A quantum measurement link clock synchronization calibration method

Abstract

The application relates to the technical field of clock synchronization and calibration, and discloses a quantum measurement link clock synchronization and calibration method, which is used for solving the problem that the traditional method lacks online separation calibration of link bidirectional propagation asymmetric drift and equipment internal time delay drift; the method first uniformly arranges bidirectional time transfer records, quantum correlation measurement records and link operation state records, and performs source layering on total time deviation to distinguish real clock deviation, link bidirectional propagation asymmetric drift and equipment internal time delay drift; then, the method combines the connection relationship of adjacent calibration periods, version updating, abnormality keeping, recovery observation and reconstruction processing to complete calibration control, avoids directly writing abnormal drift and apparent deviation into the formal calibration process, and thus realizes long-term stable maintenance of the consistency of a quantum measurement link real time and frequency reference.

Description

Technical Field

[0001] This invention relates to the field of clock synchronization and calibration technology, specifically a method for clock synchronization and calibration of a quantum measurement link. Background Technology

[0002] Clock synchronization calibration in quantum measurement links is a fundamental step in ensuring the consistency of quantum event marking, correlated measurements, measurement window alignment, and time-frequency references. In existing technologies, clock deviations at both ends of the link are typically calculated based on bidirectional time transfer results or quantum correlated measurement results, and the clock at the synchronization end is adjusted accordingly to achieve link time alignment. For example, the published invention patent application CN109547144B discloses a clock synchronization system and method based on quantum entanglement. This scheme obtains the HOM (Homo Depression) extremum through entangled photons, bidirectional transmission paths, and electrically controllable optical delay lines, thereby determining the time difference and achieving clock synchronization. Another example is the published invention patent application CN111698038B, which discloses a cascadeable quantum time synchronization method. This scheme records the photon arrival times of adjacent stations and calculates the clock difference based on the bidirectional time synchronization principle to achieve time synchronization of long-distance cascaded links. This demonstrates that existing technologies already possess the technical foundation for measuring and calibrating clock deviations in quantum links. However, current technology directly corrects the clock based on the measured total time deviation, only obtaining synchronization results and failing to distinguish and calibrate the error sources that cause this total time deviation online. In reality, during the continuous operation of the quantum measurement link, in addition to the actual clock deviation, there are factors such as bidirectional propagation asymmetric drift, internal photoelectric conversion delay drift, detection response drift, and additional delay changes caused by environmental variations. Multiple factors collectively affect the measured time deviation. Current technology cannot distinguish which of these deviations are actual clock deviations, link drift, or equipment drift; it can only directly calibrate the clock deviation based on the apparent deviation formed by the superposition of drift errors. If the system continuously outputs synchronization results under conditions of continuous link operation and environmental changes, but the synchronization results are of poor authenticity and stability, the consistency of the time and frequency reference of the quantum measurement link will gradually decrease, affecting the accuracy of the time assignment of subsequent quantum measurement results and the long-term operational reliability. Therefore, how to perform online separation and calibration for bidirectional propagation asymmetric drift and internal time delay drift of the link, and update after distinguishing between the real clock deviation and drift error, is a technical problem that urgently needs to be solved. Only by solving the above technical problems can the consistency of the real time and frequency reference of the quantum measurement link be maintained in the long term, so as to improve the accuracy, stability and engineering reliability of clock synchronization calibration. Summary of the Invention

[0003] To address the shortcomings of existing technologies, this invention provides a quantum measurement link clock synchronization calibration method, which solves the problem of the lack of online separation and calibration of bidirectional propagation asymmetric drift and internal device delay drift in traditional methods.

[0004] To achieve the above objectives, the present invention provides the following technical solution: A method for clock synchronization calibration of a quantum measurement link includes: S1. Collect bidirectional time transfer records, quantum correlation measurement records, and link operation status records of the quantum measurement link during the current calibration cycle, and perform time-series organization based on a unified time reference; S2. Based on the correspondence of the bidirectional observation records, the total time deviation within the current calibration cycle is classified and judged in layers to generate candidate quantities of real clock deviation, bidirectional propagation asymmetric drift of the link, and internal delay drift of the device. S3. Based on the deviation succession relationship between adjacent calibration cycles and the current operating status, perform calibration and determination on the candidate quantities of real clock deviation, link bidirectional propagation asymmetric drift, and device internal delay drift. S4. Perform clock synchronization calibration update based on the calibration judgment result, and record the bidirectional propagation asymmetric drift of the link and the internal delay drift of the device; S5. In case of abnormal link operation, sudden change in environmental status, or bidirectional observation mismatch, maintain the current calibration version and re-enter the next calibration cycle after recovery.

[0005] Furthermore, the bidirectional time transfer records, quantum correlation measurement records, and link operation status records of the quantum measurement link are collected during the current calibration cycle, including: The current calibration period is established according to a fixed-length window, and multiple source records from the quantum measurement link are uniformly accessed, periodically merged, and field-identified. Multi-source records must include at least a record number, record type, time stamp, direction stamp, node identifier, period identifier, and validity stamp. Record types must be distinguished into at least forward time transfer records, reverse time transfer records, quantum correlation hit records, node running status records, and environmental status records. Integrity verification and validity identification are performed on each type of record.

[0006] Furthermore, time series processing is performed based on a unified time base, including: The data sets merged into the current calibration period are time-mapped according to a unified time base, and then sorted by time markers after being merged by period identifier; Establish candidate correspondence indexes for forward time transfer records, reverse time transfer records, and quantum correlation measurement records that are close in time location; Based on the periodic set and boundary buffer, the boundary records are assigned affiliation, and abnormal records are cached and diverted to form a periodic validity mark. The results of the data processing are used to create a structured dataset, which includes bidirectional time transfer sequences, quantum correlation measurement sequences, link state sequences, anomaly record sequences, and period validity markers.

