| 摘要 |
The method includes a decomposition in elementary decodings by means of a decoder of the Soft-Input/Soft-Output (SISO) type, so that on the basis of a priori reliability information found on the transmitted symbols the values of a posteriori reliability are computed in flowchart steps (12,13,14,15) for resetting the decoding at each new iteration, and going to step (16) for updating a priori values whenn further iterations are required. The values of a posteriori reliability are computed on the basis of an extrinsic value found at step (13) by taking into account the information available on each received symbol. The method inclludes the ordering of elementary deecodings in p sets Ui of words relative to each code Ci in decreasing order according to the dependence with the received words and the capacity of correction codes, at step (12), the computation of extrinsic values of symbols represented in each set by decoding each word of the set, at step (13), the computation of a posteriori values of each symbol by finding the sum of received values and extrinsic values weighted according to the number of different symbols between closed decisions on a priori values and elementary decodings of current iteration, at step (14), and stopping iterations according to the same number of different symbols, at step (15). In the second embodiment, each set Ui of words is ordered relative to each code Ci in decreasing order of reliability, N levels comprising words of the same order of reliability are constructed, and a posteriori values of symbols of words in the levels are computed. The elementary decoding of a word relative to a code Ci is by executing an algorithm which includes sorting the reliability values by a first permutation, the determination of a code C'i equivalent to the code Ci by use of permuted purity matrix, performing a second permutation to determine useful symbols, the construction of a candidate word belonging to the code C'i, the appliccation of error patterns to the useful symbols in order to define the set of candidate words, and the computation of distance between words relative to received soft word permitted by the product of permutations. A syndrome is computed by the product in the Galois body of vector code of closed decisions on a priori values by use of parity matrix code Ci transported before computing the extrinsic values.
|
