整合訊源-通道編碼調變之研究

Abstract

在這篇論文中，我們將論文[2]的搜尋最佳或近於最佳之整合訊源-通道編碼演算法加以拓展，以尋求最佳之整合訊源-通道編碼與調變之設計，我們將之稱為整合訊源-通道編碼調變(joint source-channel coded modulation or JSCCM)。簡言之，在給定自由歐幾里得距離(free Euclidean distance)的最小允許值的前提下，運用優先權搜尋演算法(priority first search algorithm)，在特定的搜尋樹結構中，找出的平均碼長最小的整合訊源-通道編碼調變。在接收端，我們也將論文[2]所提出之針對已知傳送訊源字的個數與調變符元個數的條件下所設計的最大事後概率(maximum a posteriori)解碼演算法加以拓展，以解調所搜尋到的整合訊源-通道編碼調變。模擬結果顯示，相較於將訊源-通道與調變分別獨立設計之系統，在高訊雜比的情況下，整合訊源-通道編碼調變確實可達到更低的傳輸錯誤率。

In this thesis, the code construction algorithm that searches for an optimal or near-optimal joint source-channel code in [2] is further generalized. The new construction algorithm targets for an optimal joint design between source-channel coding and modulation, which is refereed to as joint source-channel coded modulation (JSCCM). Subject to a fixed free Euclidean distance constraint, a priority-first search algorithm is applied over a search tree for finding an optimal JSCCM with the minimum average codeword length (measured in source letters per modulation symbol). On the other hand, the sequence maximum a posteriori (MAP) decoder for joint source-channel codes is also generalized for the MAP demodulation of JSCCMs under the premise that the receiver knows both the number of transmitted codewords and the number of transmitted modulation symbols in a transmission block. Simulation results show that a joint design between source-channel coding and modulation can achieve a better performance than a design that separates source-channel coding and modulation, in particular at high SNRs.