Evolutionary Algorithm Aided Interleaver Design for Serially Concatenated Codes

Download

PDF - Published Version Download (200Kb)

Description/Abstract

In this paper, we propose an algorithm for designing the interleavers of Serially Concatenated Codes (SCCs), in order to increase the Minimum Hamming Distance (MHD) between the legitimate permutations of the encoded bit sequence and hence to improve the corresponding error floor. Unlike previous so-called Code Matched Interleaver (CMI) designs, our approach is capable of creating interleavers for serial concatenations of both irregular and non-linear codes, as well as achieving MHDs that are arbitrarily close to the maximum possible, provided that a sufficiently high off-line complexity is affordable. However, owing to the efficiency of the proposed approach, only a relatively low number of algorithm generations are required to achieve significant improvements to the error floor of low-delay wireless sensor network, speech and audio schemes, for example. Indeed, we demonstrate that our interleavers are capable of completely eradicating the error floors that would otherwise be apparent, if classic random or S-random interleavers were employed.