[0007] Furthermore, based on the correspondence between the two-way observation records, the total time deviation within the current calibration period is stratified and classified, including: The bidirectional time transfer sequence is divided into a forward record set and a reverse record set according to the direction marking, and the valid records in the forward record set, reverse record set and quantum correlation measurement sequence are merged according to a unified time reference and period identifier; Observation correspondence groups are established based on temporal proximity, and a total time deviation characterization value is generated for each observation correspondence group; Using the total time deviation as input, the system sequentially performs directional symmetry discrimination, node consistency discrimination, and intra-period continuity discrimination.

[0008] Furthermore, candidate quantities for real clock skew, link bidirectional propagation asymmetric drift, and device internal delay drift are generated, including: The deviation components after hierarchical discrimination are organized into candidate quantities to form a candidate quantity structure with real clock deviation candidate quantity, link bidirectional propagation asymmetric drift candidate quantity, device internal delay drift candidate quantity, local disturbance mark, candidate quantity source mark, candidate quantity confidence level and current calibration cycle discrimination mark; Local disturbance markers correspond to deviation components that do not form a continuous relationship, while candidate quantity confidence levels and current calibration cycle discrimination markers are used to characterize the usage conditions of candidate quantities.

[0009] Furthermore, based on the deviation succession relationship between adjacent calibration cycles and the current operating status, calibration determination is performed on the candidate quantities of real clock deviation, link bidirectional propagation asymmetric drift, and internal device delay drift, including: A continuous judgment window is established around the current calibration cycle. The candidate quantities of the real clock deviation are judged in accordance with the direction of change, the magnitude of change, and the degree of continuous stability. The candidate quantities of bidirectional propagation asymmetric drift of the link and the candidate quantities of internal delay drift of the device are constrained and judged in combination with the link operation status. Then, according to the state machine rules, the current calibration cycle is transferred to the pending update state, the observation and storage state, or the frozen and held state, and the cases of record interruption, high boundary buffer participation, and missing state records are judged to be downgraded, temporarily suspended, or rebuilt.

[0010] Furthermore, based on the calibration determination result, a clock synchronization calibration update is performed, including: Based on the current calibration cycle's determination status, a calibration update record is generated, and the actual clock deviation determination value is written into the candidate version area; Candidate versions are assigned version statuses such as pending, officially effective, under observation, and frozen. Candidate versions are then processed for official status, extended confirmation, or invalidation according to continuous confirmation conditions, single update magnitude constraints, and status transition rules.

[0011] Furthermore, the asymmetric drift during bidirectional propagation of the link and the internal delay drift of the device are recorded, including: The asymmetric drift of bidirectional propagation of the link and the internal delay drift of the device are recorded separately to form a drift record area that is independent of the formal calibration version area and is associated with the period number, version number and status flag. Observation records are generated in the suspended state, and frozen records are generated in the frozen state. The drift history for each calibration cycle is organized and updated based on drift type, drift magnitude, confidence level, associated operating status, observation reason, freeze reason, number of observations, and freeze start marker.

[0012] Furthermore, in the event of link malfunction, sudden environmental changes, or bidirectional observation mismatch, the current calibration version is maintained, and the system re-enters the next calibration cycle after recovery, including: After identifying link operation abnormalities, sudden changes in environmental state, or bidirectional observation mismatch, the current state is switched to the freeze hold state according to the rule that a severe abnormality triggers a freeze once or a weak abnormality triggers a freeze continuously. The current formal calibration version is maintained, the candidate version is stopped, and the abnormality type, freeze start information, and drift change information are recorded. Once the cause of the anomaly disappears and the pairing conditions are met again in both directions, the current state will be switched to the recovery observation state, and the recovery determination will be made based on the continuous observation conditions and the succession relationship. If no official calibration version is available, the current state is switched to the reconstruction state, and a new official calibration version is generated based on the continuous reconstruction conditions.

[0013] Compared with existing technologies, the present invention provides a quantum measurement link clock synchronization calibration method, which has the following beneficial effects: 1. This invention improves the accuracy, stability, and reliability of clock synchronization calibration of the quantum measurement link by uniformly organizing bidirectional time transfer records, quantum correlation measurement records, and link operation status records during continuous operation of the quantum measurement link. Based on this, it performs source identification, inheritance, versioning updates, anomaly retention, and recovery processing on the total time deviation. The actual clock deviation, bidirectional propagation asymmetric drift, and internal device delay drift are distinguished and calibrated separately under the same calibration process. It eliminates the need to directly write the drift errors caused by changes in link propagation conditions, internal device response, and environmental conditions into the clock calibration process after summing the apparent deviations. This improves the long-term maintenance of the consistency of the true time-frequency reference of the quantum measurement link, thus solving the problem of the lack of online separation calibration of bidirectional propagation asymmetric drift and internal device delay drift in traditional methods.

[0014] 2. This invention, by setting up a continuous judgment, freeze-hold, recovery observation and reconstruction processing mechanism, enables the quantum measurement link to perform clock calibration not directly based on the observation results during the abnormal period when the recording is interrupted, the environment changes suddenly, or the two-way observation mismatch occurs. Instead, it uses the current formal calibration version as the calibration benchmark, and re-observes and judges based on the continuity relationship after the abnormality is resolved. This ensures that the short-term deviation or distortion deviation during the abnormal period will not enter the formal calibration process, thereby improving the anti-interference capability and recovery stability capability of the quantum measurement link under complex operating conditions, and ensuring the continuous stability and controllability of the clock synchronization calibration process. Attached Figure Description

[0015] Figure 1 This is a schematic diagram of a quantum measurement link clock synchronization calibration method according to the present invention. Detailed Implementation

