Securing Practical Quantum Cryptography with Optical Power Limiters - - PowerPoint PPT Presentation

▶
securing practical quantum cryptography with optical
SMART_READER_LITE
LIVE PREVIEW

Securing Practical Quantum Cryptography with Optical Power Limiters - - PowerPoint PPT Presentation

Dec 02, 2023 628 likes •859 views

Securing Practical Quantum Cryptography with Optical Power Limiters Gong Zhang 1,* , Ignatius William Primaatmaja 2 , Jing Yan Haw 1 , Xiao Gong 1 , Chao Wang 1, , and Charles C.-W. Lim 1,2, * zhanggong@nus.edu.sg wang.chao@nus.edu.sg

slide-1
SLIDE 1

Gong Zhang1,*, Ignatius William Primaatmaja2, Jing Yan Haw1, Xiao Gong1, Chao Wang1,†, and Charles C.-W. Lim1,2,‡

*zhanggong@nus.edu.sg †wang.chao@nus.edu.sg ‡charles.lim@nus.edu.sg

1Department of Electrical & Computer Engineering,

National University of Singapore, Singapore

2Centre for Quantum Technologies,

National University of Singapore, Singapore

Securing Practical Quantum Cryptography with Optical Power Limiters

slide-2
SLIDE 2

Outline

1

❑ Background

❑ Importance of power limiter in quantum cryptography ❑ Introduction of thermo-optic defocusing

❑ Experimental and simulation results ❑ Possible attack consideration ❑ Application in plug-and-play MDI-QKD ❑ Conclusion

slide-3
SLIDE 3

Detector-blinding attack Makarov 2009, Lydersen 2010 Receiver laser damage attack Bugge 2014, Makarov 2016 Time-shift attack Qi 2007, Zhao 2008 Wavelength attack Huang 2013, Li 2011 Back-flash attack Kurtsiefer 2001 Channel calibration Jain 2011 Detector deadtime Weier 2011 Spatial efficiency mismatch Rau 2015, Sajeed 2015 Trojan-horse attack Gisin 2006, Jain 2014 Intensity information Jiang 2012 Modulation pattern effect Yoshino 2016 Source laser damage attack Huang 2020 Phase-remapping attack Fung 2007, Xu 2010 Phase information Sun 2012, 2015, Tang 2013

Hacking Practical QKD

2

Lo, H. K., et al. (2014). Nature Photonics, 8(8), 595-604. Scarani, V., et al. (2009). Reviews of modern physics, 81(3), 1301.

Solution Measurement-device-independent MDI-QKD Target: Receiver Target: Source

slide-4
SLIDE 4

Trojan-Horse Attack

3

[1] Gisin, N., et al. (2006). Physical Review A, 73(2), 022320. [2] Sajeed, S., et al. (2015). Physical Review A, 91(3), 032326.

Alice

Quantum Channel

Laser Receiver Bob Encoding Devices Eve

Trojan Horse Photon 𝒘

Current countermeasures

  • Phase randomize (Reduce 𝐽𝑓𝑤𝑓1)
  • Watchdog detector (Can be bypassed2)
  • Passive components such as isolators

(Limited degree-of-freedom, one-way application only, high isolation) Basic idea is to limit the amount

  • f unauthorized input power.

Jain, N., et al. (2014). New Journal of Physics, 16(12), 123030.

slide-5
SLIDE 5
  • Bound on the mean energy is one way to

provide a practical Semi-Device- Independent (Semi-DI) framework.

  • Use energy bound to bound the overlap

between the prepared states.

  • Energy bound could lead to certifiable

quantum randomness.

Semi-DI with Energy Bound

4

Avesani, M., et al. (2020). arXiv:2004.08344v1. Van Himbeeck, T., et al. (2017). Quantum, 1, 33.

Alice Laser Receiver Bob Encoding Devices X P

State 0 State 1

Van Himbeeck, T., et al. (2019). arXiv:1905.09117. Rusca, D., et al. (2019). Physical Review A, 100(6), 062338..

Again, a power limiting device is important here!

slide-6
SLIDE 6

❑ Provides a reliable and characterizable power limiting threshold (in the order of a few photons to hundreds of photons). ❑ If the input energy exceeds the threshold, the device will stop the communication channel. ❑ Cost-effective, passive, and easily replaceable. ❑ Power limiting effects are independent of other degree of freedoms, e.g., frequency, polarization, etc.

