Quantum key distribution system based on multi-classification learning detection and implementation method thereof

Abstract

The application provides a quantum key distribution system based on multi-classification learning detection and an implementation method thereof, comprising a sending end Alice, a receiving end Bob and a post-processing program module based on multi-classification learning; the sending end Alice modulates a quantum signal by using a Gaussian modulation method, and sends the modulated quantum signal to the receiving end Bob through a quantum channel; the receiving end Bob is a quantum key receiving end Bob, uses a homodyne detector to measure the received quantum signal, uses a power meter to measure the intensity of the local oscillator, uses a clock circuit to provide a clock signal, and finally sends the quantum signal data to the post-processing program module based on multi-classification learning through a classical quantum channel; the post-processing program module based on multi-classification learning analyzes and processes the quantum signal data to generate a final secure key. The application can detect various quantum hacker attacks on the continuous variable quantum key distribution system, thereby enhancing the security of the continuous variable quantum key distribution system under Gaussian modulation.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of quantum communication, in particular to a quantum key distribution system based on multi-classification learning detection and an implementation method thereof. BACKGROUND

[0002] Quantum key distribution technology, especially continuous variable quantum key distribution technology, plays a key role in future quantum communication applications. It allows the sender Alice (corresponding to Bob in the terminology of coherent state Gaussian modulation quantum key distribution protocol, generally referred to as continuous variable quantum key sender Alice, and receiver Bob is called Bob, and the third party eavesdropper is called Eve) and the receiver Bob to establish a shared key in an untrusted quantum channel, which cannot be encrypted and changed by the third party eavesdropper Eve.

[0003] The shared key of the continuous variable quantum key distribution technology is encoded in the quadrature components of the coherent state and decoded by coherent detection. In addition, the unconditional security of the continuous variable quantum key distribution technology is theoretically guaranteed by the principles of quantum mechanics, such as the Heisenberg uncertainty principle and the quantum no-cloning theorem. However, there is a large deviation between the theoretical framework (i.e. perfect optical devices and sufficient assumptions) and the actual physical system, which leads to security vulnerabilities, making the system vulnerable to various quantum hacker attacks.

[0004] At present, the continuous variable quantum key distribution technology has not been commercialized, and the main reason is that the continuous variable quantum key distribution system is vulnerable to various quantum hacker attacks, thereby endangering the security of the system. Therefore, how to detect whether the system is subjected to quantum hacker attacks is extremely important to the security of the system. SUMMARY

[0005] The present application provides a continuous variable quantum key distribution system based on multi-classification learning detection and an implementation method thereof, which solves the technical problem of quantum hacker attacks existing in the continuous variable quantum key distribution system and provides a new solution for improving the security of the system.

[0006] In order to achieve the above purpose, the quantum key distribution system based on multi-classification learning detection and the implementation method thereof provided by the present application, as shown in Figure 1 includes a sender Alice, a receiver Bob and a post-processing program module based on multi-classification learning;

[0007] The sender Alice modulates the Gaussian quantum signal, and sends the modulated quantum signal to the receiver Bob through the quantum channel;

[0008] The receiving end Bob is a quantum key receiving end Bob, uses a homodyne detector to measure the received quantum signal, uses a power meter to measure the local oscillator light intensity, uses a clock circuit to provide a clock signal, and finally sends the quantum signal data to a post-processing program module based on multi-classification learning through a classical quantum channel.

[0009] The post-processing program module based on multi-classification learning is used to detect the quantum signal data sent by Bob, analyzes and processes the quantum signal data through a multi-class support vector machine algorithm, and generates a final secure key according to the processing result and the key negotiation with Alice.

[0010] Preferably, the sending end Alice comprises: a sending end pulsed laser, a first polarizer, a first optical variable attenuator, a first amplitude modulator, a first phase modulator, a second optical variable attenuator, and a first polarization beam combiner.