[0016] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0017] Example 1: Figure 1 A method for clock synchronization calibration of a quantum measurement link is presented, including: S1. Collect bidirectional time transfer records, quantum correlation measurement records, and link operation status records of the quantum measurement link during the current calibration cycle, and perform time-series processing based on a unified time reference. The specific implementation is as follows: After the quantum measurement link enters the current calibration cycle, the original records participating in clock synchronization calibration within this cycle are uniformly accessed, merged, and time-mapped. The current calibration cycle is a data processing time window set for clock synchronization determination and calibration update, used to incorporate bidirectional time transfer behavior, quantum correlation measurement behavior, and changes in link operation status within the same time range into the same processing unit. The current calibration cycle is established using a fixed-length window, preferably between 0.5 seconds and 2 seconds, for example, 0.5 seconds, 1 second, or 2 seconds. In operating environments with shorter link lengths, higher quantum event arrival rates, and relatively stable environmental changes, this is appropriate. In general, the current calibration period can be selected as 0.5 seconds to 1 second; in operating scenarios with longer link lengths, lower quantum event arrival rates, or slower environmental changes, the current calibration period can be selected as 1 second to 2 seconds. The basis for determining this range is that a single calibration period should cover the minimum number of effective records required to establish subsequent bidirectional observation correspondences, and should be less than the duration of the main local changes in bidirectional propagation asymmetric drift and internal device delay drift. This ensures that sufficient observation samples can be formed within the same period without masking local drift characteristics due to an excessively long period. Therefore, the value range of 0.5 seconds to 2 seconds has a clear engineering constraint basis. Upon entering the current calibration cycle, bidirectional time transfer records, quantum correlation measurement records, and link operation status records are collected. Bidirectional time transfer records represent the arrival times of forward and backward propagation of the link; quantum correlation measurement records represent the initiation time, hit time, and correspondence of quantum correlation events; and link operation status records represent the changes in node status, link status, and environmental status within the current calibration cycle. To facilitate centralized and unified processing of data from different sources, the input data adopts a unified record structure, including at least the following fields: record number, record type, time stamp, direction stamp, node identifier, cycle identifier, and validity stamp. Record numbers can be assigned according to... The generation order is set to determine the record types generated under the same time marker. Record types include at least forward time propagation records, reverse time propagation records, quantum correlation hit records, node running status records, and environment status records. Direction markers distinguish forward propagation, reverse propagation, and non-directional status records. Node identifiers distinguish between the sender and receiver, period identifiers assign records to the corresponding current calibration period, and validity markers indicate whether a record has passed the acquisition integrity check. With a unified record structure, subsequent processing directly calls data from the period identifier, direction marker, and time marker, ensuring consistent data format, clear interfaces, and ease of implementation. Time stamps are converted and expressed using a unified time reference. This unified time reference places records from different sources within the current calibration period onto a single reference time axis, ensuring that forward records, reverse records, quantum-correlated records, and state records have comparable, pairable, and sortable time bases within the same period. Time stamps are expressed in nanosecond or picosecond increments. When the variation in link propagation delay and the device's time resolution are in the nanosecond range, the time stamp value is in the nanosecond range. When minute variations in link propagation delay and internal device delay drift place higher demands on calibration decisions, the time stamp value is in the picosecond range. The time granularity is set based on the fact that the accuracy of the time stamp needs to match the variation in link propagation delay and the resolution required for calibration decisions. Low accuracy masks small drifts, while high accuracy imposes a processing burden. Therefore, nanosecond and picosecond levels are common and practical time representation methods in this field. To ensure that the acquired data can be directly incorporated into subsequent time series processing, consistent entry conditions are applied to different records. Bidirectional time transfer records must include at least fields such as direction marker, node identifier, and time marker; quantum correlation measurement records must include at least fields such as hit time and correlation identifier; link operation status records must include at least fields such as acquisition time and status type; node operation status can be normal, restricted, or abnormal, and environmental status can be temperature status, decay status, vibration status, maintenance switching status, etc.; link operation status records, as independent records, are unified with bidirectional observation records on the same time base. Different time transfer records are mixed together, establishing a correspondence between deviation changes and node status changes, link status changes, and environmental status changes on the same time axis. This provides a preliminary data foundation for the hierarchical discrimination of real clock deviation, asymmetric drift in bidirectional link propagation, and internal device delay drift. After data collection, all records are organized into a time series based on a unified time reference to form the current calibration cycle. During time series organization, all records of the same current calibration cycle are merged and arranged by cycle identifier and then arranged in time stamp order. Forward time transfer records, reverse time transfer records, and quantum correlation measurement records that are adjacent in time position are used to establish candidate correspondence indexes according to preset time proximity rules. Node operating status and environmental status are mapped onto the same time axis as accompanying states. The organized input data of the current calibration cycle forms structured basic data for characterization of time sequence, observation correspondence, and state correlation. To ensure the reproducibility of timing organization, data temporary storage within the current calibration period employs a combination of period sets and boundary buffers. Period sets store valid records within the current period, while boundary buffers temporarily store records occurring near the current period's boundary. The boundary buffer length is preferably selected to be 5% to 10% of the current calibration period length. For example, when the calibration period is 0.5 seconds, the boundary buffer can be 25 to 50 milliseconds; when the calibration period is 1 second, the boundary buffer can be 50 to 100 milliseconds. This range is determined because the lower limit of the boundary buffer should be greater than the normal cross-window propagation jitter range of forward and reverse records and quantum correlation records near the period boundary, to avoid real cross-window records. Omissions; the upper limit of the boundary buffer should be less than the effective resolution time of the main deviation change characteristics within the current calibration period to avoid over-including records from adjacent periods that do not belong to the current period, thus affecting the accuracy of the observation correspondence in the current period; records in the boundary buffer do not directly participate in the formal sequence generation of the current period, but are assigned to a period after the main records of the current period are organized, based on the time point, direction attribute, and temporal proximity; if the boundary record has a higher temporal proximity priority to the main record of the current period, it is assigned to the current period; if the boundary record is closer to the main record of the adjacent period, it is assigned to the adjacent period; through the above processing, observation mismatches at the period boundary can be reduced, and the stability of subsequent candidate quantity generation can be improved; During the time series processing, abnormal records are triaged. Abnormal records include at least those with missing time stamps, unclear direction stamps, conflicting node identifiers, incorrect period identifiers, and records where valid stamps have failed. Abnormal records do not participate in the current period's time series construction and are first moved to an abnormal cache. Records in the abnormal cache include at least the original record number, abnormal type, and original time position for subsequent tracing and recovery. If the number of abnormal records is less than 2% to 5% of the total number of records in the current period (e.g., less than 3%), abnormal records are removed, and the current period is still treated as a normal, valid period for subsequent processing. If the number of abnormal records reaches or exceeds this range (e.g., reaches...), the abnormal records are removed. When the percentage of abnormal records reaches or exceeds 3%, the current period is marked as a weakly valid period. This range is determined based on the number of valid records in a single period, the minimum sample ratio required for subsequent candidate quantity generation, and the degree of influence of abnormal records on the confidence level of candidate quantities. When the proportion of abnormal records is at a low level, boundary buffer compensation and valid record redundancy can usually cover the impact of abnormalities. When the proportion of abnormal records reaches or exceeds the above range, the bidirectional observation correspondence and state mapping relationship in the current period will decrease significantly, making it difficult to support the generation of high-confidence candidate quantities. A weakly valid period does not mean that the current period is completely unusable, but rather that the period is only suitable as an observational input and is not suitable as a high-confidence basis for formal calibration updates. To balance online processing requirements with resource consumption, a sliding retention method is adopted to manage the period set and boundary buffer. After the current period is processed, the boundary buffers of the current period, the previous period, and the next period are retained, and all original records are not retained for a long time. If operating resources allow, the structured datasets of the most recent 5 to 20 calibration periods can be retained. If operating resources are limited, the structured datasets of the current period and the previous period can be retained. The determination of the above range mainly considers the processing requirements of boundary recovery, adjacent period continuation judgment, and online calibration continuity analysis, while controlling long-term storage overhead. The time series processing of the current period is preferably completed within one period length after the end of the current period. For example, when the calibration period is 1 second, the time series processing of the current period is completed within the next 1 second, so that the next stage processes the current period data in an online manner and avoids switching to offline processing. After the above operations, a structured dataset for the current calibration cycle can be obtained. The structured data includes at least fields such as bidirectional time transfer sequence, quantum correlation measurement sequence, link state sequence, anomaly record sequence, and cycle validity marker.

