Soft decision decoding is a difficult search problem, for which optimal algorithms are computationally intractable. Genetic algorithms (GA) are stochastic optimisation techniques that have successfully solved many difficult search problems. We have developed a high performance GA for suboptimal soft decision decoding of binary linear block codes, which gives bit error probabilities as low as 0.00183 for a [104, 52] extended quadratic residue code with a signal-to-noise ratio of 2.5 dB, exploring only 30,000 codewords, whereas the search space contains 10/sup 1/5 codewords. Success ensues from the use of a new crossover operator that exploits problem-specific knowledge.
展开▼
