Continuous variable quantum key distribution finite key analysis of composable security against coherent attacks

Institute for Theoretical
Physics
Quantum Information
Group

Continuous Variable Quantum Key
Distribution:
Finite-Key Analysis of Composable
Security against Coherent Attacks
Fabian Furrer
Leibniz Universität Hannover
PRL 109, 100502
Joint work with
T. Franz, R. F. Werner (Leibniz Universität Hannover)
M. Berta, A. Leverrier, V.B. Scholz, (ETH Zurich)
M. Tomamichel (CQT Singapore)

Qcrypt 2012, Singapore, 14.09.2012

Security of a QKD Protocol

Security Analysis
Constraints: -Key length
 Information theoretic -Certification
 Experimental of security

?

Experimental Secure
implementation Key

Minimizing the assumption and maximizing the key length!

FF, Continuous Variable QKD: Finite-Key Analysis against Coherent Attacks, Singapore, 14.09.2012 Seite 2

Security of a QKD Protocol

Constraints:
 Information theoretic
 Asymptotic key rate vs. finite uses of QM channel (finite-key
effects)
 Notion of security: composable?
 Limitation on attacks: collective (tensor product) or coherent
(general)?
 ...
 Experimental / Implementation
 Model of the measurement devices
 Model of the quantum source
 ...


Contribution: Security analysis for continuous
variable (CV) protocol based on the distribution of two-mode
squeezed states (EPR states) measured via homodyne
detection.


Contribution: Security analysis for continuous
variable (CV) protocol based on the distribution of two-mode
squeezed states (EPR states) measured via homodyne
detection.

What’s New: Computation of key length secure against
coherent attacks for achievable experimental parameters.

Security proof based on Uncertainty relation
(c.f. Tomamichel et al., Nat. Comm. 3, 634 ,2012)


Discrete Variables vs. Continuous Variables
Implementation

-Encoding in finite-dimensional -Encoding in infinite-dimensional
systems (e.g., polarization of photon) systems (bosonic modes) [1]
-Gaussian States
-Quadratures of EM-field:
Homodyne or Heterodyne
detection

Advantage:
- Compatible with standard
telecom technology
- high repetition rates for homodyne
- efficient state preparation

[1] Weedbrook et al., Reviews of Modern Physics 84, 621 (2012)

Security Analysis for CV QKD Protocols

Challenge: infinite dimensions

Finite-Key Analysis:
- Leverrier et al, Phys. Rev. A 81, 062343 (2010)
- Berta, FF, Scholz, arXiv:1107.5460 (2011)

Lifting proofs from collective to coherent (general) attacks:
 Exponential de Finetti [Renner & Cirac, PRL 102, 110504 (2009)]
Problem: Bad bounds, feasible only in the asymptotic limit
Post-selection technique,
Recent: Leverrier et al., arXiv:1208.4920 (Talk on Monday)


Security Analysis for CV QKD Protocols

Challenge: infinite dimensions

Finite-Key Analysis:
- Leverrier et al, Phys. Rev. A 81, 062343 (2010)
- Berta, FF, Scholz, arXiv:1107.5460 (2011)

Lifting proofs from collective to coherent (general) attacks:
 Exponential de Finetti [Renner & Cirac, PRL 102, 110504 (2009)]
Problem: Bad bounds, feasible only in the asymptotic limit
Post-selection technique,
Recent: Leverrier et al., arXiv:1208.4920 (Talk on Monday)
Uncertainty Relation (direct) : This Talk!
Advantage:
 one-sided device independent
 no tomography
 no additional measurements

Outline

1. Security Definition and Finite-key length formula

2. Experimental Set Up and Protocol

3. Finite-Key Rates

4. (Security Analysis)


General QKD Protocol

Eve Bob Part 1:
Alice 1) Distribution of quantum
Distribution & Measurements
state
2) Measurements
3) Parameter estimation
4) Output: Raw keys
or abort


General QKD Protocol

Eve Bob Part 1:
Alice 1) Distribution of quantum
Distribution & Measurements
state
2) Measurements
3) Parameter estimation
4) Output: Raw keys
classical post processing or abort
Part 2:
1) Error correction
2) Privacy amplification
Output: Key


Security Definitions (trace distance)
A protocol which outputs the state

is secure if it is:
 correct :

 secret:

where is the uniform distribution over all keys.

Composable Secure*

* R. Renner, PhD Thesis (ETH 2005)


Classical Post Processing
1) Error Correction:
Alice and Bob broadcast bits to match their strings.
2) Privacy amplification via two-universal hash functions:
... apply random hash function from two-universal family onto bits

Key length



Key length

Secure key of length:

Smooth min-entropy
R. Renner, PhD Thesis (2005), M. Tomamichel et al. IEEE Trans. Inf. Theory, 57 (8) (2011),
M. Berta, FF, V.B. Scholz, arXiv1107.5460 (infinite-dimensional side-information)



Key length

Secure key of length:

Smooth min-entropy
R. Renner, PhD Thesis (2005), M. Tomamichel et al. IEEE Trans. Inf. Th, 57 (8) (2011),
M. Berta, FF, V.B. Scholz, arXiv1107.5460 (infinite-dimensional side-information)

Use parameter estimation to bound min-entropy!

