Applying genetic algorithms to the information sets search problem

Error-correction coding is essentially a signal processing technique that is used to improve the reliability of digital communication systems. In the Information Set (IS) decoding a collection of information sets are used to generate candidate codewords. The procedure then selects as the decoded codeword the one which is closest to the received sequence. This decoding approach can reduces the complexity and time processing in comparison to maximum-likelihood decoding (MLD), and may presents the same decoding level. The performance of the IS algorithm depends on the number of error patterns that the collection of information sets can cover. There are no known constructive procedures for finding optimum collection of information sets. This paper presents an approach for obtaining optimum collections of information sets using genetic algorithms. Using this approach we found, in short time, collections of information sets with high covering capacity. Results from computer simulation show that the performance of the IS algorithm with optimized collections is nearly identical to the maximum-likelihood decoding.