A quantum key distribution method and system based on real-time channel evaluation

Abstract

The application discloses a quantum key distribution method and system based on real-time channel evaluation, and belongs to the technical field of quantum communication, and comprises the following steps: a communication party pre-negotiates a non-orthogonal quantum probe state pair and a theoretical jump probability thereof, and establishes an attack fingerprint library; a sending end sends a probe state, a test state and a key state according to an evaluation overhead ratio in each period; a receiving end calculates a quantum state fidelity anomaly vector according to a measurement result of the probe state; the anomaly vector is matched with the attack fingerprint library, and an identification confidence of various attacks is output; the evaluation overhead ratio and a protocol deflection angle of a next period are dynamically adjusted based on the identification confidence; finally, privacy amplification is performed in combination with the identification confidence and a measured bit error rate, and a final secure key is generated. Through real-time sensing of a channel state and dynamic adjustment of system parameters, the application can effectively identify a specific attack type, adaptively balance security and key generation efficiency, and thus improves the overall performance and attack resistance of a quantum key distribution system.

Description

Technical Field

[0001] This invention belongs to the field of quantum communication technology, and in particular relates to a quantum key distribution method and system based on real-time channel evaluation. Background Technology

[0002] Existing quantum key distribution (QKD) technologies primarily detect eavesdropping by sending and measuring single-photon quantum states and utilizing the quantum no-cloning theorem. The core of this approach involves the communicating parties (Alice and Bob) calculating the bit error rate (BER) by comparing the basis vectors of a portion of the key with the measurement results. If the BER exceeds a security threshold, eavesdropping is considered to have occurred, and the current key is discarded; otherwise, the channel is considered secure, and subsequent key purification and privacy amplification are performed. These methods mainly rely on statistically averaged BER detection and typically assume that the channel state is static during key generation.

[0003] However, existing technologies have significant drawbacks. First, their channel assessment methods are relatively simplistic and outdated, relying solely on the final bit error rate to determine the presence of an attack. This fails to detect dynamic changes in the channel state in real time and cannot effectively distinguish between channel noise and specific types of quantum attacks (such as interception and retransmission, photon number splitting, etc.), resulting in low utilization of resources (such as photons and time slots). Second, when an anomaly is detected, the system typically only adopts an "all or nothing" response strategy (i.e., continue or abandon), lacking an adaptive capability to dynamically adjust system parameters (such as encoding methods and resource allocation ratios) based on real-time assessment results. Consequently, it is difficult to maximize key generation efficiency while ensuring security. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides a quantum key distribution method and system based on real-time channel evaluation, thus solving the aforementioned problems.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a quantum key distribution method based on real-time channel evaluation, comprising: S1. The sending end and the receiving end negotiate and store a set of non-orthogonal quantum probe state pairs for identifying attacks, the theoretical transition probability of each probe state pair, and an attack fingerprint database containing feature vectors of various known attack types. S2. Divide the time into periods. In each period, the transmitter sends quantum states with the current evaluation overhead ratio. The quantum states include a probe state for attack identification, a test state for basic channel evaluation, and a key state for key generation. The key state is encoded using a coding basis vector rotated according to the current protocol deflection angle. S3. The receiver measures the received probe state and calculates the quantum state fidelity anomaly vector for the current period based on the measurement results and the transmission information subsequently published by the transmitter. S4. Match the quantum state fidelity anomaly vector of the current period with the feature vector in the attack fingerprint database, and output the recognition confidence of various attacks. S5. Based on the identification confidence level output in step S4, dynamically adjust the evaluation cost ratio and protocol deflection angle for the next cycle; S6. Generate the original key using the measurement results of all key states, and combine the identification confidence and measured bit error rate output in step S4 to perform privacy amplification to generate the final security key. The evaluation overhead ratio is defined as the proportion of the number of quantum states used for channel evaluation (including attack detection and basic testing) to the total number of quantum states transmitted in a given resource allocation period. The resources include the number of photons (number of photons used for evaluation / total number of emitted photons), time slots (number of time slots used for evaluation / total number of time slots), and time slices (time used for evaluation / total time).

[0006] Based on the above technical solutions, the present invention also provides the following optional technical solutions: Further technical solution: In step S1, the set of non-orthogonal quantum detector pairs is , Each state pair satisfies , is a preset constant; the attack fingerprint database is ,in For the first Feature fingerprint vectors for each attack type.

[0007] Further technical solution: In step S3, calculate the quantum fidelity anomaly vector. , among which, the The formula for calculating each component is:

[0008] in, The first vector representing the quantum state fidelity anomaly vector One portion, , In the first The measured state within each evaluation period arrive and the reverse quantum state transition probability, , This corresponds to the theoretical quantum state transition probability. To determine the non-orthogonal quantum probe pair The presupposed sign factor of the quantum state relation, Indicates the first The lower bound of quantum fluctuations for a probe pair, where the state pair satisfies Defined when it is a real number When the state satisfies Defined when it is a pure imaginary number .

[0009] A further technical solution: The lower limit of quantum fluctuations is determined by the inherent quantum properties of the non-orthogonal quantum probe pair and the number of transmissions, and its expression is:

[0010] in, Indicates the first The theoretical nonorthogonality degree of a pair of nonorthogonal quantum detector states. Indicates the first Within the evaluation period, the first The total number of times a non-orthogonal quantum probe pair is sent and measured.

[0011] Further technical solution: Step S4 includes the following steps: S41. Calculate the current periodic quantum fidelity anomaly vector. Compared with the attack fingerprint database Feature fingerprint vectors of attack types The similarity score is calculated using the following formula:

[0012] in, Indicates the similarity score. Represents the vector dot product. Describes the L2 norm of a vector. This represents the current periodic quantum fidelity anomaly vector. This indicates an attack on the fingerprint database. Feature fingerprint vectors of attack types; S42. Based on the current and historical similarity scores, calculate the similarity score for the first... The confidence level for identifying this type of attack is calculated using the following formula:

[0013] in, Indicates the confidence level of identification. This represents the preset weight factor of the current data and , This indicates the preset number of historical periods. Indicates the first Similarity score of cycles, This indicates the similarity score; S43, Attack Status Judgment: If at least one Make If this is the case, the channel is determined to be under attack, and will be... Largest attack type As the recognition result output, The first threshold is preset; If all ,but If so, it is determined that there is an unknown anomaly in the channel, where The second threshold is preset; If all and If the channel is deemed secure and free from attack, then it is determined that the channel is secure.

[0014] Further technical solution: In step S5, the overhead ratio is dynamically adjusted for the next cycle. The rules are: If it is determined that there is no attack, then ; If it is determined to be an unknown anomaly, then ; If it is determined to be an attack type ,but ; in, To assess the cost ratio for the next cycle, The pre-set routine assessment cost ratio, This indicates the preset reinforcement evaluation cost ratio and , This indicates the preset adjustment coefficient. This indicates the attack identified in this instance. The confidence level.

[0015] Further technical solution: In step S5, when it is determined to be an attack type When updating the protocol deflection angle for the next cycle, the calculation formula is as follows:

[0016] in, This indicates the protocol deflection angle to be updated for the next cycle. Indicates the attack type for the first... Preset sensitivity weights for each probe state pair, Represents the current anomaly vector's th... One portion, Indicates the current protocol deflection angle. This indicates the preset maximum deflection angle limit. This represents the baseline deflection angle corresponding to the intensity of each unit of attack impact. Represents the quantum fidelity anomaly vector. Indicates the first Feature fingerprint vectors for each attack type.

[0017] Further technical solution: In step S6, the compression ratio used in the privacy amplification step is... Effective bit error rate perceived by attacks The decision includes:

[0018] and

[0019] in, Indicates the compression ratio. The effective bit error rate (BER) indicates the rate at which an attack is detected. This represents the measured bit error rate for the current period. For binary Shannon entropy function, This represents the preset information leakage coefficient corresponding to the attack type. Indicates the confidence level of identification. Indicates the abnormal impact coefficient. This indicates the attenuation coefficient due to bit errors.

[0020] A quantum key distribution system based on real-time channel evaluation is provided, employing the aforementioned quantum key distribution method based on real-time channel evaluation.

