Quantum Fingerprinting: Difference between revisions

From Quantum Protocol Zoo
Jump to navigation Jump to search
m (Reverted edits by Shraddha (talk) to last revision by Bob)
Tag: Rollback
No edit summary
 
(11 intermediate revisions by 3 users not shown)
Line 1: Line 1:
Quantum fingerprinting allows two parties (two quantum clients) to collaboratively compute the value of a function using both of their inputs while maintaining the privacy of their own input. The protocol does not permit the two parties to interact directly with each other, hence they send their respective inputs to a trusted third party (quantum server) who computes the correct value of the function corresponding to both inputs while minimizing the amount of information sent by the two parties.  
This [https://arxiv.org/abs/quant-ph/0102001 example protocol] allows two quantum clients to distinguish between their quantum inputs while maintaining the privacy of their own input just by comparing the fingerprints of their inputs. The protocol does not permit the two parties to interact directly with each other, hence they send the fingerprints of their respective inputs to a trusted third party (quantum server). This server performs a test to distinguish between two unknown quantum fingerprints with a high probability. The quantum fingerprints are exponentially shorter than the original inputs.
</br></br>
 
Minimization of information is done by fingerprinting, which is a technique that associates a string to an exponentially smaller string (fingerprints) such that any two strings can by distinguished solely based on their fingerprints.
'''Tags:''' [[Fingerprinting]]
</br></br>
'''Tags:''' Fingerprinting


==Assumptions==
==Assumptions==
* The two quantum clients have no shared key in this protocol.
* The server is trusted.
* The fingerprints can consist of quantum information.


==Outline==
==Outline==
The two quantum clients want to compute a function using the input of both parties. In this protocol, we focus on a specific function, which is the equality function. This function provides boolean output 1 if the input of both the parties is the same, otherwise, the output is 0.
Here, two quantum clients want to check if their quantum inputs are distinct while also keeping their inputs secret. They prepare quantum fingerprints of their individual inputs and send these states to the server. Next stage involves the server performing a SWAP test on the fingerprints to check their equality. The server repeats this test several times on the received fingerprints to reduce the error probability.
</br>


Preparation stage involves two parties (two quantum clients) preparing the fingerprints of their individual inputs and sending their quantum states to the server (trusted third party). The two clients do not possess any shared quantum key.  
* '''Client's preparation''':
</br>
** The client prepares the fingerprint of initial input which is sized <math>n</math>-bits. This fingerprint has a length of <math>O(\log{}n)</math> bits.
** This fingerprint is prepared using particular error correcting codes, which converts the <math>n</math>-bit input to <math>m</math>-bits, where <math>m</math> is greater than <math>n</math>, and the two outputs of any two distinct inputs can be equal at atmost <math>\delta m</math> positions, where <math>\delta < 0</math>. The fingerprint has the length of <math>log{}m+1</math> bits.
** Here for error correcting code, [https://ieeexplore.ieee.org/document/1054893 Justesen codes] are used.
** The client now sends this fingerprint to the server through a quantum channel. Both the clients do this process simultaneously.


Next stage involves the server performing several operations on the fingerprints to check their equality. These operations are repeated several times on the same fingerprints to reduce the error probability.
* '''Server's test''': The server receives the two fingerprints from both the clients and performs the [[Glossary#Quantum SWAP test|quantum SWAP Test]] on these states to check if the states are distinguishable. The server independently repeats this SWAP test on fingerprints several times to reduce the error probability in detecting if the two states are different.
</br>
* '''Client's preparation''': The client prepares the fingerprint of initial input sized <math>n</math>-bits. This fingerprint has a length  of <math>\log_{}n</math> bits. The client now sends this fingerprint to the server through a quantum channel. Both the clients do this process simultaneously.


* '''Server's preparation''': The server receives the two fingerprints from both the clients and prepares the operations to distinguish them. The server independently repeats the computation process with the fingerprints several times to reduce the error probability in detecting whether the equality of the two states.
==Hardware Requirements==
 
* Authenticated Quantum channel capable of sending a pair of qubits.
* '''Measurement''': The server measures the final states and announces the result.
* Measurement devices for the server.
* A one-time quantum channel from both clients to the server.


If the initial input of the two parties are equal, this would be inferred from the corresponding fingerprints with no error probability and the outcome of the protocol would be correct. However, if the two fingerprints are of different inputs, there exists a non-zero probability that the outcome of the protocol is incorrect. Therefore, there exists a one sided error in the measurement. This one sided error is reduced by repeating the server operations several times with the same fingerprints.
==Knowledge Graph==


==Requirements==
{{graph}}
* Authenticated Quantum channel capable of sending a pair of qubits.
* Quantum memory for server to store the fingerprints.
* Measurement devices for the server.