[0018] S2. Based on the correspondence of the bidirectional observation records, the total time deviation within the current calibration period is classified into layers to generate candidate quantities for real clock deviation, bidirectional propagation asymmetric drift of the link, and internal delay drift of the equipment. The specific implementation is as follows: After the above structured dataset is formed, the total time deviation within the current calibration cycle is classified into three categories: candidate quantity of real clock deviation, candidate quantity of asymmetric drift in bidirectional propagation of the link, and candidate quantity of internal delay drift of the device, which are used as input for calibration judgment. First, the bidirectional time transfer sequence is divided into a forward record set and a reverse record set based on the direction markers. The forward record set corresponds to the time transfer behavior from the transmitting direction to the receiving direction, and the reverse record set corresponds to the time transfer behavior from the receiving direction to the transmitting direction. Then, based on a unified time reference and period identifier, the forward record set, reverse record set, and valid records in the quantum correlation measurement sequence are merged within the same period and formed into observation correspondence groups according to their temporal proximity. The observation correspondence group is a combination of records within the same calibration period that meet the preset time proximity condition and can reflect the same or the same batch of link propagation relationships. The time proximity condition is constrained by a pairing time window, which is preferably set to 10 nanoseconds to 500 nanoseconds. When the link propagation jitter range is small and the time marker resolution is high, the time proximity condition is further defined. When the force is strong and the quantum event arrival rate is high, the pairing time window can be selected from 10 nanoseconds to 50 nanoseconds; when the link propagation jitter range is large, the link is long, or the time stamp resolution is relatively low, the pairing time window can be selected from 50 nanoseconds to 500 nanoseconds. The basis for setting this range is that the lower limit of the pairing time window should cover the normal propagation jitter range and the typical arrival discrete range of quantum correlation records, so as to avoid misjudging observations that should be paired as mismatches, and the upper limit of the pairing time window should suppress irrelevant records from being mistakenly included in the same observation corresponding group, so as to reduce incorrect pairings; to ensure the reproducibility of parameters, the pairing time window is preferably pre-tuned according to the propagation jitter range under no-load conditions under stable operating conditions after link deployment, and adjusted in combination with the operation records in the frozen state or reconstruction state. After establishing the observation correspondence groups, a total time deviation characterization value is generated for each observation correspondence group within the current calibration period. The total time deviation characterization value is used to characterize the comprehensive deviation of the corresponding observation group under a unified time reference and serves as a unified input for subsequent source stratification. The number of effective pairing groups within the current calibration period is preferably 30 to 200. When the link is stable, the quantum event density is high, and the bidirectional recording integrity is good, the number of effective pairing groups can be selected as 100 to 200. When the link is long, the quantum event density is low, or the observation sample formation speed is slow, the number of effective pairing groups can be selected as 30 to 100. The basis for setting this range is that the number of effective pairing groups should meet the basic requirements of sample size for directional symmetry discrimination, node consistency discrimination, and periodic continuity discrimination, while avoiding excessively high sample thresholds that lead to excessive rejection of effective periods. If the number of effective pairing groups within the current calibration period is lower than the lower limit of this range, high-confidence candidate quantities are not formed, only observation-level candidate quantities are formed, and a low-confidence label is assigned in the candidate quantity structure. After forming the total time deviation characterization value, directional symmetry discrimination, node consistency discrimination, and intra-cycle continuity discrimination are performed sequentially. Directional symmetry discrimination is used to identify the deviation component in the total time deviation caused by changes in the bidirectional propagation conditions of the link. In specific processing, the direction and magnitude of the deviation changes of each observation group in the forward and reverse record sets are compared. If the forward and reverse deviations change in opposite directions, or if the difference between the two continues to increase within the current calibration period, the corresponding deviation is preferentially classified into the candidate quantity of bidirectional propagation asymmetric drift of the link. The continuous increase can be constrained by the average change of the deviation difference between adjacent observation groups. The preferred value for the bidirectional asymmetric discrimination threshold is 10% to 20% higher than the preset baseline. The preset baseline can be selected as the average value of the forward and reverse deviation differences within the most recent valid period, or the average value of the forward and reverse deviation differences within the most recent several stable periods. When the most recent valid cycle is in a stable operating state, this cycle can be directly used as the preset baseline. When the fluctuation of a single cycle is large, the average value of the most recent stable cycles can be used as the preset baseline. The basis for setting this threshold range is that it is necessary to distinguish between the natural fluctuation range of the positive and negative deviation difference and the deviation amplification range caused by the change of the bidirectional propagation conditions of the link, so as to avoid misjudging normal jitter as asymmetric drift of the bidirectional propagation of the link and to ensure that asymmetric changes can be identified in a timely manner. If the positive deviation and the negative deviation change slowly in the same direction in the current calibration cycle and the change amplitude does not exceed the bidirectional asymmetry discrimination threshold, the corresponding deviation is retained as the priority source of the candidate quantity of the real clock deviation. Through directional symmetry discrimination, the apparent deviation caused by the change of the bidirectional propagation conditions of the link can be preferentially removed, and the candidate quantity of the real clock deviation can be prevented from