[0021] This invention provides a quantum key distribution method and system based on real-time channel evaluation, which has the following advantages compared with the prior art: This invention introduces non-orthogonal quantum probe state pairs and an attack fingerprint database, which enables real-time and accurate calculation of quantum state fidelity anomaly vectors, thereby identifying specific attack types or unknown anomalies, significantly improving the accuracy of channel assessment and the ability to detect complex attacks. This invention dynamically adjusts the evaluation overhead ratio, which can adaptively allocate quantum state resources for evaluation and key generation according to the channel security status. When the channel is secure, the evaluation overhead is reduced to improve the key generation rate, and when an attack is detected, the evaluation overhead is increased to strengthen monitoring, thus achieving a dynamic balance between security and efficiency.

[0022] This invention dynamically adjusts the protocol deflection angle, which can actively change the encoding basis vector when attacked, increasing the difficulty and uncertainty for attackers to eavesdrop, thereby enhancing the system's anti-attack capability at the physical layer.

[0023] This invention introduces an effective bit error rate for attack awareness in the privacy amplification step, combines the attack identification confidence with the measured bit error rate, and dynamically calculates the compression ratio, so that the final generated key can ensure its theoretical unconditional security when facing attacks of different types and intensities. Attached Figure Description

[0024] Figure 1 This is a schematic diagram of the process of the present invention. Detailed Implementation

[0025] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0026] The specific implementation of the present invention will be described in detail below with reference to specific embodiments.

[0027] Please see Figure 1 The present invention provides a quantum key distribution method based on real-time channel evaluation, comprising: S1. The sending end and the receiving end negotiate and store a set of non-orthogonal quantum probe state pairs for identifying attacks, the theoretical transition probability of each probe state pair, and an attack fingerprint database containing feature vectors of various known attack types. S2. Divide the time into periods. In each period, the transmitter sends quantum states with the current evaluation overhead ratio. The quantum states include a probe state for attack identification, a test state for basic channel evaluation, and a key state for key generation. The key state is encoded using a coding basis vector rotated according to the current protocol deflection angle. S3. The receiver measures the received probe state and calculates the quantum state fidelity anomaly vector for the current period based on the measurement results and the transmission information subsequently published by the transmitter. S4. Match the quantum state fidelity anomaly vector of the current period with the feature vector in the attack fingerprint database, and output the recognition confidence of various attacks. S5. Based on the identification confidence level output in step S4, dynamically adjust the evaluation cost ratio and protocol deflection angle for the next cycle; S6. Generate the original key using the measurement results of all key states, and combine the identification confidence and measured bit error rate output in step S4 to perform privacy amplification to generate the final security key. The evaluation overhead ratio is defined as the proportion of the number of quantum states used for channel evaluation (including attack detection and basic testing) to the total number of quantum states transmitted in a given resource allocation period. The resources include the number of photons (number of photons used for evaluation / total number of emitted photons), time slots (number of time slots used for evaluation / total number of time slots), and time slices (time used for evaluation / total time).

[0028] In this embodiment of the invention, firstly, in S1, the transmitting end and the receiving end need to negotiate and store a set of non-orthogonal quantum probe state pairs for identifying attacks, the theoretical transition probabilities of each probe state pair, and an attack fingerprint database containing feature vectors of various known attack types. The non-orthogonal quantum probe state pairs can be determined in advance through experiments or theoretical calculations; for example, a set of polarization states or phase states with specific degrees of non-orthogonality can be selected. The theoretical transition probabilities of each probe state pair can be precisely calculated using quantum mechanics principles and stored in a lookup table. The attack fingerprint database can be constructed by simulating different attack scenarios, collecting data on their impact on quantum state fidelity, and extracting representative feature vectors. For example, the impact of a retransmission attack on the transition probability of a specific probe state pair can be simulated and used as the fingerprint of that attack type.

[0029] Secondly, in S2, time is divided into periods. Within each period, the transmitter sends a quantum state with the current evaluation overhead ratio. This quantum state includes a probe state for attack identification, a test state for basic channel evaluation, and a key state for key generation. The key state is encoded using a coding basis vector rotated according to the current protocol deflection angle. The time period can be set to a fixed duration, such as per second or per millisecond. The evaluation overhead ratio can be preset to a fixed proportion, for example, 10% of the total photons are used for evaluation, and the remaining 90% for key generation. The transmitted quantum state can be generated using a single-photon source or a weakly coherent light source and encoded by a modulator. The probe state, test state, and key state can be transmitted in the same channel using time multiplexing, frequency multiplexing, or polarization multiplexing. The protocol deflection angle can be initially set to zero, indicating no additional deflection. The coding basis vector can be an orthonormal basis, such as the Z-basis (horizontal / vertical polarization) and the X-basis (diagonal polarization).

[0030] Next, in S3, the receiver measures the received probe state and calculates the quantum state fidelity anomaly vector for the current period based on the measurement results and the transmission information subsequently published by the transmitter. The receiver can use a single-photon detector and a basis selector to measure the received probe state. The transmitter publishes the type of probe state it transmitted and the corresponding basis information to the receiver through a classical channel. Based on this information, the receiver compares the measured probe state transition probabilities with the preset theoretical transition probabilities, calculates the deviation for each probe state pair, and combines these deviations into a quantum state fidelity anomaly vector. For example, if the measured transition probability of a probe state pair is significantly higher than the theoretical value, this component will show a large positive value in the anomaly vector.

[0031] Subsequently, in S4, the quantum state fidelity anomaly vector of the current cycle is matched with the feature vectors in the attack fingerprint database, outputting the identification confidence level for each type of attack. The matching process can employ vector similarity calculation methods, such as calculating the cosine similarity between the anomaly vector and each attack feature vector in the fingerprint database. By comparing similarity scores, the system can determine which known attack type the current channel anomaly is most closely related to and output the corresponding identification confidence level. For example, if the anomaly vector has a high similarity to the feature vector of a "photon number separation attack," the system will output a high confidence level for that attack type.

[0032] Furthermore, in S5, based on the identification confidence level output from step S4, the evaluation cost ratio and protocol deflection angle for the next cycle are dynamically adjusted. If the system identifies a certain attack type with high confidence, the evaluation cost ratio for the next cycle can be increased accordingly to allocate more resources for channel monitoring and attack detection. Simultaneously, the protocol deflection angle can also be adjusted according to the identified attack type. For example, if an attack is identified as sensitive to a specific coding basis, the deflection angle can be adjusted to change the key state encoding method, thereby reducing the impact of the attack. If no attack is detected, the evaluation cost ratio can be restored to a lower, normal value.

[0033] Finally, in S6, the original key is generated using the measurement results of all key states. Combined with the identification confidence level and measured bit error rate output from step S4, privacy amplification is performed to generate the final secure key. The receiving end generates the original key sequence by comparing the key state measurement results with the basis vector information published by the sending end. Simultaneously, the system calculates the measured bit error rate for the current period. During the privacy amplification stage, in addition to considering the measured bit error rate, the compression ratio of the privacy amplification is adjusted based on the identification confidence level. For example, if an attack is identified, even if the measured bit error rate is not high, the compression ratio of the privacy amplification will be adjusted to be more conservative to ensure key security under potential attacks.

[0034] The following example will provide a more detailed explanation of the above technical solution: Suppose that in a quantum key distribution communication, the sender and receiver need to establish a secure key. Before the communication begins, the two parties have negotiated and stored a set of non-orthogonal quantum probe state pairs (e.g., six polarization state pairs), the theoretical transition probability of each state pair, and an attack fingerprint database containing feature vectors of known attack types such as "truncation and retransmission attack" and "photon number separation attack" according to S1.

[0035] Communication enters S2, where time is divided into 1-second cycles. In the first cycle, the sender transmits quantum states with an initial evaluation overhead ratio (e.g., 10%). These quantum states include a probe state for detecting attacks, a test state for evaluating channel quality, and a key state for generating a key. The encoding basis vectors of the key states are encoded using an initial protocol deflection angle (e.g., 0 degrees).

