[Paper Review] Improving the Finite-Length Performance of Spatially Coupled LDPC Codes by Connecting Multiple Code Chains

Pablo M. Olmos, David G. M. Mitchell+2 more|arXiv (Cornell University)|Feb 28, 2014

Error Correcting Code Techniques28 references3 citations

TL;DR

This paper proposes continuous chain (CC) transmission, a novel scheme that connects multiple long spatially coupled LDPC (SC-LDPC) code chains to improve finite-length performance over binary-input AWGN and binary erasure channels. By reorganizing codeword transmission order and leveraging structural protection in connected chains, CC transmission achieves significant block error rate gains with negligible increase in encoding/decoding complexity or delay, outperforming single-chain SC-LDPC codes in both short- and long-chain regimes.

ABSTRACT

In this paper, we analyze the finite-length performance of codes on graphs constructed by connecting spatially coupled low-density parity-check (SC-LDPC) code chains. Successive (peeling) decoding is considered for the binary erasure channel (BEC). The evolution of the undecoded portion of the bipartite graph remaining after each iteration is analyzed as a dynamical system. When connecting short SC-LDPC chains, we show that, in addition to superior iterative decoding thresholds, connected chain ensembles have better finite-length performance than single chain ensembles of the same rate and length. In addition, we present a novel encoding/transmission scheme to improve the performance of a system using long SC-LDPC chains, where, instead of transmitting codewords corresponding to a single SC-LDPC chain independently, we connect consecutive chains in a multi-layer format to form a connected chain ensemble. We refer to such a transmission scheme to as continuous chain (CC) transmission of SC-LDPC codes. We show that CC transmission can be implemented with no significant increase in encoding/decoding complexity or decoding delay with respect a system using a single SC-LDPC code chain for encoding.

Motivation & Objective

• To improve the finite-length performance of long spatially coupled LDPC (SC-LDPC) codes, which suffer from suboptimal iterative decoding thresholds.
• To address the performance gap between asymptotic thresholds and practical finite-length behavior in SC-LDPC codes.
• To design a transmission scheme that enhances decoding performance without increasing encoding/decoding complexity or delay.
• To analyze how structural connectivity between SC-LDPC chains affects decoding dynamics and error rate scaling.
• To extend finite-length performance analysis beyond the binary erasure channel (BEC) to include the binary-input AWGN (BIAWGN) channel.

Proposed method

• Proposes continuous chain (CC) transmission, where consecutive long SC-LDPC code chains are connected into a single multi-layered structure for transmission.
• Models the decoding process using peeling decoding on the Tanner graph, tracking the evolution of degree-one check nodes as a dynamical system.
• Analyzes the mean and variance of the number of undecoded degree-one check nodes to estimate block error probability and identify critical points (local minima) that limit performance.
• Compares performance of single-chain ensembles (e.g., C(3,6,L)) with connected-chain ensembles (e.g., S(3,6,L,N)) across different code lengths and connection layers.
• Employs a structured ensemble model based on random graphs rather than protograph-based designs to isolate the impact of chain connectivity on performance.
• Validates results through simulations on both the binary erasure channel (BEC) and binary-input AWGN (BIAWGN) channel, comparing block error rates across different chain layers and configurations.

Experimental results

Research questions

• RQ1 Can connecting multiple SC-LDPC code chains improve finite-length decoding performance compared to single-chain configurations?
• RQ2 How does the number of connected chains and their layering affect the decoding threshold and error rate scaling in SC-LDPC codes?
• RQ3 Does continuous chain (CC) transmission provide performance gains over conventional single-chain transmission without increasing complexity or delay?
• RQ4 To what extent do structural features such as end-node protection and local graph irregularity influence finite-length performance in connected SC-LDPC ensembles?
• RQ5 Can the performance gains observed in the BEC be extended to more practical channels like the BIAWGN?

Key findings

• Connected-chain ensembles such as S(3,6,L=50,N=3) achieve significantly better finite-length performance than single-chain ensembles of the same rate and length, particularly in the waterfall region.
• For the BEC, chain 1 in a multi-layered CC transmission setup (e.g., S(3,6,50,N=3)) shows substantial performance improvement, while chains in later layers maintain performance comparable to single-chain codes.
• On the BIAWGN channel, the first chain in a CC transmission system (e.g., S(3,6,50,N=2)) achieves a block error rate that is markedly lower than that of a single SC-LDPC chain of the same length and rate.
• The performance of the first chain in a CC transmission setup is enhanced due to stronger local protection and proximity of high-reliability regions, which reduce the likelihood of decoding failure at critical points.
• CC transmission achieves performance gains with no significant increase in encoding/decoding complexity or decoding delay, making it practical for real-time systems.
• The loop ensemble formed by connecting two chains of length L behaves like a single shorter chain in terms of decoding thresholds and finite-length scaling, demonstrating that structural connectivity enhances robustness.