| 摘要 |
A source sequence of symbols is communicated over a communications medium by encoding the source sequence according to respective first and second error correction codes to produce respective first and second encoded sequences of symbols. The first and second encoded sequences are processed to produce a communications signal, which is then communicated over the communications medium. The communicated communications signal is processed to produce first and second received sequences of symbols corresponding to the first and second encoded sequences, respectively. The first and second received sequences are selectively recursively decoded according to the associated error correction codes augmented by previous estimates of a symbol of the source sequence to repeatedly produce revised estimates of a symbol until an estimate satisfying a predetermined reliability criteria is obtained. The first and second received sequences may be selectively recursively decoded by maximum likelihood decoding the first received sequence to produce a first estimate of a symbol of the source sequence and a path metric associated therewith, and decoding the second received sequence to produce a second estimate of the first symbol if a predetermined function of the path metric associated with the first estimate is outside of a predetermined range. Similarly, the received sequences may be selectively decoded by maximum a posteriori (MAP) decoding the first received sequence to produce a first MAP estimate of a symbol of the source sequence and a reliability metric associated therewith, and decoding the second received sequence to produce a second estimate of the first symbol if the reliability metric associated with the first MAP estimate is outside of a predetermined range. Encoding of the source sequnce according to the first and second error correction codes may be preceded by encoding the source sequence according to an error detection code, e.g., a CRC, and selective recursive decoding may include decoding the first received sequence according to the first error correction code to produce an estimate of a symbol of the source sequence, and then decoding the estimate according to the error detection code to produce the reliability metric for the estimate.
|