[0036] During the current period, the receiver continuously measures the received probe states. At the end of the period, the transmitter publishes its transmitted probe state information through a classical channel. The receiver, according to S3, compares the measured probe state transition probabilities with the theoretical transition probabilities to calculate the quantum state fidelity anomaly vector for the current period. For example, if an abnormally high transition probability is detected for a probe state pair, this anomaly will be reflected in the corresponding component of the anomaly vector.

[0037] Subsequently, in S4, the receiver matches the calculated anomaly vector with the feature vectors of "intercept and retransmission attack" and "photon number separation attack" in the attack fingerprint database. By calculating the cosine similarity, the system obtains the recognition confidence of the two attacks. Suppose that the system calculates the recognition confidence of "intercept and retransmission attack" to be 0.8 and the confidence of "photon number separation attack" to be 0.1.

[0038] Based on the identification confidence level output in S4, the system dynamically adjusts the evaluation overhead ratio and protocol deflection angle for the next cycle in S5. Because the confidence level of the "intercept-and-retransmission attack" is high, the system increases the evaluation overhead ratio for the next cycle from 10% to 20% to enhance channel monitoring. Simultaneously, the protocol deflection angle is also fine-tuned according to the characteristics of the "intercept-and-retransmission attack," for example, by rotating it by 5 degrees to change the key state encoding method, thereby enhancing resistance to this attack.

[0039] In S6, the receiver generates the original key using the measurement results of all key states. Simultaneously, the system calculates the measured bit error rate (BER) for the current period. During privacy amplification, the system combines the 0.8 confidence level for identifying a "trap-and-retransmission attack" with the measured BER. Even if the measured BER is not high, the compression ratio of privacy amplification is adjusted more conservatively due to the presence of a high-confidence attack detection, ensuring that the final generated key maintains high security against "trap-and-retransmission attacks."

[0040] As can be seen from the above examples, the method provided in this embodiment can sense the dynamic changes in the channel state in real time and identify specific attack types. Compared with existing technologies that rely solely on statistical average bit error rate detection, this method achieves real-time and refined evaluation of the channel state by periodically sending probe states and calculating quantum state fidelity anomaly vectors in real time. This real-time capability enables the system to detect potential attacks or anomalies earlier.

[0041] Furthermore, this method effectively distinguishes between channel noise and specific types of quantum attacks by constructing and matching an attack fingerprint database. In the example above, the system can identify "intercepted retransmission attacks" rather than simply judging it as "channel anomalies." This ability to identify attack types enables the system to adopt more targeted countermeasures, rather than the simplistic "all or nothing" approach used in existing technologies.

[0042] Furthermore, this method can dynamically adjust system parameters, including the evaluation overhead ratio and protocol deflection angle, based on real-time evaluation results. In the example, the system adaptively increases the evaluation overhead ratio and adjusts the protocol deflection angle according to the identified attack type and confidence level. This adaptive capability allows the system to optimize resource allocation and key generation efficiency while ensuring security. For example, the evaluation overhead ratio can be reduced to improve key generation efficiency when no attack is detected, while the overhead ratio can be increased to enhance security when an attack is detected.

[0043] Preferably, in step S1, the set of non-orthogonal quantum detector pairs is: , Each state pair satisfies , The value is a preset constant; the attack fingerprint database is... ,in For the first Feature fingerprint vectors for each attack type.

[0044] The set of non-orthogonal quantum probe states represents a group of quantum state pairs used to detect channel anomalies and attack behaviors. Each state pair consists of two non-orthogonal quantum states. and Composition. Non-orthogonality is a key property of quantum information, meaning that these two states cannot be perfectly distinguished, but their overlap can be used to detect perturbations in the channel. In the set This represents the number of probe state pairs, each optimized to counter different types of attacks or channel anomalies. These probe state pairs are sent into the quantum channel. After measurement at the receiver, anomalies in the channel can be detected by comparing the actual measurement results with theoretical expectations, thus inferring the presence of an attack. As a possible implementation, polarization-coded single-photon states can be used; for example, a state pair can be a horizontally polarized state. and diagonal polarization state Another state pair could be a vertically polarized state. and anti-angle polarization state By selecting different combinations of polarization directions, multiple sets of non-orthogonal state pairs can be constructed. Alternatively, phase-encoded single-photon states can be used; for example, a state pair can be a state with a phase of 0. and phase are state By changing the phase difference, state pairs with different degrees of nonorthogonality can be constructed.

[0045] Each state pair satisfies ,in, Indicates the first The square of the overlap (or non-orthogonality) between the two states in a pair of non-orthogonal quantum detector states is a pre-defined constant. This constant reflects the difficulty of distinguishing the two quantum states and forms the basis for calculating the theoretical transition probability. This ensures that each probe state pair has deterministic and known nonorthogonality, which is crucial for subsequent calculations of theoretical jump probabilities and quantum fluctuation lower limits. It provides a quantitative basis for accurately assessing channel states and identifying attacks, through fixed... This can standardize the response characteristics of different probe state pairs, making the construction of attack fingerprints more consistent and reliable. As one possible implementation, for polarization coding, the angle between two polarization states can be precisely controlled to set... For example, if the included angle is chosen as... ,but Furthermore, for phase encoding, the phase difference between two phase states can be precisely controlled. To set For example, if the phase difference is chosen to be ,but .

[0046] The attack fingerprint database is a pre-built database that stores... Each known attack type has its own characteristic fingerprint vector. Each characteristic fingerprint vector... It is A dimensional vector, whose components Represents the first Under this type of attack, the first The fingerprint database identifies specific anomalous patterns or responses exhibited by non-orthogonal quantum probe pairs. This fingerprint database serves as a benchmark for attack identification. When the system detects a channel anomaly, it compares the measured anomaly vector with the feature vectors in the database to identify the type of attack that may be currently being attacked. One possible implementation is to simulate various known attacks (such as intercept-and-retransmit attacks, photon number splitting attacks, and side-channel attacks) in a controlled environment and record the quantum fidelity anomaly patterns of probe pairs under each attack, thereby constructing the corresponding feature fingerprint vector. Alternatively, the impact of different attacks on probe pairs can be predicted based on theoretical models or simulations, and feature fingerprint vectors can be generated accordingly. For example, for a certain attack, it may primarily affect probe pairs with specific polarization directions, while having a smaller impact on other probe pairs, thus forming a unique fingerprint.

[0047] In the aforementioned quantum key distribution method based on real-time channel evaluation, to improve the accuracy and robustness of attack identification, this application provides specific definitions for the non-orthogonal quantum probe state pairs and attack fingerprint database mentioned in step S1. This is achieved by explicitly defining the non-orthogonal quantum probe state pairs as a set. And the square of the nonorthogonality degree of each pair of states is limited to a preset constant. This ensures that the physical properties of the probe state pairs are controllable and known. This explicit definition allows the system to predict quantum state behavior in the absence of attacks based on accurate theoretical models, thus providing a reliable benchmark for subsequent calculations of quantum state fidelity anomaly vectors. Meanwhile, the attack fingerprint database is specifically defined as... Each of them It is The feature fingerprint vector of dimension , its components Corresponding to the Type of attack against the first The influence of each probe state pair. This structured fingerprint database works closely with the explicitly defined probe state pairs described above. When the receiver measures the probe state and calculates the quantum state fidelity anomaly vector for the current period... Subsequently, the dimension of the anomaly vector is matched with the dimension of the feature vectors in the fingerprint database. The system can then accurately compare the measured anomaly vector with various pre-stored attack feature fingerprint vectors in the database. This synergistic effect enables the system not only to detect anomalies in the channel, but more importantly, to identify specific attack types by matching them with known attack fingerprints. For example, if an attack primarily affects a specific probe state pair, the measured anomaly vector will show a significant deviation in that component, and the feature vectors of the corresponding attack type in the fingerprint database will also have high weights in that component. Through this refined matching mechanism, the system can effectively distinguish different attack types, avoiding the limitations of relying solely on fuzzy anomaly judgments. This provides a more accurate basis for subsequently dynamically adjusting the evaluation overhead ratio and protocol deflection angle, significantly improving the security of the quantum key distribution process.