directly absorbing the bidirectional link asymmetric change component. After determining directional symmetry, node consistency is determined by combining the node operating status records in the link state sequence. This checks whether the deviation changes within the current calibration period correspond continuously to changes in a node's detection response, gating state, time stamp stability, or photoelectric conversion state. Continuous observation windows can be evenly divided according to the current calibration period, with the preferred number being 2 to 10. When the current calibration period is short, the number of continuous observation windows can be selected as 2 to 4; when the current calibration period is long, the number of continuous observation windows can be selected as 4 to 10. The above range is based on the fact that the number of continuous observation windows needs to cover the typical duration of node state changes and also needs to correspond to the current calibration period. The length and local continuity discrimination requirements are matched to identify the local correspondence between deviation changes and node state changes within the current calibration cycle. If a deviation change occurs synchronously with a node state change within a continuous observation window, and this correspondence holds true for 2 to 5 consecutive observation windows, the corresponding deviation is included in the device's internal delay drift candidate quantity. If a node state record is missing, or the state change is not continuous, the corresponding deviation is not directly included in the device's internal delay drift candidate quantity, but is retained as a component to be discriminated and its confidence level is reduced. Through node consistency discrimination, the impact of the device's internal delay path on the total time deviation can be distinguished from the actual clock deviation for subsequent calibration judgment. After determining directional symmetry and node consistency, the remaining deviations are then subjected to intra-cycle continuity determination. Intra-cycle continuity determination involves judging whether the deviation changes have continuous continuity within the current calibration cycle. That is, according to the chronological order of the corresponding observation groups within the current calibration cycle, it is determined whether the deviation changes continue in the same direction or change steadily and slowly among multiple observation points. If it only occurs in a local short period of time, or is only caused by a few nearby observation points, and does not occur continuously in most other observation points, then the priority of this deviation as a candidate for true clock deviation is reduced. It is marked as a local disturbance, but not added as a candidate for true clock deviation or as a candidate for formal drift. It is saved as an observation mark for subsequent judgment based on the continuity relationship of adjacent calibration cycles. Thus, short-term disturbances, instantaneous interference, or temporary state fluctuations within a single calibration cycle are distinguished from deviation changes with continuous continuity characteristics. After completing the above hierarchical discrimination, a candidate quantity structure is formed. The candidate quantity structure includes at least the following fields: real clock deviation candidate quantity, link bidirectional propagation asymmetric drift candidate quantity, device internal delay drift candidate quantity, candidate quantity source marker, candidate quantity confidence level, and current calibration cycle discrimination marker. The candidate quantity source marker is used to characterize the result of the current candidate quantity's main corresponding direction symmetry discrimination, node consistency discrimination, or period continuity discrimination. The candidate quantity confidence level can be selected as high confidence, observation confidence, or low confidence. When the number of effective pairing groups is sufficient, the direction discrimination is stable, the node status is complete, and the period continuity is good, it is a high confidence candidate quantity. When the number of samples is close to the lower limit, the node status is partially missing, or the deviation change is only locally continuous, it is an observation confidence candidate quantity. When the number of samples is greatly mismatched, the bidirectional records are severely mismatched, or the status records are obviously missing, it is a low confidence candidate quantity. The current calibration cycle discrimination marker is used to characterize the usability of the candidate quantity in this cycle, and can be selected as normal use, restricted use, or observation only use. If a candidate quantity is not sufficient to generate, a null value or a low confidence marker is output. In abnormal and boundary situations, constraints are set for candidate quantity generation in the current calibration period. If the number of mismatched records in the current calibration period exceeds 20% to 30% of the number of valid paired groups, the current calibration period's discrimination flag is set to restricted use. The basis for setting this range is that after the number of mismatched records reaches this proportion, the stability of the valid observation correspondence and the confidence level of the candidate quantity will significantly decrease, and the current calibration period is no longer sufficient to support high-confidence source stratification. If the difference between the number of forward and reverse record sets exceeds a preset proportion, which can be selected as exceeding 50% of the number of records on the smaller side, the current calibration period's discrimination flag is also set to restricted use. The basis for setting this proportion is that when the number of forward and reverse observation samples reaches this imbalance level, the direction symmetry discrimination no longer has stable performance. The statistical basis is defined as follows: if the proportion of missing node status records or environmental status records exceeds 10% to 20% of the total number of status records in the current calibration cycle, then no candidate quantity for internal delay drift of high-confidence equipment will be output. The basis for setting this range is that after the proportion of missing status records reaches this level, the integrity of the continuous state mapping relationship is insufficient to support the high-confidence node consistency judgment. If there is uncertainty in the attribution of records in the boundary buffer between the current calibration cycle and adjacent cycles, they will be preferentially not included in the corresponding group of high-confidence observations, but treated as supplementary observation records. After the above constraints, the candidate quantity is established on the basis of verifiable and traceable pairing relationships. The hierarchical judgment results of the current calibration cycle can be divided into high-confidence results, restricted use results, and degradation results, and are used as input for calibration judgment.

