Quantum cryptography
Transcription
Quantum cryptography
Physics of Nanophotonic Systems Lecture 5 June, 10 th 2010 Quantum cryptography Single photons in action Prof. Jonathan J. Finley Dr. Michael Kaniber Thu, 15:30–17:30 S101 (WSI) Physics of Nanophotonic Systems SS 2010 Lecture 5 – Quantum cryptography 1 What is cryptography • The art of rendering a message unintelligible to any unauthorized party… Physics of Nanophotonic Systems SS 2010 Lecture 5 – Quantum cryptography 2 Historical overview: Cryptography http://en.wikipedia.org/wiki/History_of_cryptography Literature: Simon Singh, “The Code Book”, Fourth Estate, 2000 Physics of Nanophotonic Systems SS 2010 Lecture 5 – Quantum cryptography 3 Classical Cryptography Complexity theory Information theory (Asymmetric cryptosystem) (symmetric cryptosystems) Two keys needed – public key for encryption and private key for decryption The “public key” contains a private “decoding key” Security based on “one-way” functions (no one knows whether one way functions really exist) easy The key is secret Message is encrypted and decrypted using a decoding key: only the two partners have a copy ! Easy to crack if key is repeatedly used Only “Vernam cypher” is has been mathematically proven to be secure f(x) message + public key = cipher Difficult (unless private key known) Sometimes called RSA cryptography 128 bit key 2128 ≈ 1038 numbers to check using brute force ! Physics of Nanophotonic Systems SS 2010 Lecture 5 – Quantum cryptography 4 Vernam cipher or one-time-pad key 0101110010100011 Alice 0101110010100011 Bob Message: 1000100011110101 Key: 0101110010100011 Cipher: 1101010001010110 1000100011110101 message 1000100011110101 Sent: sum mod 2 Contains NO info! Unbreakable… unless the eavesdropper knows the key!!! How to distribute the key among the partners? Q-cryptography is a protocol for secure key distribution! i.e., purpose: establish two identical lists of secret bits. Physics of Nanophotonic Systems SS 2010 G. Vernam, J. Am. Inst. Elect. Eng. 45, 109 (1926) Lecture 5 – Quantum cryptography 5 Pro’s and con’s of the one-time-pad • Major advantage – Provides unbreakable security (can be mathematically proven) • Disadvantages – Key must be as long as the message – otherwise insecure… – Key must be absolutely random – Key must never be reused – otherwise security is compromised… – Alice and Bob must have a secure channel to exchange key beforehand… – Multi-party key exchange is a nightmare… Here QC comes into play: mathematically secure way to distribute a key between two parties QUANTUM KEY DISTRIBUTION Physics of Nanophotonic Systems SS 2010 Lecture 5 – Quantum cryptography 6 Remark: Random number generator Applications: Cryptography, confidentiality, authentication, simulation of complex systems, lotteries and gambling, etc. How to generate “randomness”? • Software solutions pseudo-random numbers • Classical physics (e.g. flip a coin) • Quantum physics (e.g. single photon emission http://www.idquantique.com/ Introductory article on idquantique.com “white paper” Create long random key to use as a one-time-pad Physics of Nanophotonic Systems SS 2010 Lecture 5 – Quantum cryptography 7 QKD – historical remarks • Founding ideas of “quantum cryptography” first proposed in the 1970’s by S. Wiesner (not published until 1982!) – advanced by Bennet and Brassard EPR Pairs (entanglement) S. Wiesner C. H. Bennett G. Brassard A. Ekert •• Fundamental tenetsofofquantum quantummechanics mechanics(N. Gisin) Fundamental “negative” “positive” tenets – – – – – You make aa quantum measurement without perturbing If youcan’t measure system, you change it…the system You can’t simultaneously measure quantum conjugatevariables variables(energy-time, with arbitrarypositionprecision Information encoded on conjugate You cannot duplicate an unknown quantum state momentum, circular-horizontal polarization) is difficult to measure unless you know the basis in which it is encoded… – Information