[0048] The following is a specific example to illustrate this. As a specific implementation, the set of non-orthogonal quantum detector states can consist of three sets of polarization state pairs, namely... =3. The first set of detector state pairs can be horizontally polarized states. and diagonal polarization state Its nonorthogonality degree squared It can be set to 1 / 2. The second set of detector state pairs can be vertically polarized states. and anti-angle polarization state Its nonorthogonality degree squared It can also be set to 1 / 2. The third set of detector state pairs can be left-handed circularly polarized states. and right-handed circularly polarized state Its nonorthogonality degree squared It can be set to 0. The attack fingerprint database can contain feature fingerprint vectors for three typical attack types, namely... =3. For example, the first type of attack It could be a "photon number separation attack," whose characteristic fingerprint vector It exhibits a similar, moderate effect on all probe state pairs, for example The second type of attack It could be a "polarization rotation attack," whose characteristic fingerprint vector It may primarily affect polarization-dependent probe state pairs (such as the first and second groups), while having a smaller impact on circularly polarized state pairs, for example... The third type of attack It could be "increased channel loss", its characteristic fingerprint vector This may manifest as a uniform, small effect on all probe state pairs, for example... The component values ​​of these characteristic fingerprint vectors can be obtained through experimental simulation or theoretical calculation and stored in an attack fingerprint database. When the system detects an anomaly, it can compare the measured anomaly vector with these preset fingerprint vectors to identify the most likely attack type.

[0049] The above technical solution clearly defines the set of non-orthogonal quantum probe state pairs and their non-orthogonality degree for attack identification, and structures the attack fingerprint database, enabling the quantum key distribution system to achieve higher accuracy and robustness in attack identification. Specifically, the preset probe state pairs and their constant non-orthogonality degree provide an accurate benchmark for quantifying channel anomalies, avoiding misjudgments caused by ambiguity in the probe state definition. Simultaneously, the structured attack fingerprint database can finely match the measured quantum state fidelity anomaly vector with the feature fingerprint vectors of various known attack types, thereby accurately distinguishing different attack types, rather than simply detecting "an attack." This precise attack identification capability provides a more reliable basis for subsequent dynamic adjustment of the evaluation overhead ratio and protocol deflection angle, significantly improving the adaptive security protection capability of the quantum key distribution process and effectively reducing security risks caused by inaccurate attack identification.

[0050] Preferably, in step S3, the quantum fidelity anomaly vector is calculated. , among which, the The formula for calculating each component is:

[0051] in, The first vector representing the quantum state fidelity anomaly vector One portion, , In the first The measured state within each evaluation period arrive and the reverse quantum state transition probability, , This corresponds to the theoretical quantum state transition probability. To determine the non-orthogonal quantum probe pair The presupposed sign factor of the quantum state relation, Indicates the first The lower bound of quantum fluctuations for a pair of probes, where the state pair satisfies... Defined when it is a real number When the state satisfies Defined when it is a pure imaginary number ; The lower limit of quantum fluctuations is determined by the inherent quantum properties of the non-orthogonal quantum probe pair and the number of transmissions, and its expression is:

[0052] in, Indicates the first The theoretical nonorthogonality degree of a pair of nonorthogonal quantum detector states. Indicates the first Within the evaluation period, the first The total number of times a non-orthogonal quantum probe pair is sent and measured.

[0053] Quantum fidelity anomaly vector Used to quantify in the current evaluation cycle Within the quantum channel, the deviation between the actual behavior of the probe state and its theoretically predicted behavior. Each of its components... This corresponds to a specific non-orthogonal quantum probe state pair, reflecting the degree of anomaly experienced by that probe state pair in the channel. By constructing such a multi-dimensional vector, minute changes in the channel state can be captured comprehensively and meticulously, providing a quantitative basis for subsequent attack identification and channel state assessment. The calculation formula for each component comprehensively considers the measured transition probability, the theoretical transition probability, the preset sign factor, and the lower limit of quantum fluctuations, aiming to accurately measure the degree of anomaly of a specific probe state pair. This calculation method can effectively filter out noise caused by quantum fluctuations and normalize the results according to the characteristics of the probe state pair, thereby making the degree of anomaly of different probe state pairs comparable and improving the sensitivity and accuracy of anomaly detection.

[0054] in, and Indicates the first Within the evaluation cycle, for the first A pair of nonorthogonal quantum probe states, from state Measured state The measured transition probability, and the state transition probability. Measured state The measured transition probabilities. These measured probabilities can be obtained by measuring and statistically analyzing the received probe states at the receiver. For example, the receiver can use a single-photon detector array to detect the received photons and statistically analyze the probability of the transition when the transmitted state is... At that time, what percentage of photons were measured as Alternatively, the receiving end can use quantum state tomography to reconstruct the received probe state and then calculate the corresponding transition probability. and This represents the theoretical quantum state transition probability corresponding to the measured transition probability. These theoretical values ​​are pre-calculated based on the principles of quantum mechanics under ideal, attack-free, and noise-free channel conditions, and are typically negotiated and stored by the transmitter and receiver at the start of the protocol. For example, they can be determined based on the inner product of the probe state pairs. The theoretical transition probability can be calculated using the square of the modulus, or obtained after modification based on the channel model.

[0055] The sign factor is based on non-orthogonal quantum probe pairs. The quantum state relation is presupposed, and its function is to provide the correct sign direction for the anomalous vector components to distinguish different types of channel perturbations or attacks. When the inner product of the state pairs... When it is a real number, define When it is a purely imaginary number, it is defined as follows: This design allows anomaly vectors to reflect specific trends in the phase or amplitude of quantum states in the channel, thus helping to identify attack types more precisely. The parameter represents the first The lower bound of quantum fluctuations for each probe pair is used to normalize the deviation between the measured and theoretical jump probabilities. Quantum fluctuations are an inherent randomness of quantum systems, existing even in attack-free ideal channels. By introducing a lower bound, deviations exceeding the normal quantum fluctuation range can be identified as anomalies, thereby improving the sensitivity of attack detection and reducing false alarms. This value is determined by the inherent quantum properties of non-orthogonal quantum probe pairs (such as non-orthogonality). (and the total number of times the probe state pair was sent and measured during the evaluation period.) A joint decision. The parameter represents the first The theoretical nonorthogonality degree of a pair of nonorthogonal quantum detector states is specifically: It quantifies the distinguishability between two non-orthogonal quantum states, is an inherent property of the probe state pair, and is determined and stored during the protocol negotiation phase. For example, for polarization-encoded quantum states, two states with polarization angles differing by a certain angle can be selected as a non-orthogonal state pair, and their non-orthogonality degree... It can be obtained by calculating the square of its inner product. The parameter represents the first time. Within the evaluation period, the first The total number of times a non-orthogonal quantum probe pair is transmitted and measured. This count is dynamic, depending on the resources allocated to that probe pair by the transmitter in each cycle. For example, the transmitter could maintain a counter to record the number of transmissions for each probe pair; the receiver could record the number of measurements for each probe pair. These two counts are summed at the end of the cycle to obtain the final count. .

[0056] The scheme in this application refines the calculation of the quantum fidelity anomaly vector in step S3 at the receiving end. Instead of simply comparing the measured transition probability with the theoretical transition probability, it calculates the corresponding quantum fidelity anomaly vector component for each non-orthogonal quantum detector state pair. The calculation process first obtains the current evaluation period. Inside, from the state arrive And the measured transition probability in the opposite direction and At the same time, the theoretical jump probability is utilized through pre-negotiation. and The deviation between the measured and theoretical values ​​is calculated. To ensure that this deviation accurately reflects the nature of channel anomalies, a preset symbol factor is introduced. This factor is determined based on the quantum state relationship of the probe state pair, thus assigning the correct directionality to the anomalous component. More importantly, this deviation is normalized by dividing by the first... The lower limit of quantum fluctuations for a single probe pair This lower bound for quantum fluctuations is based on the theoretical nonorthogonality of the probe state pairs. and the total number of times the probe state pair is transmitted and measured within the current period. This is derived through dynamic calculation. This normalization process effectively filters out inherent quantum fluctuation noise in the channel, ensuring that only significant deviations exceeding the normal fluctuation range are identified as anomalies, thereby improving the sensitivity and anti-interference capability of anomaly detection. Ultimately, the anomalous components of all probe state pairs collectively constitute the quantum fidelity anomaly vector. This vector can comprehensively and quantitatively characterize the current channel's abnormal state, providing a precise data foundation for subsequent attack identification and channel state adjustment. This refined method of anomaly vector calculation enables the system to more accurately distinguish between normal fluctuations and malicious attacks in the channel, thereby improving the security of the entire quantum key distribution protocol.