==Notation==
==Notation==
* <math>|h_x\rangle</math>, Quantum fingerprint for <math>n</math>-bit input <math>x</math>.
* <math>x</math>, <math>y</math>: inputs of both the clients
* <math>E(x)</math>, Fingerprint function associated with input <math>{x\in \{0, 1\}^n}</math> which maps <math>n</math>-bit input to <math>m</math>-bit fingerprint.
* <math>n</math>: length of inputs
* <math>\delta</math>, Fixed constant, such that <math>\delta<0</math>.
* <math>m</math>: Length of output of error correcting codes, using x and y as input.
* <math>E(x)</math>: Error correcting code associated with input <math>{x\in \{0, 1\}^n}</math>, where <math>E: \{0, 1\}^n \xrightarrow{}{} \{0, 1\}^m </math>.
* <math>|h_x\rangle</math>: <math>(log{}m+1)</math> qubit state quantum fingerprint for <math>x</math>.
<math>|h_x\rangle = \frac{1}{\sqrt{m}}\sum_{i=1}^{m} |i\rangle|E_i(x)\rangle</math>
* <math>c</math>: Parameter for error correcting code. <math>m=cn, c>1</math>
* <math>\delta</math>: Parameter for error correcting code. <math>\delta<1</math>.
* <math>\epsilon</math>: Error probability
* <math>k</math>: Number of times a fingerprint is repeated, <math>k \in O(log{}{\frac{1}{\epsilon}})</math>


==Properties==
==Properties==
* The computational complexity of this protocol is <math>\mathcal{O}(\log{}n)</math>.
* The computational complexity of this protocol is <math>\mathcal{O}(\log{}n)</math>.
* The two quantum clients have no shared key in this protocol.
* Given an <math>n</math>-bit input, the protocol requires a quantum fingerprint of minimum <math>\log{}n</math> bits which contains quantum information.
* Given an <math>n</math>-bit input, the protocol requires a quantum fingerprint of minimum <math>\log{}n</math> bits which contains quantum information.
* The quantum fingerprint is defined as the state <math>|h_x\rangle</math>, where <math>{E(x)}</math> is the fingerprint of the input <math>x</math>. <math>{E_i(x)}</math> is the <math>{i^{th}}</math> bit of <math>{E(x)}</math>. </br>
* The Hamming distance i.e. the number of positions between two strings of same length, at which the corresponding symbols are different, of <math>{E(x)}</math> and <math>{E(y)}</math> is at least <math>{(1+\delta)m}</math>.
<math>|h_x\rangle = \frac{1}{\sqrt{m}}\sum_{n=1}^{m} |i\rangle|E_i(x)\rangle</math>
* For Justesen codes, <math>\delta < 9/10 + 1/(15c)</math> for any chosen <math>c>2</math>
* The hamming distance i.e. the number of positions between two strings of same length, at which the corresponding symbols are different, of <math>{E(x)}</math> and <math>{E(y)}</math> is at least <math>{(1+\delta)m}</math>.
* Any two fingerprints, <math>|h_x\rangle</math> and <math>|h_y\rangle</math> have an inner product of at most <math>{\delta}</math>.
* Any two fingerprints, <math>|h_x\rangle</math> and <math>|h_y\rangle</math> have an inner product of at most <math>{\delta}</math>.
* The one sided error probability is at least <math>(\frac{1+\delta}{2})^k</math> if the server computation is repeated <math>k</math> times.
* The one sided error probability of the SWAP test is <math>(\frac{1+\delta}{2})</math>.
* Quantum memory would be required to store the fingerprints if the server operations are performed <math>k</math> times.
* This protocol requires no quantum memory for the Client


==Pseudocode==
==Protocol Description==
'''Input''': <math>{x \in \{0, 1\}^n}</math> to First Party and <math>{y \in \{0, 1\}^n}</math> to Second Party. </br>
'''Stage 1''': Client's preparation</br>
'''Output''': One bit by server satisfying the equality function with some error probability. </br>
'''Input''': <math>{x \in \{0, 1\}^n}, {y \in \{0, 1\}^n}</math> for first client and second client respectively. </br>
'''Stage 1''': Client's preparation
'''Output''': <math>|h_x\rangle</math>, <math>|h_y\rangle</math> sent to server
* First Party prepares the fingerprint <math>|h_x\rangle</math> from input <math>x</math>.
* First client prepares <math>|h_x\rangle</math> from <math>x</math>, <math>|h_x\rangle = \frac{1}{\sqrt{m}}\sum_{i=1}^{m} |i\rangle|E_i(x)\rangle</math>
* Second Party prepares the fingerprint <math>|h_y\rangle</math> from input <math>y</math>.
* Second client prepares <math>|h_y\rangle</math> from <math>y</math>, <math>|h_y\rangle = \frac{1}{\sqrt{m}}\sum_{i=1}^{m} |i\rangle|E_i(y)\rangle</math>
* Both parties transmit their fingerprints to the server.
* Both clients send <math>|h_x\rangle</math>, <math>|h_y\rangle</math> to server


