In this paper

fast decoding algorithms for convolutional codes

which are based on the genetic algorithm (GA)

are presented.They search from one side or both sides of the trellis respectively to implement unidirectional decoding (U-GA) or bi-directional decoding (B-GA).With good diversity

wide search region and global optimal ability of GA

the decoding performance can be improved.The effects of U- and B-GA and population size M on P

b

are analyzed.Simulation results show that

with the same decoding computation efforts and at P

b

=10

-6

B-GA achieves about 0.5 dB coding gains over M- algorithm (MA) for the coding constraint length K=19 code

and 1dB gain over Viterbi algorithm (VA) with the K=7 code.