[0019] S3. Based on the deviation succession relationship between adjacent calibration cycles and the current operating status, the candidate quantities of real clock deviation, link bidirectional propagation asymmetric drift, and device internal delay drift are calibrated and determined. The specific implementation is as follows: After the above candidate quantity structure is formed, the candidate quantities of real clock deviation, link bidirectional propagation asymmetric drift and device internal delay drift are calibrated and determined based on the link state sequence, the current formal calibration version record and the adjacent record of the current calibration cycle. The clock synchronization update is not directly performed based on the candidate quantities in the current calibration cycle. Instead, the inheritance relationship, constraint relationship and state consistency are determined first to form an updateable state, a suspended state or a frozen state. First, establish a continuous judgment window. This window is a set of several adjacent calibration cycles selected around the current calibration cycle, used to determine whether there is a continuous continuity relationship between the current candidate value and historically valid versions. The continuous judgment window must include at least the previous calibration cycle, and its length can be selected from 2 to 4 cycles. This range is set because the continuous judgment window needs to reflect the short-term continuous trend of the current link deviation change, but should not be too long to avoid outdated operating states entering the current judgment. For scenarios with stable links and small cycle fluctuations, the continuous judgment window can be selected from 3 to 4 cycles. For scenarios with rapid environmental changes or links in the recovery phase, the continuous judgment window can be selected from 2 to 3 cycles. After the continuous judgment window is established, read the deviation value, version status, effective period, and corresponding operating status of the most recent official calibration version within the window, using this as the continuity benchmark for the current calibration cycle. Continuously stable cycles are prioritized for inclusion in the continuous judgment window; abnormal cycles or frozen cycles are not used as normal continuity benchmarks. The calibration process first determines the continuity between the candidate real clock deviation and the previous official calibration version. This continuity can be judged from three aspects: direction of change, magnitude of change, and degree of continuous stability. Direction of change refers to whether the current candidate real clock deviation changes in the same direction, fluctuates slightly, or changes in the opposite direction to the previous official calibration version. Magnitude of change refers to whether the difference between the current candidate real clock deviation and the previous official calibration version is within the allowable update range. Continuous stability refers to whether the current candidate real clock deviation continues to be continuous within the continuous determination window. If the current candidate real clock deviation and the previous official calibration version have the same direction of change, the magnitude of change is less than the continuity threshold, and there is no obvious reverse abrupt change in the continuous determination window, it is considered to be continuously continuous. If the current candidate real clock deviation changes abruptly or the magnitude of change exceeds the continuity threshold, constraint analysis will continue through bidirectional propagation of asymmetric drift candidate values ​​and internal device delay drift candidate values. The acceptance threshold can be a relative threshold or an absolute threshold. The relative threshold can be selected as 5% to 15% of the deviation value of the previous formal calibration version, and the absolute threshold can be selected as 1 nanosecond to 10 nanoseconds. For short-distance, highly stable quantum measurement links with strong time resolution, a smaller range of acceptance thresholds can be used, such as 5% to 8% or 1 nanosecond to 3 nanoseconds. For long-distance quantum measurement links with high environmental sensitivity or relatively large inherent equipment jitter, a larger range of acceptance thresholds can be used, such as 8% to 15% or 3 nanoseconds to 10 nanoseconds. The basis for setting this range is that the acceptance threshold should... The threshold is matched to the natural variation range of the deviation under normal continuous operation of the link, so as to avoid misjudging normal gradual changes as abnormal sudden changes, or reducing the ability to distinguish abnormal deviation changes due to excessively large thresholds; the relative threshold is used to adapt to different reference deviation magnitudes, and the absolute threshold is used to limit the absolute offset amplitude of a single cycle. The two can be used alone or in combination according to the upper limit of the deviation fluctuation within the normal operation cycle of the link; to ensure the reproducibility of parameters, the acceptance threshold can be adjusted according to the deviation fluctuation amplitude of several consecutive stable cycles after the initial base construction, and reset in the freeze recovery or reconstruction state; After initially determining the acceptance relationship of the candidate quantities of the real clock skew, the candidate quantities of the bidirectional propagation asymmetric drift of the link are constrained. The determination of the candidate quantities of the bidirectional propagation asymmetric drift of the link focuses on their proportion relative to the change amplitude of the current candidate quantities of the real clock skew and their correspondence with changes in environmental conditions. If the candidate quantities of the bidirectional propagation asymmetric drift of the link are within an acceptable fluctuation range in the current calibration period, the acceptance of the real clock skew candidates is maintained. If the candidate quantity is close to or exceeds the acceptable fluctuation range, the update priority of the current candidate quantities of the real clock skew is reduced. The acceptable fluctuation range can be selected as less than 30% to 50% of the change amplitude of the current candidate quantities of the real clock skew, or it can be a fixed limit value based on the link length and environmental sensitivity. The basis for setting this range is that the proportion of the candidate quantities of the bidirectional propagation asymmetric drift of the link in the current skew change should be limited. The ratio is such that, under normal constraints, the candidate value of the bidirectional propagation asymmetric drift of the link does not exceed the main variation range of the real clock deviation candidate value. The lower limit corresponds to the upper bound of the statistical fluctuation of the bidirectional propagation asymmetric drift of the link within a stable operating cycle, and the upper limit corresponds to the boundary range where the asymmetric drift begins to significantly interfere with the real clock deviation acceptance judgment. For scenarios with short links and stable environments, the candidate value of the bidirectional propagation asymmetric drift of the link is preferably taken as the lower limit. For scenarios with long links, obvious temperature drift, or complex deployment environments, the limit can be appropriately relaxed under the condition that the link length increases, the temperature change rate increases, or the attenuation fluctuation amplitude increases. If the candidate value of the bidirectional propagation asymmetric drift of the link exceeds the acceptable fluctuation range in the current calibration cycle, and the environmental status record shows that temperature changes, attenuation changes, or maintenance switching are taking place, the calibration judgment of the current calibration cycle will be preferentially adjusted to a suspended state or a frozen state. After determining the candidate quantities for asymmetric drift in bidirectional propagation of the link, the internal delay drift candidate quantities of the device are then evaluated for consistency in their operating states. These internal delay drift candidate quantities reflect