A reduced complexity decoder using compact genetic algorithm for linear block codes

The compact Genetic Algorithm decoder has been introduced in [9] as an efficient decoding method of linear block codes. It requires less storage memory than Genetic Algorithms based decoders. One of its major weakness is the big number of necessary iterations to reach convergence in comparison with Genetic Algorithms (GA) based decoders. We propose, in this work, new ideas allowing us to reduce the number of iterations from about 105 to just about 103 which reduces the complexity of decoding. This, without decreasing the decoding performance. We introduce a new stopping criterion based on the soft weight of the probability vector p, a new initialization method of p and we tried to combine both methods all together. Both performance study and the calculation of the average number of iterations ensure the effectiveness of the proposed decoder.