Forced-convergence decoding for LDPC-coded modulation

Abstract

Low-density parity-check (LDPC)-coded modulation has attracted much attention for its outstanding performance in a bandwidth-limited channel. However, it involves relatively high decoding complexity. Owing to different convergence behaviors of variable nodes, the nodes detected early can be deleted from the Tanner graph, which reduces the size of the parity-check matrix iteration by iteration and thus the decoding complexity. In this paper, we propose a new decoding algorithm for LDPC-coded modulation with reduced complexity. In addition, by deleting the converged bits from the modulation constellation and updating the bit metrics of the demapper with the reduced constellation, more accurate estimation of the bits can be achieved. Simulation results show that, compared with conventional decoding schemes, the proposed method not only greatly reduces the overall decoding complexity for LDPC-coded modulation but also significantly improves decoding performance.

Large-Scale Multi-User Distributed Antenna System for 5G Wireless Communications

“…A more promising method to exploit the sparsification of the channel matrix is BP [12], [13] algorithm, which is applied in factor graph (FG)-based graphical model with Gaussian approximation of interference (GAI).In this case, the overall complexity of the detection is O (MN × K), and also doesn't need matrix inversion among the detection.…”