Proposal: Quantum Optical Fuse / Power limiter

5

The device should ideally have the following properties: It is timely to develop such devices, for we now have a wide range of security proof methods with possible energy constraints features:

Lucamarini et al 2015, Tamaki et al 2016, Van Himbeeck et al 2019, Pereria et al 2019, Primaatmaja et al 2019, Navarrete et al 2020, just to name a few.

slide-7
SLIDE 7

Review of Optical Power Limiter

6

Seo, K., et al. (2003). Furukawa Review, 24(24), 17-22. Dini, D., et al. (2016). Chemical reviews, 116(22), 13043-13233. Yan, S., et al. (2014). Scientific reports, 4, 6676.

Fiber damage

  • 102 – 103 mW level

Nonlinear effect

  • 10 – 102 mW level

Filter based

  • Using thermo-optic effect or
  • ptical force to tune the filter

center wavelength

  • Narrow operation bandwidth,

limited extinction ratio

  • 10 – 102 mW level

Thermo-optical defocusing

Sang, X., et al. (2009). Journal of optoelectronics and advanced materials, 11(1), 15. Martincek, I., et al. (2011). IEEE Photonics Technology Letters, 24(4), 297-299.

Two-photon absorption

  • 10 – 103 mW level
slide-8
SLIDE 8

7

Patent filed: SG Non-Provisional Application No.10202006635S

Our Choice: Thermo-Optical Defocusing

Input Fiber Reflective Collimator Acrylic Prism Diaphragm Output Fiber

  • Negative thermo-optic coefficient of acrylic:

𝑒𝑜 𝑒𝑈 = −1.3 × 10−4 𝐿−1

  • Higher absorbed power diverges the input light more
  • A tunable diaphragm controls the received power
  • Robust and stable performance, compact and cost-effective design

Power Limiter Module Visible filter Reflective Collimator

z r

slide-9
SLIDE 9

8

Input Fiber Reflective Collimator Acrylic Prism Diaphragm Output Fiber Power Limiter Module Visible filter Reflective Collimator

Patent filed: SG Non-Provisional Application No.10202006635S

Our Choice: Thermo-Optical Defocusing

  • Negative thermo-optic coefficient of acrylic:

𝑒𝑜 𝑒𝑈 = −1.3 × 10−4 𝐿−1

  • Higher absorbed power diverges the input light more
  • A tunable diaphragm controls the received power
  • Robust and stable performance, compact and cost-effective design
slide-10
SLIDE 10

𝜖𝜄𝑠 𝜖𝑨 = 1 𝑜 𝜖𝑜 𝜖𝑈 𝜖𝑈 𝜖𝑠

Theoretical Modeling

9

Smith, D. (1969). IEEE Journal of Quantum Electronics, 5(12), 600-607. DeRosa, M. E., et al. (2003). Applied optics, 42(15), 2683-2688.

  • Angular divergence of a paraxial light ray

passing through a refractive index gradient 𝛽𝐽 = − 𝑙 𝑠 𝜖 𝜖𝑠 𝑠 𝜖𝑈 𝜖𝑠

  • Absorbed laser power I is balanced with the

heat transfer mechanism (Assume heat transfer in r-direction only) 𝐽 𝑠, 𝑨 = 𝐽 𝑠, 0 ∙ exp −𝛽𝑨 + 𝜖𝑜 𝜖𝑈 𝑄𝑓−𝑠2

𝑏2

𝑨 − 1 𝛽 1 − 𝑓−𝛽𝑨 𝜌𝑙𝑜𝑏2

  • Laser intensity at position (r, z)

Gaussian beam shape Absorption

5000 4000 3000 2000 1000 20 40 60 80 100 0.5 1 1.5 E-field (V/m) z r Unit: mm Pin = 7.9 mW 296.5 295.5 294.5 293.5 1 2 3 4 5 Temperature (K) Pin = 7.9 mW z Unit: mm r 20 40 60 80 100

  • COMSOL simulation
slide-11
SLIDE 11

Input-Output Power Relationship

10

Lucamarini, M., et al. (2015). Physical Review X, 5(3), 031030.