encoded on quantum states is secure against copying… For a detailed review see - N. Gisin et al., Rev. Mod. Phys. 74, 145 (2002) Physics of Nanophotonic Systems SS 2010 Lecture 5 – Quantum cryptography 8 Key arguments of QKD • If Eve tries to eavesdrop a "quantum communication channel", she has to perform some measurements on individual quanta (single photon pulses) to gain any information… The "quantum communication channel" is not used to transmit a message (information), only a "key" is transmitted (no information). • But, quantum mechanics tells us: every measurement projects the quantum system into an eigenstate…. If it turns out that the key is corrupted, they simply disregard this key and no information is really lost. • Hence, Eve’s "reading" of the "quantum signal“ reduces the correlation between Alice's and Bob's data… If the key passes successfully the control test then Alice and Bob (or George and Tony) can use it safely. • Alice and Bob can thus detect any undesired third party by comparing (on a public channel) part of their "quantum signal“… Physics of Nanophotonic Systems SS 2010 Lecture 5 – Quantum cryptography 9 “Stern-Gerlach” type experiment for photons Reminder Stern-Gerlach experiment: Magnet deflects particles with spin quantum number of ½ Depending on its spin, particle is deflected up or down Analogue phenomenon for photons: polarisation Vertically polarised photon Propability to detect either |↕ or |↔ : 𝑃𝑣 = ↕ Θ 𝑃ℎ = ↔ Θ 2 = cos 2 Θ 2 = sin2 Θ Quantum state: Physics of Nanophotonic Systems SS 2010 horizontally polarised photon Normalisation condition: 𝑐𝑜𝑠𝜃 2 + 𝑠𝑖𝑛𝜃 2 =1 Lecture 5 – Quantum cryptography 10 Quantum properties of light Polarization is a quantum number for each photon The state of polarisation determines the direction of oscillation of macroscopic field 0 H-V basis 1 1 (0 1) 2 1 (0 1) 2 Physics of Nanophotonic Systems SS 2010 +45/-45 basis Lecture 5 – Quantum cryptography 11 State preparation and read-out Alice’s preparation 0 450 1 (0 1) 2 Bob’s measurement 1 RC -450 A E 1 (0 1) 2 N R 1 (0 1) 2 1 (0 1) 2 D T O A M I ! N 0 00 0 1 900 1 Physics of Nanophotonic Systems SS 2010 0 1 Lecture 5 – Quantum cryptography 12 BB84 protocol for quantum key exchange BB84 (Bennett + Brassard in 1984, first demonstrated in 1989 See also B92 protocol – C. H. Bennett, PRL 68, 3121 (1992)) – Polarization encoding system using 4 quantum states that are typically polarization states of single photons 45 , 45 H ,V – The based are “maximally conjugate” – any two states, one from each basis, have the same overlap H 45 2 V 45 2 H 45 2 V 45 2 1 2 – One attaches binary values to each quantum state within one basis H ,V Physics of Nanophotonic Systems SS 2010 0,1 45 , 45 0,1 Lecture 5 – Quantum cryptography 13 BB84 coding Alice’s bit 0= 1= 0= 1= Bob’s measurement basis |V> |H> |+45> |-45> Bob’s result or 50:50 or 50:50 or 50:50 or 50:50 Essential: use two sets of non-orthogonal states. Physics of Nanophotonic Systems SS 2010 Lecture 5 – Quantum cryptography 14 BB84 scheme in action 1) Alice sends individual, randomly chosen “spins” to Bob from the 4 basis states 0 0 0 1 1 0 0 1 0 1 0 2) Bob “measures” these spins in one of the two bases chosen at random ,, … This produces the “raw key” 0 ? 0 1 1 0 ? 1 0 ? 0 Whenever Alice and Bob’s bases randomly coincide they obtain correlated results BUT whenever they do not coincide the measurements are 50:50 (uncorrelated) The “raw” key has a high (25%) probability that an error occurs 3) The “raw” key is then “sifted” by exchanging some classical information* *Alice and Bob exchange just the choice of bases not the result Bit 2 was measured in Not OK, OK… , bit 3 was …. Sifted key Discard bits where they used different bases Physics of Nanophotonic Systems SS 2010 0 0 1 1 0 1 0 0 completely random determined by Alice and Bobs random choices Lecture 5 – Quantum cryptography 15 Apparatus for quantum cryptography Pockels cell: voltage