[0057] The following is a concrete example to illustrate this. Suppose that the sender and receiver negotiate a set of non-orthogonal quantum probe state pairs. For example, in a polarization-coded quantum key distribution system, a horizontal polarization state can be selected. and diagonal polarization state Forming a probe state pair .at this time, = Since it is a real number, the corresponding sign factor is... It can be set to +1. In the... Within each evaluation period, the sending end sent... Sub-detection state pair The receiver measures the received photons and counts the values ​​from them. Send and measure The number of times, and from Send and measure The number of times is used to calculate the measured transition probability. and At the same time, based on the preset theoretical transition probability... and Calculate the bias. The theoretical nonorthogonality of this probe state pair. , using Value and The lower bound of quantum fluctuations can be calculated. Finally, substitute these values ​​into the formula. This allows us to obtain the quantum fidelity anomaly vector component of the probe state pair. By applying all... By repeating this process with each probe state pair, a complete quantum fidelity anomaly vector can be constructed. .

[0058] Through the above technical solution, the receiver can assess the channel state in a more refined and quantitative way. This is achieved by introducing a symbol factor. This allows for the differentiation of the influence direction of different types of channel perturbations on the quantum state, enabling the anomaly vector to more accurately reflect the nature of the attack. Simultaneously, by dividing the deviation value by the lower bound of quantum fluctuations... Normalization effectively suppresses the interference of inherent random fluctuations in quantum channels on anomaly detection, significantly improving the system's sensitivity to weak attacks or anomalies and reducing the false alarm rate. This precise anomaly vector calculation method provides more reliable and accurate input for subsequent attack type identification and dynamic adjustment of channel parameters, thereby enhancing the security and robustness of the quantum key distribution protocol in complex channel environments.

[0059] Preferably, step S4 includes the following steps: S41. Calculate the current periodic quantum fidelity anomaly vector. Compared with the attack fingerprint database Feature fingerprint vectors of attack types The similarity score is calculated using the following formula:

[0060] in, Indicates the similarity score. Represents the vector dot product. Describes the L2 norm of a vector. This represents the current periodic quantum fidelity anomaly vector. This indicates an attack on the fingerprint database. Feature fingerprint vectors of attack types; S42. Based on the current and historical similarity scores, calculate the similarity score for the first... The confidence level for identifying this type of attack is calculated using the following formula:

[0061] in, Indicates the confidence level of identification. This represents the preset weight factor of the current data and , This indicates the preset number of historical periods. Indicates the first Similarity score of cycles, This indicates the similarity score; S43, Attack Status Judgment: If at least one Make If this is the case, the channel is determined to be under attack, and will be... Largest attack type As the recognition result output, The first threshold is preset; If all ,but If so, it is determined that there is an unknown anomaly in the channel, where, The second threshold is preset; If all and If the channel is deemed secure and free from attack, then it is determined that the channel is secure.

[0062] Among them, similarity score Used to quantify the current channel anomaly state and pre-stored specific attack types. The degree of matching between feature patterns is defined as follows: if a channel is attacked, the resulting anomalous patterns should be highly similar to the feature fingerprint of that attack. Similarity scores are typically calculated using cosine similarity, which effectively measures the directional proximity of two vectors, unaffected by their magnitude, thus providing a standardized matching metric. The vector dot product is a fundamental vector operation used to calculate the sum of the products of corresponding components of two vectors. In similarity calculations, the dot product reflects the magnitude of the projections of two vectors onto each other in the same direction; the more aligned the directions of the two vectors, the larger their dot product. The L2 norm of a vector, also known as the Euclidean norm, represents the length or magnitude of a vector. It is obtained by calculating the square root of the sum of the squares of the vector's components. In similarity calculations, the L2 norm is used to normalize the vectors, ensuring that the similarity score reflects only the directional similarity, not their magnitude. Identification confidence. This indicates that the system is suffering from a specific type of attack related to the current channel state. The degree of certainty is determined by considering not only the similarity score of the current period but also incorporating information from historical periods, thus providing a more stable and reliable attack identification indicator. By combining historical data, it can effectively smooth out instantaneous fluctuations, reduce false positives, and improve the detection capability of persistent attacks. The preset weighting factors for the current data... It is a constant between 0 and 1, used to balance the relative importance of the current period similarity score and the historical period similarity score when calculating the recognition confidence. A higher [score / value] indicates a higher [score / value]. A higher value means the system focuses more on the latest channel state information, resulting in a faster response to attacks; while a lower value... The value indicates that the system relies more on historical trends and has better robustness to short-term fluctuations. This weighting factor can be adjusted according to the actual application scenario to meet the requirements for response speed and stability. Preset number of historical periods. The time window length for historical similarity scores considered when calculating identification confidence is defined. By incorporating information from multiple historical periods, the evolution trend of attack behavior can be effectively captured, and the detection capability for intermittent or slowly evolving attacks can be enhanced. For example, it can be set according to the dynamic change rate of the channel environment or the typical duration of the attack. The value of. The first. Periodic similarity score This refers to the current cycle. The previous Within a historical period, the current channel anomaly vector and attack type The similarity scores between the feature fingerprint vectors are used. These historical scores, together with the current scores, form the basis for calculating the recognition confidence, providing the system with information on attack patterns over time. A preset first threshold is used. It is a pre-set threshold used to determine whether the identification confidence level has reached a level sufficient to determine that the channel has suffered a specific attack. When the identification confidence level for any attack type... When this threshold is exceeded, the system considers the channel to be under that specific attack. This threshold can be calibrated based on the system's tolerance for false alarms and false negatives. A preset second threshold... It is a pre-set threshold value used to determine whether the overall amplitude of the quantum fidelity anomaly vector exceeds the normal fluctuation range, even if no specific attack type (i.e., all) is identified. If the L2 norm of the outlier vector If this threshold is exceeded, it indicates that there is some kind of unexpected anomaly in the channel, which may be an unknown attack or environmental interference. This threshold can be set according to the normal noise level of the channel and the system's sensitivity to unknown anomalies.

[0063] The scheme in this application measures and calculates the quantum state fidelity anomaly vector for the current period at the receiving end. Subsequently, in order to accurately identify the type of threat faced by the channel, firstly, the anomaly vector is calculated. Compared with each known attack type pre-stored in the attack fingerprint database Feature fingerprint vector Similarity score between This similarity score uses cosine similarity quantification, which effectively reflects the directional matching degree between the current channel anomaly pattern and known attack patterns. Furthermore, to improve the stability and accuracy of identification, the system does not rely solely on the similarity score of the current period, but further combines the current score with other similarities. and the past Similarity score of each historical period The weighted average was used to calculate the effect of each attack type. Recognition confidence Among them, weighting factors This allows the system to adjust the emphasis on current and historical data based on actual needs. Finally, these are used to identify confidence levels. The overall magnitude of the quantum fidelity anomaly vector The system then determines the attack status. Specifically, if any attack type exists... Recognition confidence Exceeding the preset first threshold If so, the channel is determined to be under attack, and the attack type with the highest confidence level is identified. If the confidence level for all known attack types does not exceed However, the overall magnitude of the anomaly vector Exceeding the preset second threshold This indicates an unknown anomaly in the channel; only when all identification confidence levels are below a certain threshold will the channel exhibit this anomaly. Furthermore, the overall magnitude of the anomaly vector is also lower than Only when the channel is deemed secure and attack-free is the channel determined to be in a secure state. Through this multi-dimensional, time-series analysis and decision-making mechanism, the scheme of this application can effectively distinguish between normal fluctuations, known attacks, and unknown anomalies in the channel, thereby providing accurate threat assessment for subsequent dynamic adjustments and key generation.