[0011] The sending end pulsed laser prepares pulsed coherent light, which is sent to the input end of the first polarizer; the first polarizer separates the received quantum signal into two beams of light, including signal light and local oscillator light, wherein the first beam of signal light is sent from the first output end of the first polarizer to the input end of the first optical variable attenuator, the output end of the first optical variable attenuator is connected to the input end of the first amplitude modulator, the output end of the first amplitude modulator is connected to the first input end of the first phase modulator, the output end of the first phase modulator is connected to the input end of the second optical variable attenuator, and the output end of the second optical variable attenuator is connected to the second input end of the first polarization beam combiner; the second beam of local oscillator light is connected from the second output end of the first polarizer to the first input end of the first polarization beam combiner, the first polarization beam combiner combines the two beams of light into one beam of light, which is sent to the receiving end Bob through a quantum channel, and the output end of the first polarization beam combiner is connected to the input end of the receiving end Bob.

[0012] Preferably, the receiving end Bob comprises: a first polarization beam splitter, a first optical switch, a first beam splitter, a first homodyne detector, a second beam splitter, a second amplitude modulator, a first PIN photodiode, a first power meter, and a first clock circuit.

[0013] The first polarization beam splitter receives the quantum signal sent by the sending end Alice, divides the quantum signal into two beams of light, including signal light and local oscillator light, the first beam of signal light is sent from the first output end of the first polarization beam splitter to the input end of the first optical switch, the output end of the first optical switch is connected to the first input end of the first beam splitter, the output end of the first beam splitter is connected to the input end of the first homodyne detector, and the output end of the first homodyne detector is connected to the first input end of the post-processing program module based on multi-classification learning; the second beam of local oscillator light is sent from the second output end of the first polarization beam splitter to the input end of the second beam splitter, the first output end of the second beam splitter is connected to the input end of the second amplitude modulator, the output end of the second amplitude modulator is connected to the second input end of the first beam splitter, the second output end of the second beam splitter is connected to the input end of the first PIN light-emitting diode, the first output end of the first PIN light-emitting diode is connected to the input end of the first power meter, the output end of the first power meter is connected to the second input end of the post-processing program module based on multi-classification learning, the second output end of the first PIN light-emitting diode is connected to the input end of the first clock circuit, and the output end of the first clock circuit is connected to the third input end of the post-processing program module based on multi-classification learning.

[0014] Preferably, the post-processing program module based on multi-classification learning includes a PC (personal computer) end, which analyzes and processes the collected quantum signal data to generate a final security key.

[0015] Preferably, the PC end trains 70% of the received data as training data by analyzing and processing the collected quantum signal data, marks different types of quantum hacker attacks, classifies and detects the remaining 30% of the data, and performs parameter estimation, reverse negotiation and privacy amplification operations on the data after classification detection and the sending end Alice to generate a final security key.

[0016] Preferably, the data analysis and processing process is specifically performed according to the following steps:

[0017] Step P1. The receiving end Bob collects quantum signal data and normal signal data under different quantum hacker attacks, adjusts 10% of the maximum attenuation by the first optical switch, measures the shot noise variance N0, and calculates the mean value of the measurement value with 90% of the minimum attenuation and variance (V y ); lo ;

[0018] V y =ηT(V A N0+ξ)+N0+V el ,

[0019] wherein η is the detection efficiency of the homodyne detector, VA is the modulation variance of the sending end Alice, ξ is over noise, V el is electric noise, N0 is the variance of shot noise;

[0020] Step P2.PC end pre-processes data, adopts discrete standardization to linearly transform data:

[0021] X * =(x-Min) / (Max-Min)

[0022] Wherein, Max is the maximum value of data, Min is the minimum value of data, x is the data before processing, X * is the data after processing;

[0023] Then 70% of the data after processing is trained, 30% of the data is tested, and the trained multi-class support vector machine algorithm model is saved for identification and detection of quantum hacker attack data, and the model structure is as follows:

