Seminar, Purdue Quantum Center
May 17, 2019
Quantum-Enabled Sensing and Secure Communications
Franco N. C. Wong Massachusetts Institute of Technology
Funding: AFOSR, ONR, ARO Outline
. Quantum entanglement, loss, and noise • optical parametric amplifier: generator of photonic entanglement • adverse effects of loss/noise in realistic scenarios
. Concept of quantum illumination • can entanglement still be useful in a lossy and noisy environment?
. Quantum sensing of faint target bathed in noise • quantum illumination vs. best classical method
. Floodlight quantum key distribution • breaking “rules” and gaining speed • demonstration at Gbps secret-key rates Entanglement and applications Entanglement: correlation beyond classical physics allows Spontaneous parametric Ultra-sensitive metrology downconversion
Biphoton generation
10 ω ~10 s χ(2) ω p ωi Nature Phys. 10, 582 (2014) Very low efficiency Quantum computing
Idler Pump
Signal NIST Quantum decoherence
Environmental loss & noise destroy entanglement
Noise Noise
Noise
Noise
Major challenge for implementing quantum information applications Classical illumination: target detection
Noise Laser Probe Noise source Reference Noise Conventional detection Noise Quantum illumination (QI): target detection Multi-mode spontaneous parametric downconversion
Noise Entanglement Signal (probe) Noise source Idler (reference) Noise Joint quantum measurement Optical Noise parametric amplifier
QI references: S. Lloyd, Science 321, 1463 (2008) S.-H. Tan et al., Phys. Rev. Lett. 101, 253601 (2008) S. Guha and B.I. Erkmen, Phys. Rev. A 80, 052310 (2009) E.D. Lopaeva et al., Phys. Rev. Lett. 110, 153603 (2013) Z. Zhang et al., Phys. Rev. Lett. 111, 010501 (2013) Entanglement source Spontaneous parametric downconversion (SPDC)
Idler Signal Signal modes modes Pump Idler
Output state:
Low-brightness limit: Phase correlation:
Classical limit Optical parametric amplifier receiver Optical parametric amplification (OPA)
Signal Signal OPA receiver converts signal phase modulation Pump to idler intensity modulation
Idler Idler
S. Guha and B.I. Erkmen, Phys. Rev. A 80, 052310 (2009) Quantum sensing experiment
Target
BS: beam splitter DM: dichroic mirror OPA: optical parametric amplifier SPDC: spontaneous parametric downconverter
Z. Zhang et al., Phys. Rev. Lett. 114, 110506 (2015) Quantum sensing experiment
ASE: amplified spontaneous emission; BS: beam splitter; CWDM: coarse wavelength-division multiplexer; D: detector; DCF: dispersion-compensating fiber; DM: dichroic mirror; DSF: dispersion-shifted LEAF fiber; EDFA: erbium-doped fiber amplifier; OPA: optical parametric amplifier; PC: polarization controller; PM: phase modulator; Pol: polarizer; SMF: single-mode fiber; SPDC: spontaneous parametric downconverter; Z: zoom-lens systems
Z. Zhang et al., Phys. Rev. Lett. 114, 110506 (2015) Classical sensing experiment
Target
BS: beam splitter D: detector
Z. Zhang et al., Phys. Rev. Lett. 114, 110506 (2015) Signal-to-noise ratios (SNRs)
Quantum sensing Classical sensing
channel transmissivity probe brightness (photons/mode) in the limit of noise brightness (photons/mode) no experimental nonidealities Typical spectrum analyzer traces
Target Target absent present SNR comparison vs. transmissivity
Quantum results
Classical limit
Environmental loss: 14 dB; Noise background: 75 dB Quantum sensing has a 20% SNR enhancement
Z. Zhang et al., Phys. Rev. Lett. 114, 110506 (2015) Quantum illumination for sensing
Noise Entanglement Signal (probe) Noise source Idler (reference) Noise Joint quantum measurement Noise
. Quantum illumination protocol is loss and noise tolerant – enhancement comes from much stronger cross-phase correlation than classical states and use of many modes . Demonstrated entanglement-based quantum sensing – in the presence of entanglement-breaking loss and noise – outperform optimum classical sensing by 20% in SNR . QI practical applications in lossy and noisy environment – other quantum techniques? Secure communication: one-time pad encryption One-time pad encryption is impossible to crack Encoding by Bob ciphertext = plaintext ⊕ random binary key …1001101… = …1101000… ⊕ …0100101… Ciphertext is a completely random binary string Decoding by Alice ciphertext ⊕ same key = Bob’s plaintext …1001101… ⊕ …0100101… = …1101000… • Alice and Bob must share same “truly random” keys • keys must be as long as the plaintext message • use the secret keys only once . Quantum key distribution • key distribution with ability to detect eavesdropper Eve • reduce Eve’s knowledge of generated keys to practically zero Quantum key distribution (QKD): concept