[0064] As a specific implementation method, in the quantum key distribution process, when the receiver completes the measurement of the probe state and calculates the current period... Quantum fidelity anomaly vector The vector is then fed into a processing module for attack identification. This module first extracts feature fingerprint vectors from a pre-established attack fingerprint database for different attack types (e.g., intercept-and-retransmission attacks, photon number splitting attacks, side-channel attacks, etc.). For each type of attack The processing module will calculate and The cosine similarity between them is used to obtain a similarity score. For example, if and If the directions of the intercepted retransmission attack are highly consistent, then... It will approach 1. Then, the processing module will utilize the currently calculated... And combined with the storage history period (e.g., recent) Similarity score (5 cycles) , ..., Each attack type was calculated using a weighted average formula. Recognition confidence For example, it can be set =0.7 indicates that the current data accounts for 70% of the weight, and historical data accounts for 30% of the weight. Finally, the processing module will determine the weight based on the preset first threshold. (e.g., 0.8) and the second threshold (For example, 0.5) to make a judgment, if found Reaching 0.9, exceeding The system will immediately determine that the channel has been subjected to an intercept and retransmission attack. If all All below ,but Reaching 0.6, exceeding If the system detects an unknown anomaly, it will issue an alarm. If all indicators are below the threshold, the channel is considered safe.

[0065] Through the above technical solution, this application overcomes the limitations of attack identification based solely on a single-period anomaly vector. By introducing the calculation of similarity scores, quantitative matching of channel anomaly patterns with known attack characteristic patterns is achieved, thereby improving the accuracy of attack identification. Furthermore, combining current and historical similarity scores to calculate identification confidence effectively smooths out instantaneous fluctuations, enhances the system's robustness against persistent or evolving attacks, and reduces the false alarm rate. In addition, by setting multi-level decision thresholds, this application's scheme can not only identify specific known attack types but also effectively detect unknown channel anomalies, avoiding missed detections due to an incomplete attack fingerprint database. This refined attack identification and state decision mechanism provides a more accurate basis for subsequent evaluation of overhead ratios and dynamic adjustment of protocol deflection angles, thereby significantly improving the security and adaptability of the quantum key distribution system.

[0066] Preferably, in step S5, the dynamic adjustment of the cost ratio for the next cycle is evaluated. The rules are: If it is determined that there is no attack, then ; If it is determined to be an unknown anomaly, then ; If it is determined to be an attack type ,but ; in, To assess the cost ratio for the next cycle, The pre-set routine assessment cost ratio, This indicates the preset reinforcement evaluation cost ratio and , This indicates the preset adjustment coefficient. This indicates the attack identified in this instance. Confidence level; In step S5, when it is determined to be an attack type When updating the protocol deflection angle for the next cycle, the calculation formula is as follows:

[0067] in, This indicates the protocol deflection angle to be updated for the next cycle. Indicates the attack type for the first... Preset sensitivity weights for each probe state pair, Represents the current anomaly vector's th... One portion, Indicates the current protocol deflection angle. This indicates the preset maximum deflection angle limit. This represents the baseline deflection angle corresponding to the intensity of each unit of attack impact. Represents the quantum fidelity anomaly vector. Indicates the first Feature fingerprint vectors for each attack type.

[0068] Dynamically adjust the assessment cost ratio for the next cycle The rules aim to intelligently allocate resources based on the channel security status (no attack, unknown anomaly, specific attack type) to balance security and efficiency. This can be achieved by configuring a control module at the transmitting end, which queries a preset evaluation cost ratio lookup table based on the received channel state decision result and adjusts the ratio of probe state to test state when transmitting quantum state in the next cycle accordingly; or it can be achieved through an adaptive algorithm, which predicts and sets the optimal evaluation cost ratio based on feedback from historical evaluation cost ratios and the current channel state through a machine learning model. The evaluation overhead ratio is used when the channel is determined to be secure and attack-free. Its purpose is to ensure that the system can perform necessary channel monitoring with low resource consumption under normal and threat-free channel conditions, thereby optimizing the efficiency of key distribution. This value can be determined through experimental testing or simulation under ideal or near-ideal channel conditions, and can be a fixed value that meets the basic channel evaluation requirements while minimizing resource overhead. It can also be assigned by expert experience. This is a higher evaluation overhead ratio used by the system to enhance security monitoring when the channel is determined to have an unknown anomaly or be under specific attack. Its function is to increase the number of quantum states used for detection and testing, thereby improving the ability to perceive potential threats and the speed of response. This value can be set as an empirical value, and it should be significantly higher than [previous value]. This is to ensure that more channel information can be captured in abnormal situations, but not so much that it would lead to low key generation efficiency. It is a preset adjustment factor used to adjust the evaluation cost ratio as a function of attack confidence when a specific attack type is identified. The sensitivity to change allows the system to adjust resource allocation more precisely according to the severity of the attack. This coefficient can be manually set by the system designer based on the risk assessment of different attack types and the system response requirements. It can be determined by expert experience or by fitting historical data.

[0069] Update the protocol deflection angle for the next cycle. The calculation formula aims to proactively resist or evade specific types of attacks by altering the encoding method of the quantum state, thereby enhancing the robustness of the protocol. This formula calculates a new deflection angle by combining the current deflection angle, the sensitivity weight of the attack type to the probe state pair, the current anomaly vector components, and the intensity of the attack's impact. This can be executed by the processor at the transmitting end, adjusting the quantum state encoding device based on the calculation result. It is an attack type For the The preset sensitivity weights of each probe pair reflect the specific attack type. For the The influence or sensitivity of a non-orthogonal quantum probe pair is used to guide the system to prioritize the abnormal information reflected by the probe pairs that are most sensitive to the currently identified attack type when adjusting the protocol deflection angle. These weights can be assigned by expert experience or set by theoretical analysis or experimental simulation of different attack types. It is the maximum range that the protocol deflection angle can be adjusted. Its function is to prevent the deflection angle from being adjusted too large, which would lead to a decrease in protocol performance or exceed the physical implementation capability, thereby maintaining the stability and effectiveness of the protocol. This limit can be set according to the physical implementation limitations of the quantum key distribution system and the theoretical security analysis of the protocol. It is the baseline deflection angle corresponding to each unit of attack impact intensity, defining the basic proportional relationship between the attack impact intensity and the protocol deflection angle adjustment range. Its function is to quantify the driving force of attack intensity on deflection angle adjustment. This parameter can be assigned by expert experience or observed through simulation or experimentation under different attack intensities, and a suitable proportional coefficient can be set empirically accordingly.

[0070] When the system is based on the quantum state fidelity anomaly vector The attack fingerprint database identifies the channel status (no attack, unknown anomaly, or specific attack type). At that time, a dynamic adjustment mechanism will be triggered. This is relevant for evaluating the cost-benefit ratio. If no attack is detected, the system will revert the overhead ratio to a lower, normal level. To conserve resources. If an unknown anomaly is detected, the system will immediately switch to a higher enhancement evaluation overhead ratio. To gather more information and enhance monitoring. If a specific attack is identified. The system not only adopts enhanced evaluation cost ratio It will also be based on the attack confidence level By adjusting the coefficients Further increasing the assessment overhead means that the higher the confidence level for a specific attack, the greater the investment of assessment resources, thereby enabling a more thorough investigation and deployment of countermeasures against that attack. Simultaneously, when an attack type is determined... At that time, the protocol deflection angle It will also be adjusted. The direction of the adjustment items is changed from... The decision indicates whether the attack caused a deviation in the sensitive probe state. The adjustment range is related to the attack's impact. Proportional and limited by the maximum deflection angle. The constraints ensure that the system can proactively adjust its coding basis vectors to counteract the effects of identified attacks, with the adjustment magnitude proportional to the perceived strength of the attack, while preventing excessive or unstable adjustments. This integrated approach enables the quantum key distribution system not only to detect threats but also to intelligently and adaptively reconfigure its operating parameters (resource allocation and coding strategy) in real time, thereby optimizing the trade-off between security and efficiency under different channel conditions and attack scenarios.

[0071] The following is a concrete example to illustrate this. Suppose that during the operation of a quantum key distribution system, its normal evaluation overhead is higher than... Set at 5% to enhance the evaluation cost ratio Set to 20%, adjustment coefficient Set to 0.5, maximum deflection angle limit Set as Reference deflection angle Set to 0.1.