0.0 0.5 1.0 1.5 2.0

  • 20
  • 10

10 20 30 Power (dBm) Diaphragm width (mm)

Maximum Output Power Input Power

5 10 15 20 25 30

  • 10

10 20 Power (dBm) Length (cm)

Maximum Output Power Input Power

Prism length Diaphragm width

  • 10

10 20 30

  • 30
  • 20
  • 10

10 Output Power (dBm) Input Power (dBm)

20 cm 10 cm 5 cm 2.5 cm 1 cm

  • 10

10 20 30

  • 20
  • 10

10 Output Power (dBm) Input Power (dBm)

1.6 mm 0.8 mm 0.52 mm 0.4 mm 0.2 mm 41 dBm 12.8 W

Fiber damage threshold

slide-12
SLIDE 12

Shorter pulse Higher output power ?

Response Time Consideration

11

0.0 0.1 0.2 0.3 0.4 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 Output Power (mW) Time (s)

Pin = 64.6 mW Pin = 32.4 mW Pin = 16.2 mW Pin = 6.5 mW Pin = 2.0 mW

0.0 0.1 0.2 0.3 0.4 0.2 0.5 2 5 1 10 Output Power (mW) Time (s)

Pin = 196.3 mW Pin = 80.4 mW Pin = 38.0 mW Pin = 20.1 mW Pin = 7.2 mW

Experimental Results Simulation Results

slide-13
SLIDE 13
  • Assume 20 mW average input

power (Based on prior experiment)

  • Pulsed input experiences greater

power-limiting effect comparing to the continuous-wave cases

Pulsed Response Simulation

12

2 4 6 8 10 293 294 295 296 297 298 Maximum Temperature (K) Time (s)

Pin = 196.3 mW Pin = 125.7 mW Pin = 80.4 mW Pin = 38.0 mW Pin = 20.1 mW Pin = 7.2 mW

0.01 0.1 1 0.25 0.30 0.35 0.40 0.45 0.50 0.55 Output Average Power (mW) Duty Cycle Time Power Input pulses with different duty cycle

slide-14
SLIDE 14

0.4 0.8 1.2 1.6 2.0

  • 200
  • 150
  • 100
  • 50

Power (dB/cm) Wavelength (μm)

Pin Pout No filter

Wavelength Dependence

13

Zhang, Z., et al. (2006). Polymer, 47(14), 4893-4896. Beadie, G., et al. (2015). Applied optics, 54(31), F139-F143.

Thermo-optic coefficient 𝑈𝑃𝐷 = 𝑒𝑜 𝑒𝑈 = 𝑜2 − 1 𝑜2 + 2 6𝑜 (Φ − 𝛾)

  • Electronic polarizability Φ > 0 typically
  • Volumetric expansion 𝛾 is dominant in

polymer Wavelength (nm) dn/dT (x104 /K) 472.9

  • 1.37

780.4

  • 1.37

1055.7

  • 1.30

1308.9

  • 1.33

1550

  • 1.3

Material absorption

0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 2.2 5 10 15

Loss (dB/cm) Wavelength (μm)

1550 nm 1310 nm Minimum loss 0.15 dB/cm

  • Consider fiber damage threshold 12.8W
  • Silicon absorber limit visible light

0.4 0.8 1.2 1.6 2.0

  • 200
  • 150
  • 100
  • 50

Power (dB/cm) Wavelength (μm)

Pin Pout No filter Pout 100 μm Pout 1 mm Pout 10 mm

Zhang, X. 2020

Zhang, X., et al. (2020). Applied Optics, 59(8), 2337-2344. Lucamarini, M., et al. (2015). Physical Review X, 5(3), 031030.

slide-15
SLIDE 15

Laser Damage Attack

14

Berrie, P. G., et al (1980). Optics and Lasers in Engineering, 1(2), 107-129. M Taha, R. (2014). Diyala Journal of Engineering Sciences, 7(1), 30-39.

290 390 490 590 690 200 400 600 800 Max Temperature (K) Input Laser Power (mW)

Property Value Melting Point (K) 404 Boiling Point (K) 473 Evaporation rate (g/s) log w = 5.87- 6.77x103/T

  • Material could be melted and

evaporated under strong laser beam. As a result of the evaporation and assist gas pressure, the material is thrown out of the hole.

