Reliability-driven framework for multidimensional reconciliation in CV-QKD

In multidimensional reconciliation (MR) of continuous-variable quantum key distribution (CV-QKD) systems, the performance of conventional schemes is often limited by unreliable Gaussian-modulated coherent states (GMCSs). In this paper, we propose a reliability-driven preprocessing framework for MR. The proposed framework introduces a multi-step sample interleaver that redistributes GMCSs to increase intra-vector reliability while reducing the reliability imbalance across vectors. In addition, a bit-allocation strategy is designed to align the reliability of the soft information generated by MR with the reliability requirements of the employed forward error correction (FEC) codes. To further reduce implementation overhead, simplified log-likelihood ratio (LLR) estimation methods are also investigated. Simulation results demonstrate that the proposed scheme significantly improves the frame error rate (FER) performance compared with conventional schemes, while introducing only negligible additional latency, thereby extending the achievable transmission distance of CV-QKD systems.