Soft-decision decoding of a linear block code using the most reliable basis corresponding to each received word is investigated. Based either on probabilistic properties or on the structure of the code considered, three improvements to the algorithm devised by Fossorier and Lin (see ibid., vol.41, no.9, p.1379-1396, 1995) are presented. These modifications allow large computation savings or significant decoding speedup with little error performance degradation. First, a reduced probabilistic list of codeword candidates is associated with order-i reprocessing of a given code. It results in a large reduction of the maximum number of computations with a very small degradation in performance. Then, a probabilistic stopping criterion is introduced for order-0 reprocessing. This new test significantly decreases the average number of computations when appropriately implemented. Finally, the application of the algorithm to coset decoding is considered for |u|u+v| constructed codes. In addition to the conventional coset decoding, a new adaptive practically optimum coset decoding method is presented where at each reprocessing stage, the number of surviving cosets decreases. Suboptimum closest coset decoding is also investigated. It is shown that two-stage decoding with the algorithm of Fossorier and Lin offers a large variety of choices, since the reprocessing order of each stage can be determined independently.
展开▼
机译：研究了使用与每个接收到的字对应的最可靠的基础对线性分组码进行软判决解码。基于概率性质或所考虑的代码结构，对Fossorier和Lin设计的算法进行了三处改进（同上，第41卷，第9期，第1379-1396页，1995年）。这些修改可节省大量计算或显着提高解码速度，而不会降低错误性能。首先，将减少的码字候选者概率列表与给定代码的i阶重新处理相关联。这会导致最大计算数量的大幅减少，而性能的下降却很小。然后，引入了概率停止准则以进行零阶重新处理。如果适当实施，此新测试将大大减少平均计算量。最后，考虑| u | u + v |在同集解码中的应用。构造代码。除了常规的陪集解码之外，还提出了一种新的实用的，最佳的自适应陪集解码方法，其中，在每个再处理阶段，尚存的陪集数量都会减少。还研究了次最佳最近陪集解码。结果表明，采用Fossorier和Lin算法进行的两阶段解码提供了多种选择，因为每个阶段的重新处理顺序可以独立确定。
展开▼
