Transceiver Design with Iterative Decoding of Capacity-Approaching codes over Fading channels

Abstract

Firstly, we consider LDPC and turbo coded transmissions with pilot-symbol-assisted (PSA) channel estimation over time correlated Rayleigh channels with known statistics. The correct conceptual approach is presented for deriving the LLR metric, which is shown to outperform the conventional metric in both error performance and convergence speed. Our work demonstrates the importance of incorporating the knowledge of channel estimation accuracy into iterative decoding processes.

We also propose a joint maximum-likelihood detector based on soft-output Viterbi algorithm for LDPC coded transmissions over block-wise static fading channels with unknown fading statistics. Computer simulations show that the proposed detector has substantially better error performance than the conventional system with PSA channel estimation.

Finally, a doubly multistage decoder (DMD) is proposed for LDPC codes over integer residue rings. Two variants of the DMD are considered and shown to achieve significant coding gains over standard BP decoding, with remarkably lower computational complexity.