[0024]

[0025]

[0026] Wherein, ω m It is the optimization parameter corresponding to the mth sub-classifier, C is a penalty parameter, l represents l sample data, Indicates a slack variable, φ(X i ) indicates that the kernel function is mapped to a high-dimensional feature space, bm indicates a bias term, Y i =m indicates that the Y i Sample data belongs to the category corresponding to the mth sub-classifier; i is the sample data number;

[0027] Step P3.The multi-class support vector machine algorithm model saved in the PC end can be directly used for quantum hacker attack detection, so as to identify whether the continuous variable quantum key distribution system is subjected to quantum hacker attack.

[0028] The continuous variable quantum key distribution system based on multi-classification learning detection and the implementation method thereof provided by the application, the post-processing program module based on multi-classification learning first detects the collected signal data through the multi-class support vector machine algorithm, and once it is found that the system is subjected to quantum hacker attack, the transmission of quantum key is immediately terminated, therefore, the application can realize real-time quantum hacker attack detection, accurately identify and detect the quantum hacker attack existing in the system, and further enhance the actual security of the system. BRIEF DESCRIPTION OF DRAWINGS

[0029] Figure 1A structural block diagram of an implementation of a preferred embodiment of the quantum key distribution system based on multi-classification learning detection and the implementation method thereof;

[0030] Figure 2 A functional module diagram of a preferred embodiment of the quantum key distribution system based on multi-classification learning detection and the implementation method thereof. DETAILED DESCRIPTION

[0031] In order to make the technical problems, technical solutions and advantages to be solved by the present application clearer, specific embodiments will be described in detail below with reference to the drawings and specific embodiments.

[0032] The present application provides a quantum key distribution system based on multi-classification learning detection and an implementation method thereof to solve the existing problems.

[0033] In one embodiment, as shown in Figure 1 and Figure 2 The quantum key distribution system based on multi-classification learning detection and the implementation method thereof include a sending end Alice, a receiving end Bob, and a post-processing program module based on multi-classification learning.

[0034] The sending end Alice modulates a quantum signal with Gaussian and sends the modulated quantum signal to the receiving end Bob through a quantum channel.

[0035] The receiving end Bob is a quantum key receiving end Bob, uses a homodyne detector to measure the received quantum signal, uses a power meter to measure the local oscillator light intensity, uses a clock circuit to provide a clock signal, and finally sends the quantum signal data to the post-processing program module based on multi-classification learning through a classical quantum channel.

[0036] The post-processing program module based on multi-classification learning is used to detect the quantum signal data sent by Bob, analyzes and processes the quantum signal data through a multi-class support vector machine algorithm, and generates a final secure key according to the processing result and the key negotiation with Alice.

[0037] The sending end Alice includes a sending end pulsed laser 1, a first polarizer 2, a first optical variable attenuator 3, a first amplitude modulator 4, a first phase modulator 5, a second optical variable attenuator 6, and a first polarization beam combiner 7.

[0038] The sending end pulse laser 1 prepares pulse coherent light, and sends the pulse coherent light to the input end of the first polarizer 2; the first polarizer 2 separates the received quantum signal into two beams of light, including signal light and local oscillator light, wherein the first beam of signal light is sent from the first output end of the first polarizer 2 to the input end of the first optical variable attenuator 3, the output end of the first optical variable attenuator 3 is connected to the input end of the first amplitude modulator 4, the output end of the first amplitude modulator 4 is connected to the first input end of the first phase modulator 5, the output end of the first phase modulator 5 is connected to the input end of the second optical variable attenuator 6, and the output end of the second optical variable attenuator 6 is connected to the second input end of the first polarization beam combiner 7; the second beam of local oscillator light is connected from the second output end of the first polarizer 2 to the first input end of the first polarization beam combiner 7, the first polarization beam combiner 7 combines the two beams of light into one beam of light, and sends the one beam of light to the receiving end Bob through a quantum channel, and the output end of the first polarization beam combiner 7 is connected to the input end of the receiving end Bob.