. Entangled photons provide true randomness singlet state of polarization-entangled biphotons • V − |ψ 〉 = |H1V2 – V1H2〉/√2 HV basis A D
|ψ−〉 = |D A – A D 〉/√2 AD basis 1 2 1 2 H
. Quantum mechanics allows detection of eavesdropping (Eve) • state of an unknown qubit (polarization of Bob’s photon) cannot be fully determined • If Eve tries to determine the unknown qubit, it causes a disturbance in the biphoton state that is detectable by Alice and Bob • QKD protocols take advantage of such detectable disturbances to bound Eve’s ability to steal information • bottom line: laws of quantum physics can guarantee QKD security
No-cloning theorem makes QKD quantum! State-of-the-art QKD performance
. State-of-the-art QKD demonstrations • 1.26 Mbit/s for a 10-dB loss fiber channel using decoy-state BB84 Shields et al., Appl. Phys. Lett. 104, 021101 (2014) • 1.1 kbit/s for LEO satellite to ground using decoy-state BB84 Pan et al., Nature 549, 43 (2017)
Alice
Bob . BB84 typically operates under “photon-starved” conditions • send <1 photon/bit to prevent photon-number splitting attack • component loss + propagation losses ⇒ low flux at receiver • low secret-key rates (kbit/s to Mbit/s)
Mbit/s is far below acceptable, usable internet speed! Potential paths to much higher secret-key rates
. BB84 scaled up with massive multiplexing • 100 independent channels (with 10x rep rate) yield Gbps rates • complexity: timing, control, and post-processing for each channel • costly: 100x components (e.g., single-photon detectors)
. Solutions for classical communication in photon-starved situations • send more photons per bit for redundancy • use EDFAs to replenish and extend transmission range • pre-amp to restore SNR at detection stage
. Similar techniques to QKD without compromising security? • more photons per bit to mitigate channel loss • use EDFA to maintain SNR and extend range (mitigate loss) • do away with single-photon detectors Floodlight QKD protocol: a radically different paradigm
FL-QKD: a two-way protocol―breaks “standard” rules
. Use many photons over many more modes per bit • mean photon # per mode < 1 maintains no-cloning security
. Use standard PIN detectors • bit-value decoding based on homodyne reception • single-photon detectors are not needed for key distribution
. Use EDFA to mitigate channel loss • bit value can still be decoded with high probability • security is maintained
FL-QKD theory: PRA 94, 012322 (2016) Proof-of-principle demo: PRA 95, 012332 (2017) Gbps experiment: Quantum Sci. Technol. 3, 025007 (2018) Floodlight QKD: a two-way protocol 1550-nm ASE noise source: 0 or π One-way Bob-to-Alice protocol EDFA (no input) + 18-nm CWDM filter phase shift one-way channel loss ∼10 dB
ASE passive reference Eve delay line Eve’s BPSK receiver LO
homodyne amplifier receiver (gain + noise)
Alice standard Eve Bob InGaAs detectors 40 dB gain listens passively EDFA Bit error rates for direct communication (passive Eve)
opt Pr(e)Eve (BER)
10 Pr(e)Alice
log @ 200 photons/bit: 0.01 photon per mode over 20,000 modes (2 THz bandwidth)
Photons per bit Alice fully decodes at 100 Mb/s Eve has high BER if Alice uses less than ∼200 photons/bit Security against a smarter Eve (active eavesdropping) Eve injects her own light (SPDC, optimally) to decode Bob’s message Active monitor of channel integrity: inject SPDC light; measure singles and coincidences; Alice’s reference tap
idler single-photon single-photon SPDC active detector Eve detector signal beam ASE tap tap combiner passive reference Eve delay line Eve’s BPSK single-photon receiver LO detector