controlled waveplate/polariser Alice applies voltages to her Pockels cell and produces polarization rotations - 0° = |↕ - 90° = |↔ - 45° = |↗ - 135°= |↙ Bob randomly applies voltages to his Pockels cell and switches between the H/V- and +45/-45-basis Physics of Nanophotonic Systems SS 2010 Lecture 5 – Quantum cryptography 16 First experimental demonstration The first prototype implementation of quantum cryptography (IBM, 1989) C. H. Bennett, Proceedings of IEEE International Conference on Computers, Systems and Signal processing (1984) - Quantum Cryptography: Public key distribution and coin tossing - Physics of Nanophotonic Systems SS 2010 Lecture 5 – Quantum cryptography 17 Eve is on the line… • Let us now consider the security of this “ideal” protocol Beam-splitter Alice sends her qubit on a supposedly secure line… Bob chose the “good basis” according to his conversation with Alice Eve intercepts the photon and tries to gain information about it She has a 50/50 chance of measuring in the wrong basis and, thus, resending the wrong basis to Bob nevertheless he got the wrong result? There are some errors in the communication? After revealing the bases, Alice and Bob reveal also some results, and check the correlations. By measuring the error, they can estimate the amount of information Eve received Eve cannot make a copy of the original photon sent by Alice due to the NO-CLONING theorem of quantum mechanics – it is impossible to “clone” a quantum state Imperfections in the losses in the quantum channel, Alice’s single photon generator and Bob’s single photon detector also introduce errors – how to distinguish from the action of Eve ? Physics of Nanophotonic Systems SS 2010 Lecture 5 – Quantum cryptography 18 Intercept-resend attack Alice sends her qubits on a quantum channel that has been “tapped”… Beam-splitter or 50:50 On average HALF of the qubits intercepted by Eve are wrongly measured Alice and Bob discover Eve’s intervention for 50% of all qubits sent between them since they get uncorrelated results If Eve uses this “interceptresend” strategy, she gets 50% information, while Alice and Bob have about a 25% error rate in their sifted key Bob chose the “good basis” according to his conversation with Alice Eve intercepts all photons from Alice (exactly as Bob does in the secure case) EVE is lucky Bob gets the “right result” EVE chooses the wrong basis Bob obtains the “wrong result” has a 50/50 chance of measuring in the correct basis she then resends a photon to Bob in the “correct basis” When she chooses the “wrong basis” she gains no information and resends in the wrong basis However, if Eve applies this strategy to only a fraction of the communication (say, 10%) then the error rate will be only ~2.5%, whilst Even ends up with 5% of the qubits sent Even if they eliminate the cases where they used incompatible bases there is still 25% error in the sifted key and they “detect” Eve’s attack Physics of Nanophotonic Systems SS 2010 How can Alice and Bob counter such attacks ? Lecture 5 – Quantum cryptography 19 Potential error sources Random error deletion: • Absorption/scattering • Inefficient light collection • Limited detector QE Not lowering security! Birefringence: • Medium in which the photons propagate from Alice to Bob change the polarization Bob obtains wrong result even if the basis was correct! Alice and Bob exchange bits and perform error correction: Shannon’s noisy channel coding theorem C. E. Shannon, Bell Syst. Tech. J. 27, 379 (1948) Needed bits for error correction Number of bits Detector dark counts: • Random photons due to thermal noise Physics of Nanophotonic Systems SS 2010 Error rate 100 dark counts/s Detector gating! Lecture 5 – Quantum cryptography 20 Error correction and privacy amplification • Alice and Bob share a sifted key that contains errors which are caused by technical imperfections as well as by Eve’s intervention – – • Realistic error rates in the sifted key using