[0072] If the system determines that the channel is secure and free from attack within a given period, the evaluation overhead for the next period is reduced compared to... This will be set to 5%. This means the system will dedicate its main resources to key generation while maintaining necessary channel monitoring.

[0073] If the system detects an unknown anomaly in the channel, such as the norm of the quantum fidelity anomaly vector... Exceeded the preset second threshold However, if no known attack type is matched, the evaluation cost for the next cycle is higher than... This will immediately increase the rate to 20%. This allows the system to allocate more resources to detection and testing in order to uncover the root cause of the anomaly.

[0074] If the system identifies a specific type of attack For example, photon number splitting (PNS) attacks, and their identification confidence. The cost is 0.8. At this point, the evaluation cost for the next cycle is higher than... Will be calculated as =0.28, or 28%. Meanwhile, the protocol deflection angle... Adjustments will also be made, assuming the current protocol deflection angle It is 0, and depends on the attack type. Sensitivity weights for probe state pairs and the current anomaly vector Calculated +1, attack impact strength If the value is 0.5, then the adjustment amount for the deflection angle is... =0.05, therefore, the protocol deflection angle for the next cycle It will be updated to 0 + 0.05 = 0.05. In this way, the system can dynamically adjust resource allocation and coding strategies according to the type and intensity of the attack to enhance its resistance to specific attacks.

[0075] By dynamically adjusting the evaluation overhead ratio and protocol deflection angle based on real-time channel evaluation results, this application enables adaptive optimization of the quantum key distribution system under different channel security states. When the channel is secure, the system can reduce evaluation overhead and improve key generation efficiency. When an unknown anomaly or specific attack is detected, the system can rapidly increase evaluation resource investment and proactively adjust protocol parameters to resist the attack, thereby significantly improving the robustness, security, and resource utilization efficiency of the quantum key distribution process.

[0076] Preferably, in step S6, the compression ratio used in the privacy amplification step is... Effective bit error rate perceived by attacks The decision includes:

[0077] and

[0078] in, Indicates the compression ratio. The effective bit error rate (BER) indicates the rate at which an attack is detected. This represents the measured bit error rate for the current period. For binary Shannon entropy function, This represents the preset information leakage coefficient corresponding to the attack type. Indicates the confidence level of identification. Indicates the abnormal impact coefficient. This indicates the attenuation coefficient due to bit errors.

[0079] The compression ratio used in the privacy enhancement step Effective bit error rate perceived by attacks Privacy amplification is a key post-processing step in quantum key distribution protocols, designed to improve key security by reducing redundant information in the original key and eliminating any information that an eavesdropper could obtain. Compression ratio This is a crucial parameter used to determine the final key length during the privacy amplification process. It directly affects the length and security of the final key. Traditionally, the compression ratio is mainly calculated based on the measured bit error rate. In this application, the determination of the compression ratio introduces the "attack-aware effective bit error rate". This means that when calculating the compression ratio, not only are bit errors caused by random noise in the channel taken into account, but also the impact of potential attacks or anomalies on key security, thus making the privacy amplification process more robust and secure.

[0080] The effective bit error rate of attack detection It is a comprehensive indicator, based on traditional measured bit error rate. Building upon this foundation, the additional risk of information leakage due to channel anomalies or attacks is added. The aim is to quantify the potential threat of attacks or anomalies to key security and incorporate it into privacy amplification considerations. By introducing this concept, even the measured bit error rate... The effective bit error rate is low, but if there are significant channel anomalies or signs of attack, the effective bit error rate will be lower. This will also increase accordingly, prompting Privacy Amplifier to adopt a more conservative strategy to ensure key security.

[0081] The The formula specifically defines the effective bit error rate for attack detection. The calculation method. Among them, This represents the measured bit error rate for the current period, reflecting errors caused by random noise and losses in the channel. It is the anomaly impact coefficient, used to adjust the contribution of channel anomalies to the effective bit error rate. Its value can be preset based on experience or simulation results. For example, it can be set to a positive decimal to balance the impact of anomalies on the bit error rate. It is the L2 norm of the quantum fidelity anomaly vector, which quantifies the degree to which the current periodic channel state deviates from the theoretical ideal state. The larger the value, the more significant the anomaly. It is a preset second threshold used to distinguish between normal fluctuations and significant anomalies. Only when the degree of anomaly exceeds this threshold will it be included in the calculation of the effective bit error rate. This ensures that a gain is applied to $E_{t}$ only when the anomaly exceeds a threshold, thus avoiding an overreaction to normal channel fluctuations.

[0082] The The formula gives the privacy amplification compression ratio. The specific calculation method. Among them, It is the binary Shannon entropy function, which represents the effective bit error rate. Below is the maximum amount of information that an eavesdropper might obtain. The basic compression ratio that constitutes privacy enhancement is the third term in the formula. attack type Compensation for any additional information leaks caused. Is it related to attack type The corresponding preset information leakage coefficient reflects the specific attack type. The potential threat level to key security can be set based on theoretical models or experimental data of different attacks, or determined by expert experience. attack type The higher the confidence level of the attack, the greater the likelihood that the attack exists. It is the bit error rate attenuation coefficient, used to adjust the measured bit error rate. The impact on compensation items for information leakage caused by attacks, for example, when At higher levels, compensation for information leakage may need to be appropriately attenuated, as channel noise itself already causes significant information loss. This can be achieved by taking... This ensures that Privacy Amplification can perform the most conservative compression on the attack types with the highest current confidence level, thereby maximizing protection against potential eavesdropping.

[0083] The binary Shannon entropy function It is a fundamental concept in information theory used to quantify the uncertainty or information content of a binary random variable. In quantum key distribution, it is often used to estimate the amount of information about the key that an eavesdropper might obtain from a public channel at a given bit error rate.

[0084] The preset information leakage coefficient corresponding to the attack type This is a pre-defined parameter for each known attack type, used to quantify the potential information leakage caused by that attack type. Different attack strategies (such as intercept-and-retransmit attacks, photon number splitting attacks, etc.) pose different threats to key security, therefore requiring different... Values ​​are used to reflect this difference. These coefficients can be determined through expert experience assignment, theoretical analysis, simulation, or experimental data to ensure that privacy amplification provides sufficient security margins against specific attacks.

[0085] The abnormal influence coefficient It is an adjustable parameter used to control channel anomalies (determined by the L2 norm of the quantum fidelity anomaly vector). (Reflects) on the effective bit error rate The intensity of the impact. Larger. A value of 100% means that even slight anomalies can significantly increase the effective bit error rate, leading to more stringent privacy amplification; a smaller value... A higher value indicates a higher tolerance for anomalies. The setting of this coefficient needs to balance key security and key generation efficiency.

[0086] The bit error affects the attenuation coefficient. It is used to adjust the measured bit error rate. The parameter affecting the compensation term for attack information leakage is related to the bit error rate of the channel itself. A higher value may indicate poor channel quality, which may increase the difficulty for attackers to obtain additional information through specific attacks. Therefore, the compensation term for information leakage can be appropriately reduced. The introduction of this coefficient allows the privacy amplification process to adapt more flexibly to different channel conditions and avoids over-compressing the key when the channel quality is extremely poor.

[0087] The solution in this application introduces an effective bit error rate that is aware of attacks. The privacy amplification process was optimized by adding an additional information leakage compensation term based on attack identification confidence. Within each evaluation period, the privacy amplification process is first optimized based on the measured bit error rate of the current period. L2 norm of quantum fidelity anomaly vector To calculate Specifically, when the channel anomaly level (by...) (Measurement) Exceeds the preset second threshold At that time, the measured bit error rate Will be affected by abnormal coefficient Magnification, thus obtaining a higher . This comprehensively reflects the impact of channel noise and potential anomalies on key security. Subsequently, the privacy amplification compression ratio... The calculation is not only based on Binary Shannon entropy function Furthermore, it subtracted one that was related to the identified attack type. and its confidence level The relevant compensation item is based on a preset information leakage coefficient. And the impact of bit error rate on attenuation coefficient This mechanism quantifies the potential information leakage caused by a specific attack and compensates for the highest leakage risk among all attack types. This ensures that privacy amplification dynamically adapts to channel conditions and attack threats. Even when the measured bit error rate is low but there are signs of an attack, it can more rigorously compress the key by increasing the effective bit error rate and adding compensation terms, thereby effectively reducing the probability of eavesdroppers obtaining information and ensuring the security of the final key.

