| abstract |
To decode binary signals prepared according to an error-correcting convolutional code, the path metrics are transformed from the arithmetic to the logical operation area. The path metric with the extreme value is subtracted from all other path metrics. All possible binary signal states occurring on the encoder side are simulated in real time and used for addressing hypothetical source data channels. Such a hypothetical source data channel, which is fed by path decision bits, is used to update a buffer memory as well as the simulated binary signal states on the encoder side. The output of decoded data from a buffer memory is carried out in accordance with the addressing, specifically for the extreme value of the path metric. When branching in the trellis diagram, those buffer memories whose associated trellis paths do not survive are overwritten. |