Improved linear-time metric sifting in breadth-first decoding of convolutional codes

We describe a linear-time method for finding the best metrics during breadth-first reduced search decoding of convolutional codes. This method is an improved version of one of the merge sifting methods described by Kot (1993). As an example, in the M-algorithm decoding of rate 1/2 convolutional codes, the present method reduces the number of comparisons used to provide a sorted set of survivor metrics from 3M-1 to 2M+1. This is a significant improvement over comparison-based sorting, which requires O(M log M) comparisons, and offers a similar improvement over well-known comparison-based selection methods