Source-linked AI summary

Field test of a practical secure communication network with decoy-state quantum cryptography

Teng-Yun Chen, Hao Liang, Yang Liu, Wen-Qi Cai, Lei Ju, Wei-Yue Liu, Jian Wang, Hao Yin, Kai Chen, Zeng-Bing Chen, Cheng-Zhi Peng, Jian-Wei Pan

arXiv:0810.1264v4quant-ph

TL;DR

Weak coherent pulses expose practical QKD to photon-number-splitting attacks, while earlier network implementations lacked decoy states or had short secure spans. The authors operate a decoy-state QKD system with a trusted relay linking three nodes over two approximately 20 km telecom-fiber spans. The three-node network demonstrated real-time one-time-pad voice calls between any two users and encrypted broadcasts from one node to the other two.

  • Problem

    Weak coherent pulses expose practical QKD to photon-number-splitting attacks, while earlier network implementations lacked decoy states or had short secure spans.

    Source: §1 · pp. 3–4
  • Method

    The authors operate a decoy-state QKD system with a trusted relay linking three nodes over two approximately 20 km telecom-fiber spans.

    Source: §1 · p. 4, §2 · p. 6, §3 · p. 8
  • Results

    The three-node network demonstrated real-time one-time-pad voice calls between any two users and encrypted broadcasts from one node to the other two.

    Source: §1 · pp. 4–5, §4 · p. 10
  • Takeaways & Limitations

    The chained topology forwards secret keys hop by hop through a trusted middle node for one-time-pad information transmission.

    Source: §4 · p. 10

    Simultaneous two-way communications for N nodes require (N − 1) kbps quantum keys.

    Source: §3 · p. 10

Abstract

from arXiv · show

We present a secure network communication system that operated with decoy-state quantum cryptography in a real-world application scenario. The full key exchange and application protocols were performed in real time among three nodes, in which two adjacent nodes were connected by approximate 20 km of commercial telecom optical fiber. The generated quantum keys were immediately employed and demonstrated for communication applications, including unbreakable real-time voice telephone between any two of the three communication nodes, or a broadcast from one node to the other two nodes by using one-time pad encryption.

1. Introduction

Practical QKD must address security limits from weak coherent pulses, while deployment has largely remained point-to-point. This work presents a decoy-state, three-node trusted-relay network for real-time one-time-pad voice and broadcast applications.

1. Introduction

  • Decoy-state methods improve practical QKD key rates and distances, addressing weak-coherent-pulse security threats and limits of earlier security analyses.Weak coherent pulses can contain multiple photons, leaving systems vulnerable to photon-number-splitting attacks.Source: §1 · p. 3
  • Earlier networks showed multiuser architectures, but DARPA, NIST, and Beijing implementations lacked decoy states, while DARPA and NIST secure spans were under 10 km and 1 km, respectively.Source: §1 · p. 4
  • The proposed three-user decoy-state network uses a trusted USTC relay and two approximately 20 km telecom-fiber links for on-demand key creation.Source: §1 · p. 4
  • The network demonstrated real-time one-time-pad voice between any two users and encrypted broadcasts from one node to the other two.Source: §1 · pp. 4–5
  • The authors emphasize a complete, compact, low-cost system using true random number generators and small-volume InGaAs detectors for practical deployment.The text contrasts these detectors with upconversion detectors’ high background counts and superconducting detectors’ cryogenic-cooling requirement.Source: §1 · pp. 5–6

2. Experimental setup

The experimental system uses commercial telecom fiber to connect two decoy-state QKD links through USTC. Integrated control and application hardware supports quantum-key generation and real-time audio communication.

