Spatially Coupled 5G LDPC Codes via Superposition

We propose in this paper to enhance the 5G low-density parity-check (LDPC) codes by transmitting the codewords in a block Markov superposition transmission (BMST) manner, resulting in a class of spatially coupled LDPC codes. We present a generalized extrinsic information transfer (EXIT) chart for performance analysis, showing that the decoding thresholds can be improved by increasing the memory size (coupled width) <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$m$</tex>. However, for m > 1, the receiver requires a relatively large window size for the sliding window decoding (SWD) algorithm, potentially causing unacceptable decoding latency. To address this issue, we propose an adaptive sliding window decoding (ASWD) algorithm, in which the decoding window size depends on the decoding state of the BMST system. The proposed EXIT chart analysis can also effectively guide the setting of the maximum decoding window in the ASWD algorithm, as well as the setting of the superposition fractions for the BMST-5G-LDPC system. Simulation results show that: 1) the ASWD algorithm can effectively reduce the average decoding window size in the medium to high signal-to-noise ratio (SNR) region, without the performance loss compared to the conventional SWD algorithm; 2) the proposed BMST-5G-LDPC code can achieve about 0.4, 0.5 and 1.7 dB performance gains compared to the 5G LDPC code over the AWGN channel, the fast fading channel, and the quasi-static block fading channel, respectively.