[0039] The receiving end Bob comprises a first polarization beam splitter 8, a first optical switch 9, a first beam splitter 10, a first homodyne detector 11, a second beam splitter 12, a second amplitude modulator 13, a first PIN light-emitting diode 14, a first power meter 15, and a first clock circuit 16.

[0040] The first polarization beam splitter 8 receives the quantum signal sent by the sending end Alice, separates the quantum signal into two beams of light, including signal light and local oscillator light, the first beam of signal light is sent from the first output end of the first polarization beam splitter 8 to the input end of the first optical switch 9, the output end of the first optical switch 9 is connected to the first input end of the first beam splitter 10, the output end of the first beam splitter 10 is connected to the input end of the first homodyne detector 11, the output end of the first homodyne detector 11 is connected to the first input end of the post-processing program module 17 based on multi-classification learning, the second beam of local oscillator light is sent from the second output end of the first polarization beam splitter 8 to the input end of the second beam splitter 12, the first output end of the second beam splitter 12 is connected to the input end of the second amplitude modulator 13, the output end of the second amplitude modulator 13 is connected to the second input end of the first beam splitter 10, the second output end of the second beam splitter 12 is connected to the input end of the first PIN light-emitting diode 14, the first output end of the first PIN light-emitting diode 14 is connected to the input end of the first power meter, the output end of the first power meter is connected to the second input end of the post-processing program module 17 based on multi-classification learning, the second output end of the first PIN light-emitting diode 14 is connected to the input end of the first clock circuit 16, and the output end of the first clock circuit 16 is connected to the third input end of the post-processing program module 17 based on multi-classification learning.

[0041] The post-processing program module 17 based on multi-classification learning includes a PC terminal which analyzes and processes the collected quantum signal data to generate a final security key.

[0042] The PC terminal analyzes and processes the collected quantum signal data, trains 70% of the received data, marks different types of quantum hacker attacks, and classifies and detects the remaining 30% of the data. The data after classification detection and the sending terminal Alice perform parameter estimation, reverse negotiation, and privacy amplification operations to generate a final security key.

[0043] The data analysis and processing process is specifically performed according to the following steps:

[0044] Step P1. The receiving terminal Bob collects quantum signal data and normal signal data under different quantum hacker attacks, adjusts 10% of the maximum attenuation through the first optical switch 9, measures the shot noise variance N0, and calculates the mean value of the measurement value with 90% of the minimum attenuation and variance (V y ); lo ;

[0045] V y = ηT(V A N0+ ξ) + N0+ V el ,

[0046] where η is the detection efficiency of the homodyne detector, V A is the modulation variance of the sending terminal Alice, ξ is the over noise, V el is the electrical noise, and N0 is the shot noise variance.

[0047] Step P2. The PC terminal pre-processes the data and linearly transforms the data using discrete standardization:

[0048] X * = (x-Min) / (Max-Min)

[0049] where Max is the maximum value of the data, Min is the minimum value of the data, x is the data before processing, and X * is the data after processing. Then, 70% of the processed data is trained, and 30% of the data is tested. The trained multi-class support vector machine algorithm model is saved for identification and detection of quantum hacker attack data, and the model structure is as follows:

[0050]

[0051]

[0052] Wherein, ωm is the mth sub-classifier corresponding optimization parameter, C is a penalty parameter, and l represents l sample data, The relaxation variable is represented by φ (X i The kernel function mapping to the high-dimensional feature space is represented by b m The bias term is represented by Y i The mth sub-classifier corresponding category is represented by Y i The mth sub-classifier corresponding category is represented by Y i The sample data number is i;

[0053] The trained multi-class support vector machine algorithm model saved in the PC end can be directly used for quantum hacker attack detection, so as to identify whether the continuous variable quantum key distribution system is subjected to quantum hacker attack.