'''Stage 2''': Server's preparation
 
* Server prepares an ancilla qubit <math>|0\rangle</math> for final measurement purpose, and thus starts with the state <math>|0\rangle|h_x\rangle|h_y\rangle</math>.
'''Stage 2''': Server's test
* Server creates an entangled state by applying the gate <math>G</math>.
'''Input''': <math>|h_x\rangle</math>, <math>|h_y\rangle</math> </br>
<math>G = {(H\otimes I)(c-SWAP)(H\otimes I)}</math>
'''Output''':  SWAP test result
* The server measures the first qubit and transmits the output to both the parties.
* Server prepares an ancilla qubit <math>|0\rangle</math> for SWAP test and starts with state <math>|0\rangle|h_x\rangle|h_y\rangle</math>
* Server applies gate <math>G = {(H\otimes I)(c-SWAP)(H\otimes I)}</math>, resulting in final state <math>\frac{1}{2}|0\rangle(|h_x\rangle|h_y\rangle + |h_y\rangle|h_x\rangle) + \frac{1}{2}|1\rangle(|h_x\rangle|h_y\rangle - |h_y\rangle|h_x\rangle)</math>
* The server measures the first qubit and transmits the result to both the clients.


==Further Information==
==Further Information==
* To reduce the error probability to any <math>\epsilon</math>, the fingerprint of <math>x</math> should be set to <math>|h_x\rangle^{\otimes k}</math>, for a suitable <math>k \in O(log{}{\frac{1}{\epsilon}})</math>
==Related Papers==
==Related Papers==


<div style='text-align: right;'>''*contributed by Rhea Parekh''</div>
<div style='text-align: right;'>''*contributed by Rhea Parekh''</div>

Latest revision as of 16:28, 16 October 2019

This example protocol allows two quantum clients to distinguish between their quantum inputs while maintaining the privacy of their own input just by comparing the fingerprints of their inputs. The protocol does not permit the two parties to interact directly with each other, hence they send the fingerprints of their respective inputs to a trusted third party (quantum server). This server performs a test to distinguish between two unknown quantum fingerprints with a high probability. The quantum fingerprints are exponentially shorter than the original inputs.

Tags: Fingerprinting

Assumptions[edit]

  • The two quantum clients have no shared key in this protocol.
  • The server is trusted.
  • The fingerprints can consist of quantum information.

Outline[edit]

Here, two quantum clients want to check if their quantum inputs are distinct while also keeping their inputs secret. They prepare quantum fingerprints of their individual inputs and send these states to the server. Next stage involves the server performing a SWAP test on the fingerprints to check their equality. The server repeats this test several times on the received fingerprints to reduce the error probability.

  • Client's preparation:
    • The client prepares the fingerprint of initial input which is sized -bits. This fingerprint has a length of bits.
    • This fingerprint is prepared using particular error correcting codes, which converts the -bit input to -bits, where is greater than , and the two outputs of any two distinct inputs can be equal at atmost positions, where . The fingerprint has the length of bits.
    • Here for error correcting code, Justesen codes are used.
    • The client now sends this fingerprint to the server through a quantum channel. Both the clients do this process simultaneously.
  • Server's test: The server receives the two fingerprints from both the clients and performs the quantum SWAP Test on these states to check if the states are distinguishable. The server independently repeats this SWAP test on fingerprints several times to reduce the error probability in detecting if the two states are different.

Hardware Requirements[edit]

  • Authenticated Quantum channel capable of sending a pair of qubits.
  • Measurement devices for the server.
  • A one-time quantum channel from both clients to the server.

Knowledge Graph[edit]

Notation[edit]

  • , : inputs of both the clients
  • : length of inputs
  • : Length of output of error correcting codes, using x and y as input.
  • : Error correcting code associated with input , where .
  • : qubit state quantum fingerprint for .

  • : Parameter for error correcting code.
  • : Parameter for error correcting code. .
  • : Error probability
  • : Number of times a fingerprint is repeated,

Properties[edit]

  • The computational complexity of this protocol is .
  • Given an -bit input, the protocol requires a quantum fingerprint of minimum bits which contains quantum information.
  • The Hamming distance i.e. the number of positions between two strings of same length, at which the corresponding symbols are different, of and is at least .
  • For Justesen codes, for any chosen
  • Any two fingerprints, and have an inner product of at most .
  • The one sided error probability of the SWAP test is .

Protocol Description[edit]

Stage 1: Client's preparation
Input: for first client and second client respectively.
Output: , sent to server

  • First client prepares from ,
  • Second client prepares from ,
  • Both clients send , to server


Stage 2: Server's test Input: ,
Output: SWAP test result

  • Server prepares an ancilla qubit for SWAP test and starts with state
  • Server applies gate , resulting in final state
  • The server measures the first qubit and transmits the result to both the clients.

Further Information[edit]

  • To reduce the error probability to any , the fingerprint of should be set to , for a suitable

Related Papers[edit]

*contributed by Rhea Parekh