  • A reflection structure could be

implemented to permanently fuse the

  • ptical path.
slide-16
SLIDE 16

Laser Damage Attack

15

290 390 490 590 690 200 400 600 800 Max Temperature (K) Input Laser Power (mW)

  • Material could be melted and

evaporated under strong laser beam. As a result of the evaporation and assist gas pressure, the material is thrown out of the hole.

  • A reflection structure could be

implemented to permanently fuse the

  • ptical path.

Property Value Melting Point (K) 404 Boiling Point (K) 473 Evaporation rate (g/s) log w = 5.87- 6.77x103/T

Berrie, P. G., et al (1980). Optics and Lasers in Engineering, 1(2), 107-129. M Taha, R. (2014). Diyala Journal of Engineering Sciences, 7(1), 30-39.

slide-17
SLIDE 17
  • Plug-and-play phase-encoding

measurement-device-independent (MDI) QKD

  • Robust performance with simple

setup.

  • Common laser source for all users,

enables identical central wavelength and accurate clock synchronization.

  • Automatically compensate for any

birefringence effects and polarization-dependent losses in

  • ptical fibers.
  • The average Trojan photon number 𝜉

could provide Eve with information about the encoded phase

Application: Plug-and-Play MDI-QKD

16

Xu, F. (2015). Physical Review A, 92(1), 012333.

Alice Charlie 𝒘 Eve Bob Faraday Mirror Modulator Attenuator Power Limiter Detector Laser Bell state measurement

Lucamarini, M., et al. (2015). Physical Review X, 5(3), 031030. Tamaki, K., et al. (2016). New Journal of Physics, 18(6), 065008.

Patent filed: SG Non-Provisional Application No.10202006635S

slide-18
SLIDE 18

Secret Key Rate against THA

17

50 100 150 200 10

  • 7

10

  • 6

10

  • 5

10

  • 4

10

  • 3

10

  • 2

10

  • 1

Secret key rate per pulse Distance between Alice and Bob (km)

 =   = 

−

Consider a repetition rate of 1 GHz, the Trojan-horse photon power is about 1.28 x 10-10 mW

  • Assume average Trojan photon leakage 𝜉 from coherent state (CW and Pulse).
  • Proof technique taken here:

Primaatmaja, I. W., et al. (2019). Physical Review A, 99(6), 062332. Parameters Value Detector efficiency 70% Dark count rate 10−7 Misalignment error 2% Fiber loss 0.2 dB/km

slide-19
SLIDE 19

❑ To do: Security analysis of MDIQKD with untrusted light source ❑ To do: Measurement with visible wavelength and high-power laser

Conclusions and Outlooks

18

✓ Passive power limiter at mW level. Using additional attenuation for few- photon level limitation. ✓ If the input energy exceeds the threshold, the output power will be limited, and start decrease. ✓ Cost-effective, passive, and easily replaceable. ✓ Power limiting effects for both CW and pulsed light, wavelength and polarization independent. ❑ Provides a reliable and characterizable power limiting threshold (in the order of a few photons to hundreds for photons). ❑ If the input energy exceeds the threshold, the device will stop the communication channel. ❑ Cost-effective, passive, and easily replaceable. ❑ Power limiting effects are independent

  • f other degree of freedoms, e.g.,

frequency, polarization, etc. Ideal model Our scheme

slide-20
SLIDE 20

Acknowledgement

19

We are hiring postdoctoral researchers (theory/experiment)! Please contact us at charles.lim@nus.edu.sg for more information.

slide-21
SLIDE 21

Gong Zhang1,*, Ignatius William Primaatmaja2, Jing Yan Haw1, Xiao Gong1, Chao Wang1,†, and Charles C.-W. Lim1,2,‡

*zhanggong@nus.edu.sg †wang.chao@nus.edu.sg ‡charles.lim@nus.edu.sg

1Department of Electrical & Computer Engineering,

National University of Singapore, Singapore

2Centre for Quantum Technologies,

National University of Singapore, Singapore

Securing Practical Quantum Cryptography with Optical Power Limiters