摘要 |
In order to decode a convolution coded data sequence containing errors at least a first and a second version are generated which are decoded essentially simultaneously in order to generate a decoded sequence. The decoding is carried out in periods 21-24, 24-27, 27- 30, 30-33 in accordance with the bind length of the convolution coder, where at the end of the period the most probable is chosen of the decoding paths generated by the different versions which come to the respective trellis states of the decoding operation. At the start of a new period the decoding continues from each trellis state of the decoding operation in accordance with all versions. <IMAGE>
|