Cross-entropy based symbol selection and partial iterative decoding for serial concatenated convolutional codes

Cross-entropy based symbol selection and partial iterative decoding for serial concatenated convolutional codes (SCCC) is developed in this paper. We first apply symbol classification in terms of detection convergence status to all coded bits of the outer code. After that, computations are focused on symbols that have not reached convergence, by applying short windows around these symbols at both the inner and outer component decoders. By utilizing the converged soft output from their neighboring symbols and exchanging extrinsic information only among the windowed symbols, the component decoders performs very closely to standard full iterations but at only a small fraction of the original computational complexity.