[0088] As a specific implementation method, it is assumed that within a certain evaluation period, the quantum key distribution system detects the measured bit error rate. This is within the normal range, for example, 2%. However, through measurement and analysis of the probe state, the L2 norm of the calculated quantum fidelity anomaly vector is... Significantly higher than the preset second threshold This indicates the presence of some anomaly in the channel. Simultaneously, the attack identification module, based on matching results from the attack fingerprint database, identifies a specific type of attack. The probability is relatively high, and its identification confidence level is [high]. It is also relatively high. In this case, even Low effective bit error rate for attack detection Will because Exceed This is amplified, for example, from 2% to 3%. Next, the privacy amplification compression ratio is calculated. At that time, in addition to the improvements made after this upgrade In addition to calculating the Shannon entropy term, an extra compensation term is subtracted. The magnitude of this compensation term is determined by the recognition confidence level. and attack types Corresponding preset information leakage coefficient And the attenuation coefficient due to bit error rate Joint decision. For example, if It is an attack known to cause significant information leakage, and If the value is high, then the compensation term will be large, leading to a higher final compression ratio. Further reduction results in shorter but more secure keys. Conversely, if Not exceeding And all Both are relatively low, then Approaching Furthermore, the compensation term will be very small, and privacy amplification will be mainly based on the measured bit error rate, thereby maximizing the key rate while ensuring security.

[0089] Through the above technical solution, this application can dynamically adjust the privacy amplification strategy based on real-time channel assessment results and attack identification. By introducing an effective bit error rate (BER) for attack awareness, even when the measured BER is low but channel anomalies or attack signs exist, the actual risk of information leakage can be assessed more accurately. Furthermore, by combining the confidence level for identifying specific attack types, the privacy amplification compression ratio can be further refined to ensure that sufficiently conservative privacy amplification measures are taken when facing different attack threats, thereby effectively reducing the probability of eavesdroppers obtaining key information. This not only significantly improves the security of the final key and avoids key leakage due to underestimating attack risks, but also avoids over-compression when there are no attacks or minor anomalies, thus optimizing key generation efficiency and achieving a balance between security and efficiency.

[0090] A quantum key distribution system based on real-time channel evaluation is provided, employing the aforementioned quantum key distribution method based on real-time channel evaluation.

[0091] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus.

[0092] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A quantum key distribution method based on real-time channel evaluation, characterized in that, include: S1. The sending end and the receiving end negotiate and store a set of non-orthogonal quantum probe state pairs for identifying attacks, the theoretical transition probability of each probe state pair, and an attack fingerprint database containing feature vectors of various known attack types. S2. Divide the time into periods. In each period, the transmitter sends quantum states with the current evaluation overhead ratio. The quantum states include a probe state for attack identification, a test state for basic channel evaluation, and a key state for key generation. The key state is encoded using a coding basis vector rotated according to the current protocol deflection angle. S3. The receiver measures the received probe state and calculates the quantum state fidelity anomaly vector for the current period based on the measurement results and the transmission information subsequently published by the transmitter. S4. Match the quantum state fidelity anomaly vector of the current period with the feature vector in the attack fingerprint database, and output the recognition confidence of various attacks. S5. Based on the identification confidence level output in step S4, dynamically adjust the evaluation cost ratio and protocol deflection angle for the next cycle; S6. Generate the original key using the measurement results of all key states, and combine the identification confidence and measured bit error rate output in step S4 to perform privacy amplification to generate the final security key. The evaluation overhead ratio is defined as the ratio of the number of quantum states used for channel evaluation to the total number of quantum states transmitted in a given resource allocation period, where the resources include the number of photons, time slots, and time slices.

2. The quantum key distribution method based on real-time channel evaluation according to claim 1, characterized in that, In step S1, the set of non-orthogonal quantum detector state pairs is: ， Where each state pair satisfies , is a preset constant; the attack fingerprint database is ,in For the first Feature fingerprint vectors for each attack type.

3. The quantum key distribution method based on real-time channel evaluation according to claim 2, characterized in that, In step S3, the quantum fidelity anomaly vector is calculated. , among which, the The formula for calculating each component is: in, The first vector representing the quantum state fidelity anomaly vector One portion, , In the first The measured state within each evaluation period arrive and the reverse quantum state transition probability, , This corresponds to the theoretical quantum state transition probability. To determine the non-orthogonal quantum probe pair The presupposed sign factor of the quantum state relation, Indicates the first The lower bound of quantum fluctuations for a probe pair, where the state pair satisfies Defined when it is a real number When the state satisfies Defined when it is a pure imaginary number .

4. The quantum key distribution method based on real-time channel evaluation according to claim 3, characterized in that, The lower limit of quantum fluctuations is determined by the inherent quantum properties of the non-orthogonal quantum probe pair and the number of transmissions, and its expression is: in, Indicates the first The theoretical nonorthogonality degree of a pair of nonorthogonal quantum detector states. Indicates the first Within the evaluation period, the first The total number of times a non-orthogonal quantum probe pair is sent and measured.

5. The quantum key distribution method based on real-time channel evaluation according to claim 2, characterized in that, Step S4 includes the following steps: S41. Calculate the current periodic quantum fidelity anomaly vector. Compared with the attack fingerprint database Feature fingerprint vectors of attack types The similarity score is calculated using the following formula: in, Indicates the similarity score. Represents the vector dot product. Describes the L2 norm of a vector. This represents the current periodic quantum fidelity anomaly vector. This indicates an attack on the fingerprint database. Feature fingerprint vectors of attack types; S42. Based on the current and historical similarity scores, calculate the similarity score for the first... The confidence level for identifying this type of attack is calculated using the following formula: in, Indicates the confidence level of identification. This represents the preset weight factor of the current data and , This indicates the preset number of historical periods. Indicates the first Similarity score of cycles, This indicates the similarity score; S43, Attack Status Judgment: If at least one Make If this is the case, the channel is determined to be under attack, and will be... Largest attack type As the recognition result output, The first threshold is preset; If all ,but If so, it is determined that there is an unknown anomaly in the channel, where The second threshold is preset; If all and If the channel is deemed secure and free from attack, then it is determined that the channel is secure.

6. The quantum key distribution method based on real-time channel evaluation according to claim 2, characterized in that, In step S5, the overhead ratio is dynamically adjusted for the next cycle. The rules are: If it is determined that there is no attack, then ; If it is determined to be an unknown anomaly, then ; If it is determined to be an attack type ,but ; in, To assess the cost ratio for the next cycle, The pre-set routine assessment cost ratio, This indicates the preset reinforcement evaluation cost ratio and , This indicates the preset adjustment coefficient. This indicates the attack identified in this instance. The confidence level.

7. The quantum key distribution method based on real-time channel evaluation according to claim 2, characterized in that, In step S5, when it is determined to be an attack type When updating the protocol deflection angle for the next cycle, the calculation formula is as follows: in, This indicates the protocol deflection angle to be updated for the next cycle. Indicates the attack type for the first... Preset sensitivity weights for each probe state pair, Represents the current anomaly vector's th... One portion, Indicates the current protocol deflection angle. This indicates the preset maximum deflection angle limit. This represents the baseline deflection angle corresponding to the intensity of each unit of attack impact. Represents the quantum fidelity anomaly vector. Indicates the first Feature fingerprint vectors for each attack type.

8. The quantum key distribution method based on real-time channel evaluation according to claim 2, characterized in that, In step S6, the compression ratio used in the privacy amplification step is... Effective bit error rate perceived by attacks The decision includes: and in, Indicates the compression ratio. The effective bit error rate (BER) indicates the rate at which an attack is detected. This represents the measured bit error rate for the current period. For binary Shannon entropy function, This represents the preset information leakage coefficient corresponding to the attack type. Indicates the confidence level of identification. Indicates the abnormal impact coefficient. This indicates the attenuation coefficient due to bit errors.

9. A quantum key distribution system based on real-time channel evaluation, characterized in that, The quantum key distribution method based on real-time channel evaluation as described in any one of claims 1-8 is adopted.

Images