[0054] The continuous variable quantum key distribution system based on multi-class learning detection and the implementation method thereof provided by the application are as follows: the post-processing program module based on multi-class learning first detects the collected signal data through a multi-class support vector machine algorithm, and once it is found that the system is subjected to quantum hacker attack, the transmission of quantum key is immediately terminated, so that the application can realize real-time quantum hacker attack detection, accurately identify and detect the quantum hacker attack existing in the system, and further enhance the actual security of the system.

[0055] The system provided by the application is as follows: the sending end Alice generates 1550nm coherent light from an external telecommunication diode at a repetition frequency of 1MHz; a first polarization beam splitter 8 divides the quantum signal into 10% signal light and 90% local oscillator light; a first optical switch 9, 10% is used for randomly setting maximum attenuation, and 90% is used for setting no attenuation; a first homodyne detector 11 is used for detecting the quantum signal, calculating the mean value and variance of the measurement value; the post-processing program module based on multi-class learning adopts a multi-class support vector machine algorithm, solves the optimal solution of the problem by solving multiple quadratic programming problems, can avoid generating new security vulnerabilities, and in addition, the multi-class support vector machine algorithm only needs to use a small part of support vectors to make hyperplane decision, and does not depend on all quantum hacker attack data.

[0056] The above is the preferred embodiment of the application, and it should be pointed out that, for ordinary skilled persons in the technical field, some improvements and refinements can be made without departing from the principles of the application, and these improvements and refinements should also be regarded as the protection scope of the application.