the impact of changes in the node's internal detection response, gating state, time stamp stability, and photoelectric conversion state on the current deviation. Their priority is directly related to the node's operating state. If the current node is in a state of detection hysteresis, gating restriction, time stamp instability, or photoelectric conversion state fluctuation, the priority of the internal delay drift candidate quantities is increased. When a candidate quantity forms a continuous correspondence with the node's abnormal state and meets the requirements for the number of state correspondences and duration, the current deviation change is preferentially determined as internal device drift, rather than as a change in the actual clock deviation. Here, the continuous correspondence of states is prioritized for determination, and the candidate quantity amplitude is not used as the sole determination condition; the candidate quantity change amplitude is only used as an auxiliary judgment criterion. If the current node's operating state remains normal, and the internal delay drift candidate quantities do not form a continuous correspondence within the continuous determination window, their priority is reduced. The number of state correspondences can be selected as consistent correspondences occurring within 2 to 5 consecutive observation sub-windows, and the duration can be selected as no less than 30% to 50% of the current calibration cycle length, where the observation sub-windows are based on the sub-window division described above. The range of state correspondences is set to cover the minimum duration of node state changes to exclude the interference of occasional state jitter on the judgment and to match the density of the observation sub-window division. The range of duration is set to ensure that the correspondence between node state changes and deviation changes covers a representative continuous time period within the current calibration cycle. Its lower limit is used to exclude instantaneous disturbances, and its upper limit is used to avoid misjudging state fluctuations that only occur in local short periods as continuous constraints for the entire cycle. If the candidate quantity of internal delay drift has formed a continuous state correspondence and the node state is obviously abnormal, the current calibration cycle should not be directly judged as an updatable state, but should at least be judged as a suspended state. If the candidate quantity forms a high-constraint state for multiple consecutive calibration cycles, it should be directly judged as a frozen state. After performing continuity and constraint analyses on the candidate quantities of real clock skew, link bidirectional propagation asymmetric drift, and device internal delay drift, a calibration judgment result for the current calibration cycle is formed. The calibration judgment result can be categorized into three states: updatable, suspended, and frozen. The updatable state corresponds to a stable continuity between the candidate quantities of real clock skew and the previous formal calibration version, the link bidirectional propagation asymmetric drift candidate quantities not exceeding acceptable limits, the device internal delay drift candidate quantities not forming a continuous correspondence with abnormal node states and meeting the freezing condition, and no restrictive anomalies appearing in the current operating state. The suspended state corresponds to the candidate quantities of real clock skew having continuity, but the link bidirectional propagation asymmetric drift candidate quantities or the device internal delay drift candidate quantities approaching the judgment threshold, or the current operating state being in a transitional phase. The frozen state corresponds to the link bidirectional propagation asymmetric drift candidate quantities exceeding limits, or the device internal delay drift candidate quantities forming a continuously high-constraint state, or significant anomalies appearing in the current node state, link state, or environmental state. After the above processing, the analysis results at the candidate quantity level are converted into directly callable calibration judgment results. To ensure the auditability and feasibility of calibration judgment rules, state transitions are expressed using a state machine. When the current calibration cycle initially enters the judgment phase, it is in the normal observation state. When the update condition is met, the state transitions from normal observation to pending update. When the postponement condition is met, the state transitions from normal observation to observation storage. When the freeze condition is met, the state transitions from normal observation, pending update, or observation storage to freeze, with freeze having higher priority than pending update and observation storage. If the previous cycle within the continuous judgment window was already in observation storage, and the current calibration cycle is again judged to be in postponement, then the observation storage state is maintained and the number of observations is accumulated. When the accumulated number of observations reaches a preset upper limit, it transitions to freeze. The preset upper limit can be selected from 2 to 5 times. This range is set to match the typical duration of link edge fluctuations and the blocking requirements of the observation storage state on the formal update path. If the previous cycle was in freeze, the current calibration cycle does not directly transition to the pending update state, but continues judgment based on recovery conditions. The output structure generated in this stage includes at least the following fields: calibration determination status, real clock deviation determination flag, link bidirectional propagation asymmetric drift determination flag, device internal delay drift determination flag, succession relationship flag, and operating status constraint flag. The calibration determination status characterizes the processing result of the current calibration cycle; the real clock deviation determination flag characterizes the reception status of the current real clock deviation candidate; the link bidirectional propagation asymmetric drift determination flag characterizes the constraint degree of the current link asymmetric drift; the device internal delay drift determination flag characterizes the constraint degree of the current node's internal delay on the deviation determination; the succession relationship flag characterizes the reception status of the current real clock deviation candidate compared to the previous formal calibration version; and the operating status constraint flag indicates the constraint degree of the current operating status on the calibration determination. The operating status constraint flag can be normal, observation, restricted update, or prohibited update. Based on the output structure, version updates, observation and temporary storage, or frozen storage can be performed directly. Interruptions and boundary conditions are handled according to the same rules within the current calibration cycle. If there is a record interruption between the current calibration cycle and the previous cycle, and the interruption duration does not exceed one calibration cycle, the most recent official calibration version is allowed to be used as the basis for acceptance, and the acceptance relationship of the current calibration cycle is downgraded to weak acceptance. If the interruption duration exceeds one calibration cycle, the acceptance relationship is adjusted to require reconstruction, and the current calibration cycle does not enter the updatable state. If the proportion of boundary buffer records to the number of effective pairing groups exceeds the preset upper limit, which can be selected from 10% to 20%, the judgment result of the current calibration cycle is at most in a suspended state. The basis for setting this range is that this proportion corresponds to the upper limit range of the acceptance judgment result of the current calibration cycle where the boundary undetermined records have not yet dominated. If there are missing status records, but the candidate quantity stratification results remain stable, the observation attributes of the real clock deviation candidate quantity can be temporarily retained, and the internal delay drift candidate quantity of the device is not judged as highly reliable.

