Conference Paper

Optimized symbol mapping for bit-interleaved coded modulation with iterative decoding

Sch. of Inf. Eng., Commun. Univ. of China, Beijing
DOI: 10.1109/ICCT.2008.4716215 Conference: Communication Technology, 2008. ICCT 2008. 11th IEEE International Conference on
Source: IEEE Xplore

ABSTRACT Symbol mapping is very crucial for the performance of bit-interleaved coded modulation with iterative decoding (BICM-ID). In this paper, a new scheme of symbol mapping called Cross-8PSK-Quasi-Gray is proposed. The optimization scheme of Cross-8PSK is composed of two QPSK with different radius and phases. Through analysis and comparisons with three previously mentioned symbol mappings in terms of channel capacity and error performance, simulation results show that Cross-8PSK-Quasi-Gray mapping significantly outperforms set partitioning, semi set partitioning labeling maps in the aspects of both channel capacity and error performance. The overall performance of Cross-8PSK-Quasi-Gray mapping is nearly close to gray mapping in conventional 8PSK. Thus, Cross-8PSK-Quasi-Gray mapping is a good scheme for designing the power-efficient BICM.

  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: We propose a block coded Turbo DeCodulation scheme, combining iterative demodulation, channel decoding, and source decoding at the receiver. Turbo DeCodulation resembles a serial concatenation of bit-interleaved coded modulation (BICM) and iterative source-channel decoding (ISCD) where the two iterative loops are connected by a single convolutional code. In our contribution, explicit channel (de)coding by a convolutional code, used in most Turbo systems, is omitted. Instead, highly redundant index assignments and multi-dimensional mappings, both based on short block codes, introduce the artificial redundancy. Thus, the proposed receiver possesses only a single iterative loop. The simulations results show a competitive or even superior performance compared to previously known systems. Furthermore, we present an extension to a flexible multi-mode system and a novel, simple design method for the index assignments.
    Turbo Codes&Related Topics; 6th International ITG-Conference on Source and Channel Coding (TURBOCODING), 2006 4th International Symposium on; 05/2006
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: In this paper, we investigate a coded OFDM system employing an iterative receiver, which is a serial concatenation of an inner and outer component decoder. The inner component decoder includes pilot and data aided channel estimation (CE) and demodulation whereas the outer component decoder consists of the channel code decoder. Between these two components extrinsic log-likelihood ratio values of the code bits are iterated to reduce the bit-error rate (BER) of the information bits. The main contribution of this work is to increase convergence speed by feeding the CE with a-posteriori instead of extrinsic data estimates. Besides bit-error rate (BER) plots, we use extrinsic information transfer (EXIT) charts to analyze and predict the performance of the iterative receiver. Further, we investigate hard decision feedback to the CE and the exclusion of a-priori knowledge in either the demodulator or CE.
    Wireless Personal Multimedia Communications (WPMC) Symposium 2006; 09/2006
  • [Show abstract] [Hide abstract]
    ABSTRACT: We present multi-dimensional binary mappings for bit-interleaved coded modulations on ergodic multiple antenna channels with iterative decoding. We established a closed form expression for the pairwise error probability under ideal maximum likelihood decoding. The design criterion for mapping optimization is established from the maximum likelihood performance of the ideally interleaved channel. It coincides with the gure of merit derived from the genie condition when the iterative receiver converges to perfect a priori information. We present multi-dimensional mapping constructions that exhibit high signal to noise ratio gains without increasing the complexity of the a posteriori probability detection. Also, the proposed multi-dimensional mappings allow for a reduced decoding complexity: they achieve near turbo code performance with a single convolutional code.