Simple method for limiting delay of optimized interleavers for turbo-codes

Abstract

An iterative interleaver growth algorithm is extended to allow the delay and required memory of designed interleavers to be halved with negligible performance loss. The original algorithm is efficient for two-component parallel concatenated turbo-codes with given constituent encoders that are optimum with regard to a cost function satisfying some mild conditions. However, it is only actually optimum if the selected set of patterns is representative of low-weight turbo-codewords. The new interleaver uses all terminating error patterns having an input weight not greater than a fixed IWX and single-coder output weight not greater than WX is proposed.