摘 要 迴旋碼的解碼法則中，以最大相似解碼法則---腓特比演算法為公認的有效率之近似解碼演算法。本文主要是針對數位電視地面廣播系統中所訂定的間空迴旋碼，希望設計實現出一個符合系統解碼速度的腓特比解碼器，且研究於軟判決的解碼方式，以希望提升錯誤保護之能力。 軟判決解法方式，是依照最大相似解碼法則衍生出最小距離法則，再按照量化位階的不同給予不同的分支路徑計量値，本論文內分析比較硬判決以及軟判決量化位階為Q=8、Q=16的效能，且提供硬體實現上，參數設定的依據。硬體實現上，比較不同的實作方式，採用全平行的架構，為了能夠更加快速度且節省資源，使用模數歸一的正規化方式；最後，以較簡單的3-指標偶數區塊追溯法來完成整個解碼器。在計算分支路徑計量値時，使用量化位階為Q=16的方式。 ABSTRACT Viterbi algorithm is the well-known and efficiently maximum likelihood decoding algorithm applied to decode convolutional code. In this discussion, it concentrates on the standard of punctured convolutional encoder in DVB-T system. To design and implement a Viterbi decoder agreed with the decoding throughput which limited by the system, on the other hand, to research the soft-decision method for decoding process with the result in promoting the bit error rate！ The way used to soft-decision is by the minimum distance algorithm . Bit branch metrics have different value if the quantized level is not the same for soft-decision decoding. The analysis is focusing on the performance of soft-decision quantized level Q=8 and Q=16 and hard-dec ision and a basis for the coming IC design of Viterbi decoder. In the implementation issue, full parallel architecture and modulo normalization are used to save the source and decoding throughput, then the 3-pinter even block method is used to trace back and decoding the data. At the last, we implement the Viterbi decoder by above-mentioned ways and the soft-decision quantized level is Q=16.
