arXiv:2609.20596
Experimental demonstration of finite-size general security via discrete-modulated CVQKD with real time postprocessing
Sven Bodenstedt, Carlos Pascual-García, Nil Canta i Pujol, Martí Sales-Moragues, Mariana Navarro, Pau Gómez Kabelka, Sebastián Etcheverry, Saeed Ghasemi
Abstract
Continuous-variable quantum key distribution (CVQKD) is compatible with telecommunication infrastructure, but implementing composable security with experimentally practical resources has remained challenging, particularly for discrete-modulated (DM) protocols. We report the first experimental demonstration of a DM CVQKD system that generates composable secret keys against general attacks with finite-size block lengths on the order of $\sim 10^{6}$ rounds via a quadrature phase shift keying (QPSK) system. Our implementation follows a variable-length, general security framework enabled by modern entropy accumulation techniques and conic optimization, whose experimental pipeline allows real-time operation on near-commercial hardware.
AI Summary
The paper addresses whether discrete-modulated coherent-state CVQKD can generate finite-size, composably secure keys against general attacks with online classical postprocessing. It implements QPSK modulation and heterodyne detection, combining a variable-length security framework based on marginal-constrained entropy accumulation with conic optimization and a trusted-detector-noise model. Computationally expensive entropy bounds and trade-off functions are precomputed in a configuration database. At runtime, observed categorical frequencies and reconciliation leakage determine the key length through inexpensive inner products; the largest candidate among entries satisfying calibration-based validity conditions is selected. Database construction took roughly two weeks on a consumer CPU, replacing per-session optimization reported to take about 45 minutes. The experiment reports positive keys with blocks of order $10^6$ rounds up to 10 km equivalent loss. With $10^8$ rounds, full-protocol rates range from approximately 10 kbit/s at 5 km to 100 bit/s at 40 km, with total secrecy-plus-correctness parameter at most $10^{-10}$. Distances are emulated using attenuation. For a representative $10^8$-round session at 5 km, measurement takes 157.3 s and LDPC reconciliation 379.4 s; initialization, key-length decision and privacy amplification together take under 3 s. Security remains conditional on a photon-number cutoff assumption, set to $N_c=10$, calibrated trusted detector parameters, and perfect symmetric authentication using a pre-shared key. Removing the cutoff and accounting for authentication-key consumption are left to future work. Reconciliation dominates runtime; public communication occurs only after quantum transmission, without pipelining.
Central claim may not hold: The abstract's general-attack security claim is conditional on an assumed photon-number cutoff N_c = 10 on Bob's received state, without an implemented test or a correction accounting for violations. The implemented maximization of key length over database rules chosen from current-session statistics lacks a simultaneous secrecy bound or security-budget allocation, so Theorem 1 does not establish the stated secrecy parameter for the implemented selection rule.
Written by gpt-6.1-sol from the paper's full text. Automated: it can be wrong.
- Comments
- 21 pages, 5 figures