Claims

1. A quantum key distribution system based on multi-classification learning detection and an implementation method thereof, characterized by, It comprises a sending end Alice, a receiving end Bob and a post-processing program module based on multi-classification learning; The sending end Alice comprises a sending end pulse laser (1), a first polarizer (2), a first optical variable attenuator (3), a first amplitude modulator (4), a first phase modulator (5), a second optical variable attenuator (6) and a first polarization beam combiner (7); The receiving end Bob comprises a first polarization beam splitter (8), a first optical switch (9), a first beam splitter (10), a first homodyne detector (11), a second beam splitter (12), a second amplitude modulator (13), a first PIN light-emitting diode (14), a first power meter (15) and a first clock circuit (16); The post-processing program module based on multi-classification learning (17) comprises a PC end; The sending end Alice modulates a quantum signal in a Gaussian mode and sends the modulated quantum signal to the receiving end Bob through a quantum channel; The receiving end Bob is a quantum key receiving end Bob, uses a homodyne detector to measure the received quantum signal, uses a power meter to measure the intensity of the local oscillator light, uses a clock circuit to provide a clock signal, and finally sends the quantum signal data to the post-processing program module based on multi-classification learning through a classical quantum channel; The post-processing program module based on multi-classification learning is used for detecting the quantum signal data sent by Bob, analyzing and processing the quantum signal data through a multi-class support vector machine algorithm, and generating a final secure key according to the processing result and a key negotiation with Alice; The PC end analyzes and processes the collected quantum signal data, trains 70% of the received data as training data, marks the types of different quantum hacker attacks, performs classification detection on the remaining 30% of the data, and performs parameter estimation, reverse negotiation and privacy amplification operations on the data after classification detection and the sending end Alice to generate a final secure key. The data analysis and processing process is specifically performed according to the following steps: Step P1. The receiving end Bob collects quantum signal data under different quantum hacker attacks and normal signal data, adjusts 10% of the maximum attenuation randomly through the first optical switch (9), measures the shot noise variance N0; 90% of the minimum attenuation, calculates the mean value of the measured value and variance (V y ), the power meter measures the local oscillator light intensity I lo ; V y = ηT(V A N0+ξ)+N0+V el , where η is the detection efficiency of the homodyne detector, V A is the modulation variance of the transmitter Alice, ξ is the excess noise, V el is the electrical noise, and N0is the shot noise variance. Step P2. The PC end pre-processes the data and linearly transforms the data by using discrete standardization: X * = (x-Min) / (Max-Min) where Max is the maximum value of the data, Min is the minimum value of the data, x is the data before processing, and X * is the data after processing; Then 70% of the processed data is trained, 30% of the data is tested, the trained multi-class support vector machine algorithm model is saved, and is used for identification and detection of quantum hacker attack data. The model structure is as follows: wherein ω m is an optimization parameter corresponding to the mth sub-classifier, C is a penalty parameter, and l represents l sample data, is a slack variable, φ(X i ) represents a kernel function mapping to a high-dimensional feature space, b m represents a bias term, Y i = m represents that the Y i th sample data belongs to a category corresponding to the mth sub-classifier; i is a sample data number; Step P3. The trained multi-class support vector machine algorithm model saved in the PC end can be directly used for quantum hacker attack detection, so as to identify whether the continuous variable quantum key distribution system is subjected to quantum hacker attack. 2.The quantum key distribution system based on multi-classification learning detection and implementation method thereof according to claim 1, wherein, The sending end pulse laser (1) prepares pulse coherent light, and sends the pulse coherent light to the input end of the first polarizer (2); the first polarizer (2) separates the received quantum signal into two beams of light, including signal light and local oscillator light, wherein the first beam of signal light is sent from the first output end of the first polarizer (2) to the input end of the first optical variable attenuator (3), the output end of the first optical variable attenuator (3) is connected to the input end of the first amplitude modulator (4), the output end of the first amplitude modulator (4) is connected to the first input end of the first phase modulator (5), the output end of the first phase modulator (5) is connected to the input end of the second optical variable attenuator (6), and the output end of the second optical variable attenuator (6) is connected to the second input end of the first polarization beam combiner (7); the second beam of local oscillator light is connected from the second output end of the first polarizer (2) to the first input end of the first polarization beam combiner (7), the first polarization beam combiner (7) combines the two beams of light into one beam of light, and sends the one beam of light to the receiving end Bob through a quantum channel, and the output end of the first polarization beam combiner (7) is connected to the input end of the receiving end Bob. 3.The quantum key distribution system based on multi-classification learning detection and implementation method thereof according to claim 1, wherein, The first polarization beam splitter (8) receives the quantum signal sent by the sending end Alice, and separates the quantum signal into two beams of light, including signal light and local oscillator light, the first beam of signal light is sent from the first output end of the first polarization beam splitter (8) to the input end of the first optical switch (9), the output end of the first optical switch (9) is connected to the first input end of the first beam splitter (10), the output end of the first beam splitter (10) is connected to the input end of the first homodyne detector (11), and the output end of the first homodyne detector (11) is connected to the first input end of the post-processing program module (17) based on multi-classification learning; the second beam of local oscillator light is sent from the second output end of the first polarization beam splitter (8) to the input end of the second beam splitter (12), the first output end of the second beam splitter (12) is connected to the input end of the second amplitude modulator (13), the output end of the second amplitude modulator (13) is connected to the second input end of the first beam splitter (10), the second output end of the second beam splitter (12) is connected to the input end of the first PIN light-emitting diode (14), the first output end of the first PIN light-emitting diode (14) is connected to the input end of the first power meter, the output end of the first power meter is connected to the second input end of the post-processing program module (17) based on multi-classification learning, the second output end of the first PIN light-emitting diode (14) is connected to the input end of the first clock circuit (16), and the output end of the first clock circuit (16) is connected to the third input end of the post-processing program module (17) based on multi-classification learning.

4. The quantum key distribution system based on multi-classification learning detection and implementation method thereof according to claim 3, characterized in that, The PC end analyzes and processes the collected quantum signal data to generate a final security key. The PC end analyzes and processes the collected quantum signal data to generate a final security key.

Images