Why one quantum-security experiment reports two very different key rates
A new preprint compares two treatments of the same fiber experiment. Its smaller figure includes finite-data penalties, but neither establishes a deployment-ready secure service.
Reading tools for text size, copying, printing and your reading list

A November 12 preprint from researchers in Brazil and the Netherlands compares two treatments of one quantum-security experiment. Accounting for limited measurement data reduced the estimated key rate from 1.421 to 0.364 megabits per second, a drop of about 74%. Both rates divide the estimated key length by the time spent transmitting signals and exclude offline processing time.[1][2]
Those are model-dependent results from offline processing of a 12-kilometer fiber experiment, not live encrypted-message throughput. The authors explicitly say their analysis is not a complete composable finite-key security proof, the kind needed to establish how security guarantees hold when a protocol is used within a larger system.[1]
The study tests continuous-variable quantum key distribution, or CV-QKD, which uses measurements of light to help two parties establish encryption keys. Its one-dimensional protocol places information in one of two optical-field components, called quadratures, while measurements of the other constrain the security calculation.[1]
Two estimates from one 12 km fibre link
Both estimates use signal-transmission time. The smaller estimate accounts for uncertainty in the available measurements and a privacy-amplification penalty. Offline processing time is excluded from both.
See the authors’ 12 km setup diagram · Try sampling a qubit in the readout lab
The larger rate uses an asymptotic treatment, effectively setting aside the statistical uncertainty associated with a limited sample. The smaller figure includes finite-sample parameter-estimation bounds and an additional privacy-amplification correction. Privacy amplification compresses the processed bits to limit information potentially available to an eavesdropper under the model.[1]
The experimental contribution is integrating independent lasers with practical error reconciliation, parameter estimation and privacy amplification. It is not the first single-quadrature QKD experiment. The analysis assumes a calibrated, trusted receiver and Gaussian collective attacks; the demonstration also used computer-generated pseudorandom modulation rather than a quantum random-number source.[1]
The protocol could eventually simplify the transmitter. This setup still needed an in-phase-and-quadrature modulator to accommodate its lasers and detector.[1]
When would a user actually have the key?
The finite-data figure corresponds to about 364,000 usable key bits per second of signal transmission. The authors divide the output bits by the time spent transmitting quantum signals, including transmissions from which no usable key survived. To measure delivery to a user, divide usable output bits by the full elapsed time until the key is ready. The paper processes its data offline, so its rates cannot supply that wall-clock throughput.[1]