| 摘要 |
A decoding method is started with initialization of a forward regression probability function vector alpha 0 and a backward regression probability function vector beta N. Transition probability matrixes GAMMA (Rk) and GAMMA i(Rk) are then determined depending on each received code of a sequence R1<N>. Subsequently, the value of a vector alpha k corresponding to the received Rk is determined depending on the GAMMA (Rk). A plurality of multiplications related to the GAMMA (Rk) and GAMMA i(Rk) are performed in parallel with determination of the alpha k. After receiving a complete code sequence R1<N>, all values of a backward regression probability function vector beta k (k=1, 2, ..., N-1) are determined in parallel utilizing the results of matrix multiplication and a logarithmic likelihood ratio related to each decoded bit dk (k=1, 2, ..., N) is determined in parallel. The circuit performs continuous decoding in parallel using a series of regular matrix operations.
|