homodyne amplifier receiver (gain + noise)
Alice Eve Bob Floodlight QKD experimental setup
1570-nm idler SNSPD coinc SNSPD
PPLN CWDM coinc SNSPD 1550-nm Alice’s terminal signal wave- φ 0.1% EDFA CWDM attn + tap 98/2 BS shaper ASE mod tap 1550 nm 18-nm bandwidth >99% is ASE message attn delay line
dispersion Eve’s injection compensator CWDM
LO homodyne tunable CWDM EDFA receiver interferometrically delay line stable with Bob’s terminal BERT phase-lock loops SPDC-tap coincidence measurements 1570-nm idler SNSPD coinc SNSPD PPLN CWDM 1550-nm signal ASE CWDM + tap
accidental coincidences
Accidental measurements with & without SPDC signal sent to Bob yields SPDC signal ∼0.6% of total power
PRA 95, 012332 (2017) Received signal after homodyne detection
1550-nm 100 Mb/s ASE φ mod homodyne EDFA receiver
BERT
bit rate: 100 Mb/s ∼50 bits in 500-ns trace
Bob’s message bits Alice’s measured bits P Active monitoring of Eve’s injected light injected of Eve’s monitoring Active Alice f E SPDC power) SPDC in change (no injection Eve’s by unaffected remains rate bit Alice’s tap her own it with normalizing by Eve’s fraction f Alice estimates
= P
Eve Eve Eve’s f tap E
/ = P P
1 Total -error Eve – P
Total
True True
True coinc. rate (idlerTrue coinc. rate
Alice’s bit-error rateest ( rate coinc. 0.02 Alice’s 0.04 estimate 0.06 fE 0.08 0.10 0 0.6 0.4 0.2 0 idler Fraction Fraction Fraction Fraction Alice’s -Alice’s -Bob’s tap)/ Bob’s tapsingles rate of Eve’s injected Eve’s light of of Eve’s injected Eve’s light of tap) / Alice’s tap tap Alice’s
singles singles f f E actual E actual rate
Floodlight QKD secret-key rates
Secret key Reconciliation Mutual information Bob, Eve shared capacity efficiency between Alice, Bob Holevo information
modulation rate: 100 Mb/s IAB UBact β = 94% χEB LBpass ΔIAB UBpass χEB assuming only passive Eve, secret-key rate > 66 Mb/s
assuming active Eve, LBact ΔIAB secret-key rate > 55 Mb/s estimated fE = 0.27%
FL-QKD initial demo: PRA 95, 012332 (2017) Experimental setup for achieving Gbps rates
• High-speed phase modulator: 7 Gbps pseudo-random bit sequence
• High-speed balanced homodyne receiver: 14-GHz bandwidth
• Receiver noise: dominated by ASE noise from Bob’s EDFA
• 10-dB channel loss, each way
Gbps demo: Quantum Sci. Technol. 3, 025007 (2018) Results from high-rate experimental measurements
• 7 Gbps mod. rate: 143-ps bit duration, >900 temporal modes/bit
• 1.3 Gbps secret-key rate @ 20 transmitted photons/bit
• Bound on Eve’s injection fraction fE = 0.25% (No actual injection)
Eve’s Holevo information Alice-Bob mutual information
secret-key rate
Quantum Sci. Technol. 3, 025007 (2018) Practical FL-QKD implementation issues
. Two-way protocol and homodyne reception • phase stabilization of delayed-LO and signal paths • successful LL-campus-LL fiber loop stabilization [CLEO’17 FTu4F.6] • much tougher for free-space transmission paths
. Short coherence length • large bandwidth => short coherence time (0.1 ps) for matching paths
. Dispersion compensation and matching • not critical issue
. Detector saturation lengthens required channel monitor times • channel is secured only when monitor has enough measurements • single-photon detectors with fast recovery times • 4-detector setup: >100 Mcounts/s rate [OE 21, 1440 (2013)]
Summary
. Quantum illumination for sensing – experiment shows 20% improvement over best classical technique – quantum microwave radar: wishful thinking or real applications?
. QI-inspired floodlight QKD – novel protocol breaks conventional rules, overcomes channel loss – achieves >1 Gbps secret key rates, 103 better than BB84 system
. Quantum techniques are beneficial but application-specific – entanglement: useful even if there is significant loss and noise – no-cloning theorem: key to communication security Key contributors
. Theory (concept, protocol, security proofs) – Jeff Shapiro – Zheshen Zhang (now faculty at U. of Arizona) – Quntao Zhuang (now faculty at U. of Arizona) . Experiments – Zheshen Zhang (now faculty at U. of Arizona) – Sara Mouradian (now postdoc at UC Berkeley) – Changchen Chen, Jane Heyes (graduate students)