Summary form only given. Maximum-likelihood soft decision decoding of an (n, k, d) binary linear block code is performable by complementing the hard-detected version of the received word in at most m = n - k positions. The set of positions, or pattern, is selected according to the least sum of reliabilities of the associated bits. The author has developed a method whereby out of all the patterns with cardinality m, m - 1, and m - 2, fewer explicitly described patterns have to be scored than previously, provided that d ≥ 3. This approach allows decoding of the (15,11,3) code by at most 51 real additions, compared with the previous best 83 additions required in the worst case by a different search scheme. Decoding of the (31, 26, 3) and (32, 26, 4) codes is accomplished by fewer than 200 additions in the worst case versus more than 1000 additions performed by any previously published decoder. Application of reduced lists of patterns to coset decoding of medium-rate codes has also been addressed.
|Number of pages||1|
|State||Published - 1990|
|Event||1990 IEEE International Symposium on Information Theory - San Diego, CA, USA|
Duration: 14 Jan 1990 → 19 Jan 1990
|Conference||1990 IEEE International Symposium on Information Theory|
|City||San Diego, CA, USA|
|Period||14/01/90 → 19/01/90|