2. Experimental setup

  • Two links connect Binhu–USTC and USTC–Xinglin over the running China Netcom telecom fiber, with integrated transmitter or receiver functions at each node.The laser source runs at 4 MHz, matching the rate of the commercially available true random number generators.Source: §2 · p. 6
  • The transmitter’s time-division Mach–Zehnder interferometer generates BB84 states; chosen attenuation sets signal and decoy intensities for each link.Polarizing and ordinary beam splitters route photons between Alice’s and Bob’s arms to avoid a 3 dB loss for useful photons.Source: §2 · p. 6
  • Synchronization information is transmitted through a WDM apparatus over 20 km of dark fiber; at the receiver, a synchronized clock controls phase-modulator measurement-basis choice and InGaAs detector gates.Synchronization electronics, detection logic, and signal acquisition are integrated on an FPGA board running at 4 MHz; detector gates are 2 ns wide.Source: §2 · p. 7
  • Active polarization and phase feedback compensate fiber polarization changes and interferometer phase fluctuations during transmission.A reference laser pulse at 1550.12 nm is inserted between signal pulses to stabilize the interferometer arm lengths.Source: §2 · p. 7
  • True random number generators drive signal/decoy attenuation and the transmitter’s state and receiver’s measurement-basis choices.Source: §2 · p. 7

3. Secure key generation and applications

The network used decoy-state QKD with finite-key post-processing to generate secure keys, then demonstrated real-time voice calls and one-time-pad broadcasts. The two links produced secure keys at more than 1.5 kbps each, supporting these applications with a stated feedback limitation for broadcasts.

Secure key generation and applications

  • More than 1.5 kbps final secure key rate was achieved on both links after real-time error correction and privacy amplification.Measurements over 30 minutes recorded sifted-key rates above 10.5 kbps and 9.0 kbps, with QBERs of about 1.6% and 1.4%, respectively.Source: §3 · p. 8
  • The decoy-state key-rate method estimates a lower bound on single-photon gain and an upper bound on its error rate to calculate secure key generation.The implementation considered finite key length and statistical fluctuations; privacy amplification used universal2 functions represented by Toeplitz matrices.Source: §3 · pp. 8–9
  • 1.7 kbps final key rate was obtained for a typical 120-second run on the USTC-Xinglin link.The estimate accounted for statistical fluctuations within 10 standard deviations and reported a confidence interval of about 1−1.5×10^-23.Source: Table 2 · p. 9, §3 · p. 9
  • About 120 Mbits of final key passed the reported NIST 800-22 and Diehard randomness tests.NIST tests had a minimum pass rate of 95%; reported Diehard p-values ranged from 0.009 to 0.989.Source: §3 · p. 9
  • Real-time two-way voice calls used one-time-pad encryption with 0.6 kbps audio compression and more than 1.2 kbps of available keys.The authors reported clear audio after one hour of continuous operation and no problems during a half-month test over a 20 km telecom fiber.Source: §3 · p. 9
  • One-time-pad broadcasts reached the other two nodes, but simultaneous two-way communication among N nodes would require 1.2 × (N − 1) kbps of quantum keys.The authors state that feedback from both receiving nodes was not then possible because of limited key-generation rates.Source: §3 · pp. 9–10

4. Conclusion

The experiment demonstrates an operational three-user network for real-time voice calls or one-to-many broadcasts using one-time-pad encryption. Its chained QKD links forward keys hop by hop through a trusted relay, and the setup can be expanded to more nodes.

4. Conclusion

  • The operational network supports real-time voice telephone between any two of three users or one-to-two broadcasts using one-time-pad encryption.The authors also state that unconditional authentication and encryption for information transmission using one-time pad become possible.Source: §4 · p. 10
  • Chained QKD links forward secret keys hop by hop, while the middle node acts as a trusted relay and increases key generation compared with direct connections.The authors describe direct connection key generation as decreasing exponentially with distance.Source: §4 · p. 10
  • The setup can be expanded to many nodes, with key demand increasing slowly; future work targets higher key rates using better detectors and faster true random number generators.For video conferencing when one-time-pad key rates are insufficient, the authors expect AES with frequent key refresh could maintain a desired security level.Source: §4 · p. 10
Loading 0810.1264v4…