sum-Product one.
