An alternative to metric rescaling in Viterbi decoders

In the Viterbi algorithm, the negative log-likelihood estimates, accumulated distances, or path metrics are unboundedly increasing functions of time. For implementation, all variables must be confined to a finite range. The following properties of the Viterbi algorithm can be exploited for this purpose: (1) path selection depends only on differences of metrics; an (2) the difference between metrics is bounded. In the rescaling scheme, at each iteration the minimum metric is subtracted from all metrics. The use of two´s complement arithmetic is proposed as an alternative to the rescaling method. This scheme avoids any kind of rescaling subtractions. Obvious advantages in implementation are hardware savings and a speedup inside the metric update loop, which is critical to the decoder´s computational throughput