Spatially Coupled LDPC Codes via Partial Superposition and Their Application to HARQ

This paper proposes a new class of spatially coupled codes, which are constructed by sending codewords of a low-density parity-check block code (LDPC-BC) in a block Markov superposition transmission (BMST) manner, resulting in the BMST-LDPC codes. Different from the conventional spatially coupled LDPC (SC-LDPC) codes, the proposed BMST-LDPC codes can have encoder/decoder implemented using the structure of the corresponding LDPC-BCs. The proposed construction is universal, which is applicable to any existing LDPC-BCs including the 5 G New Radio (5G-NR) LDPC-BCs to obtain extra coding gains. The proposed BMST-LDPC codes are also a special class of BMST codes, which have lower error floors even with an encoding memory of one, inheriting a low decoding latency. Moreover, partial superposition, where a fraction of the coded bits are superimposed onto the following codewords, is introduced to alleviate the decoding error propagation. The construction can be optimized, for random long codes, by a protograph-based extrinsic information transfer (EXIT) chart analysis, or by a one-dimensional search for structured block codes. Furthermore, the BMST-LDPC scheme is integrated with the hybrid automatic retransmission request (HARQ) over the block fading channel, improving the throughput performance. Numerical results are presented to validate our analysis and demonstrate the performance advantage of the BMST-LDPC codes over the LDPC-BCs. They also show that the proposed HARQ scheme can yield a throughput improvement of up to 10% over the conventional HARQ scheme.

Automated summary

This paper introduces a new class of spatially coupled codes, BMST-LDPC codes, which are constructed by sending codewords of LDPC-BC in a BMST manner. The proposed codes offer lower error floors and decoding latency, and introduce partial superposition to alleviate decoding error propagation. By integrating BMST-LDPC scheme with HARQ over block fading channel, the throughput performance is improved with potential up to 10% improvement over conventional HARQ scheme.