Experimental Set Up
Source:
two-mode squeezed
state (EPR state)

Measurements:
homodyne detection,
randomly either
amplitude or phase
(synchronized via LO)

Entanglement based!

Cerf, N. J., M. Levy, and G. van Assche, 2001, Phys. Rev. A 63, 052311

Measurements
Source:
Correlated outcomes if both measure
two-mode squeezed
amplitude or phase:
state

Measurements:
homodyne detection,
randomly either
amplitude or phase
(synchronized via LO)

Entanglement based!


Measurements
Binning of the Outcome Range:

 Spacing parameter:
 Cutoff parameter:

Outcome Range:


Protocol
1. Performing 2N measurements

2. Sifting: approx. N data points left

3. Parameter estimation:
pick random sample of k data points and check correlation:
Hamming distance:

4. Classical post-processing on remaining strings :


Protocol


Hamming distance:


A secret key of length

can be extracted


Protocol


Hamming distance:


A secret key of length Statistical correction

can be extracted Monotonic function
Complementarity of amplitude and phase
measurement: depending on spacing parameter

Finite-Key Length
The key is ...

 composable secure

 provides security against coherent attacks

Experimental constraints:

 Alice’s measurements are modeled by projections onto spectrum of
quadrature operator for amplitude and phase (parameter: )

 subsequent measurements commute

 trusted source in Alice’s lab of Gaussian states (can be relaxed)

 No assumptions about Bob’s measurements: one-sided device
independent


Finite-Key Rates
Key Rate depending on symmetric losses for two-mode squeezed state
 input squeezing/antisqueezing 11dB/16dB *
 error correction efficiency of 95%
* T. Eberle et al., arXiv:1110.3977
 excess noise of 1% *
 additional symmetric losses of ...

0%

4%
6%

Plot: FF et al., PRL 109, 100502 (2012)


Key Rate versus Losses
Key rate versus losses for N=10^9 sifted signal:

Asymptotic optimal rate with
perfect error correction

Coherent attacks Gaussian collective attacks

symmetric losses

Plot: FF et al., PRL 109, 100502 (2012)


Security Analysis Based on Uncertainty Relation
Extractable key length:

Goal: bound for

Key ingredient: Uncertainty relation with side-information*

* Tomamichel & Renner, Phys. Rev. Lett. 106, 110506 (2011)


Entropic Uncertainty Relation with Side
Information
: Amplitude
Eve
: Phase
Alice Bob


Information
: Amplitude
Eve
: Phase
Alice Bob

uncertainty Eve has about uncertainty of Bob about
outcome of Alice outcome of Alice

complementary of the
M. Tomamichel , PhD Thesis, measurements
M. Berta, FF, V.B. Scholz, arXiv1107.5460 (2011)


Information
: Amplitude
Eve
: Phase
Alice Bob

Data processing inequality

M. Tomamichel et al., Phys. Rev. Lett. 106,110506 (2011),
M. Tomamichel , PhD Thesis (2012); M. Berta, FF, V.B. Scholz, arXiv1107.5460 (2011)


Information
: Amplitude
Eve
: Phase
Alice Bob

Data processing inequality
Correlation betw. Alice & Bob
M. Tomamichel et al., Phys. Rev. Lett. 106,110506 (2011),
M. Tomamichel , PhD Thesis (2012); M. Berta, FF, V.B. Scholz, arXiv1107.5460 (2011)


Correlation between Alice & Bob
Correlation between Alice and Bob can be bounded in terms of the
Hamming distance of a random sample

via

Combining with Uncertainty Relation:


Conclusion
Advantage:
 one-sided device independent (e.g. local oscillator included)
 direct approach (no additional measurements compared to post-
selection approach)
 no state tomography
 robust under small deviations of experimental parameters
Problems:
 very sensitive to noise
 asymptotically not optimal: Uncertainty relation not tight for the
Gaussian states used in the protocol

Implementation in Leibniz University in Hannover:
Crypto on Campus: T. Eberle, V. Händchen, J. Duhme, T. Franz, R. F. Werner, and R.
Schnabel


Thank you for your attention!


Continuous variable quantum key distribution finite key analysis of composable security against coherent attacks

More Related Content

What's hot (7)

Similar to Continuous variable quantum key distribution finite key analysis of composable security against coherent attacks (20)

More from wtyru1989 (20)

Recently uploaded (20)

Continuous variable quantum key distribution finite key analysis of composable security against coherent attacks