疊代型整合式訊源/通道解碼其主要概念,是藉由訊源解碼器(source decoder)與通道解碼器(channel decoder)彼此之間資訊的交換,來達到降低解碼之錯誤率。在疊代解碼中,一般常使用軟式輸出(soft output)BCJR演算法[6],獲得最佳錯誤更正之效能,但其計算複雜度會隨著格狀結構(trellis-based)上之節點個數增加而呈指數成長,故使用循序式解碼演算法(Sequential Decoding Algorithm),可降低解碼時之計算複雜度。本論文主要探討樹狀結構(tree-based)解碼演算法和格狀結構(trellis-based)解碼演算法,兩者間計算複雜度和效能之差異。
The main concept of the Iterative Joint Source-Channel Decoding is to achieve lower decoding error rate by the source decoder and the channel decoder in exchange of the information between each other. In the iterative decoding, we usually use the BCJR soft output algorithm to get the better error correcting performance. However, the decoding complexity will increase exponentially as the numbers of trellis states gradually becomes larger. Hence, the trellis-based sequential and its soft output decoding algorithms are proposed to reduce the computational complexity. Both the computational complexity and the performance between the tree-based and the trellis-based decoding algorithms are compared thoroughly.