today’s technologies (SPS, Q-channels and SP-detectors) are typically a few percent How to distinguish between Eve and errors ? Consequently the final steps in all QKD protocols are: 1. Error correction 2. Reduce Eve’s information on the final key used via “privacy amplification” “SIMPLE” ERROR CORRECTION Alice randomly chooses pairs of bits and announces publically to Bob their XOR value (sum modulo 2) Bit 1 Bit 2 XOR 0 0 0 If they get the same XOR value then Alice and Bob keep the first of the two bits and disregard the second 0 1 1 1 0 1 They continue to repeat this protocol until they have a shorter sifted key that now contains no errors… 1 1 0 Physics of Nanophotonic Systems SS 2010 Lecture 5 – Quantum cryptography 21 Classical error correction in QKD 1001110111001101110010 0 REPEAT 0010100101010101010001 0⊕0=0 “accept” “reject” 0010110110010001010011 Alice and Bob end up with a shorter, but error free, key 0 0 1 0 … In reality XOR operations are performed on larger bit sets but principles are similar Physics of Nanophotonic Systems SS 2010 Lecture 5 – Quantum cryptography 22 Error correction and privacy amplification • Alice and Bob share a sifted key that contains errors which are caused by technical imperfections as well as by Eve’s intervention – – • Realistic error rates in the sifted key using today’s technologies (SPS, Q-channels and SP-detectors) are typically a few percent How to distinguish between Eve and errors? Consequently the final steps in all QKD protocols are: 1. Error correction 2. Reduce Eve’s information on the final key used via “privacy amplification” “SIMPLE” PRIVACY AMPLIFICATION “SIMPLE” ERROR CORRECTION Alice again randomly chooses pairs of bits and computes their XOR value Alice randomly chooses pairs of bits and announces publically to Bob their XOR value (sum modulo 2) She does NOT announce the XOR value, but instead only the bits she chose (e.g. bits number 7 and 555) If they get the same XOR value then Alice and Bob keep the first of the two bits and disregard the second They continue to repeat this protocol until they have a shorter sifted key that now contains no errors… Physics of Nanophotonic Systems SS 2010 Alice and Bob then replace the two bits by their XOR value and, thus, shorten their key whilst keeping it error free BUT – Eve only has partial information on the two bits, her information on the XOR value is even less. Example: If Eve knows the value of both bits chosen with 60% probability then the probability that she correctly guesses the XOR value is only 0.62+0.42=52% by repeating many times Eve’s information is continuously reduced Lecture 5 – Quantum cryptography 23 Privacy amplification? 1000110100101010010 REPEAT 10100101 Error free 10100101 10 Physics of Nanophotonic Systems SS 2010 10 Lecture 5 – Quantum cryptography 24 Summary of QKD • Quantum cryptography is really a method to reliably send a quantum key that is subsequently used with the “one time pad” Vernam cipher • Many “protocols” have been developed for QKD each of which use nonorthogonal basis sets – E.g. BB84 which uses (H,V) and (+45,-45) polarization states of light – “Send and measure” with some classical communication resources – Final steps consist of a “classical” error correction and privacy amplification • To detect the effects of Eve, Alice selects a random subset of the qubits to be announced as test bits. Alice and Bob compare these bits to learn the error rate. • If the error rate is small enough, the test is “passed” and Alice announces the error correction information so Bob can correct his errors. • Now, Alice and Bob have the same strings, but Eve may have some information. Alice announces privacy amplification information to reduce Eve’s information to zero. Physics of Nanophotonic Systems SS 2010 Lecture 5 – Quantum cryptography 25 Lecture 5 J. J. Finley Fundamentals of photonic crystals Physics of Nanophotonic Systems SS 2010 Lecture 5 – Quantum cryptography 26