High-speed reconciliation with convergence-based early termination for CV-QKD using SC-LDPC codes

The information reconciliation (IR) stage is a critical component of post-processing in continuous-variable quantum key distribution (CV-QKD), which is essential to ensure high secret key rates and long transmission distances. Currently, improving reconciliation efficiency is a key research focus in the CV-QKD field. Additionally, the low-density parity-check (LDPC) decoding speed in IR is a post-processing bottleneck in high-speed CV-QKD. In this paper, we propose a construction scheme for spatially coupled (SC) LDPC codes in CV-QKD systems and maintain a constant code rate when extending the code length as required. Furthermore, we investigate the relationship between the syndrome variation pattern (SVP) and the frame error rate. Adaptively adjusting the number of decoding iterations according to the distinct convergence behaviors of the SVP, the convergence-based early termination (C-ET) scheme is proposed, which can reduce invalid iterations. The simulation results show that the proposed construction scheme for SC-LDPC codes can improve the reconciliation efficiency to 97% and achieve a flexible code length. The C-ET scheme achieves a 292% improvement in information throughput compared to the original ET scheme when the reconciliation efficiency reaches 97%. Moreover, the C-ET scheme also achieves a reduction in decoding latency compared to the existing syndrome-based early termination scheme. These advances provide a substantial step toward improving reconciliation efficiency and breaking the post-processing bottleneck in CV-QKD systems.