[0020] S4. Perform clock synchronization calibration update based on the calibration judgment result, and record the bidirectional propagation asymmetric drift of the link and the internal delay drift of the device. The specific implementation is as follows: After the above calibration judgment results are generated, clock synchronization calibration is updated according to the calibration judgment status, and the bidirectional propagation asymmetric drift of the link and the internal delay drift of the device are recorded respectively. This process does not directly use the official calibration version, but uses version management to generate candidate versions, observe or freeze records. When entering the calibration update branch, the current official calibration version is not directly overwritten. Instead, a calibration update record corresponding to the current calibration cycle is generated first, and the real clock deviation judgment value is written into the candidate area of ​​the current calibration version. The calibration update record includes at least the generation cycle number, the corresponding official version number, the candidate version number, the real clock deviation written value, the link bidirectional propagation asymmetric drift record value, the device internal delay drift record value, the version status, the write time, and the effective condition flag. The candidate version number is generated using an incrementing integer, which can be selected to increment by 1 based on the current official version number with each update action. The version status can be divided into pending effectiveness, officially effective, observation and temporary storage, and frozen retention. Pending effectiveness means that the current calibration cycle has met the update conditions, but has not yet completed continuous confirmation, so it does not replace the current official calibration version. Officially effective means that the current version has passed continuous confirmation and is used as the reference for subsequent clock synchronization calls. Observation and temporary storage means that the current calibration cycle only retains the result for subsequent continuous judgment and does not form a formal progress. Frozen retention means that the current calibration cycle does not perform an update on the official version, but only keeps the current official version valid. The effective condition flag includes at least the consecutive confirmation count condition and the prohibition triggering condition; the consecutive confirmation count condition is the number of consecutive valid periods required for the candidate version to be transformed into the official effective version, and the prohibition triggering condition is the condition under which the candidate version cannot be transformed into the official effective version; the prohibition triggering condition includes at least the frozen state trigger, single update range exceeding the limit, and continuous observation state; the effective condition flag can directly write single-period occasional deviations into the official calibration version. To ensure parameter reproducibility and controllable update behavior, candidate versions must meet continuous confirmation conditions before entering formal implementation. The continuous confirmation condition is preferably set to being determined as updatable for 2 to 3 consecutive calibration cycles. For highly stable link scenarios, 2 consecutive calibration cycles can be used to shorten update latency. For scenarios with large environmental fluctuations, long links, or frequent switching of operating states, 3 consecutive calibration cycles can be used to improve the robustness of version switching. This range is based on the fact that the number of continuous confirmations needs to meet the continuous stability requirements of formal version switching while also taking into account the response latency constraints of online updates. During continuous confirmation, the current candidate version remains in a pending-implementation state and is not directly used for external clock synchronization. If an updatable state reappears during continuous confirmation, the candidate version confirmation count is refreshed. If a suspended state occurs, the current candidate version is retained, and its state is adjusted to observation and temporary storage, while the continuous confirmation cycle is extended. If a frozen state occurs, the current candidate version is canceled and no longer participates in the formal implementation judgment. Before the actual clock skew value is included in the candidate version, a single update amplitude verification is required. The single update amplitude limits the allowable variation of the current calibration cycle relative to the previous official version, to prevent excessively large skew changes from being directly written to the candidate version when the link is in a state of edge fluctuation or when the candidate value changes abruptly. The single update amplitude can be selected as 10% of the deviation value of the previous official version, or a fixed absolute value of 2 nanoseconds to 20 nanoseconds. For scenarios with short links and high clock stability, the single update amplitude can be selected as 2 nanoseconds to 5 nanoseconds or 5% to 10% of the deviation value of the previous official version. For scenarios with long links and significant fluctuations in operating conditions, the single update amplitude can be selected as 5% to 10% of the deviation value of the previous official version. In typical scenarios, the single update increment can be selected as 5 nanoseconds to 20 nanoseconds or 10% of the deviation value of the previous official version. The above range is based on the fact that the real clock deviation should not change significantly beyond the upper limit of the normal drift in a single calibration cycle under normal continuous operation. The relative threshold is used to adapt to the magnitude of change of the deviation base value of different official versions, and the absolute threshold is used to limit the absolute write increment in a single cycle. If the difference between the current candidate real clock deviation value and the previous official version exceeds the maximum single update increment, the write value will not be included in the version to be effective in the current calibration cycle, but will be transferred to the observation record and confirmed in subsequent calibration cycles. Asymmetric drift during bidirectional propagation of the link and internal device delay drift are recorded separately at this stage, and are not written as actual clock deviation values. Each separate record must include at least the period number, drift type, drift magnitude, confidence level, corresponding node or link direction marker, associated operating status marker, and record time field. The drift type represents asymmetric drift during bidirectional propagation of the link and internal device delay drift; the drift magnitude represents the value or level of the drift item under the unified time reference in the current calibration period; and the confidence level represents the reliability of the drift item. The corresponding node or link direction marker indicates the location of the drift item, and the associated operating status marker associates the drift record with the node status, environmental status, or maintenance status within the current calibration period. Captured drift candidates are recorded separately and then directly invoked in subsequent calibration periods. In the current stage, priority is given to maintaining the formal calibration version area and the sub-item drift record area separately. The formal calibration version area is used to store the formal version currently available for clock synchronization and the version to be activated, and includes at least the following fields: version number, previous version number, activation status, actual clock deviation value, activation start period, activation termination condition, and acknowledgment count. The sub-item drift record area is used to store the link bidirectional propagation asymmetric drift records and device internal delay drift records for the past few cycles, and includes at least the following fields: record number, cycle number, drift type, drift magnitude, confidence level, node or direction marker, and associated status marker. The formal calibration version area and the sub-item drift record area are logically independent of each other and are associated through the cycle number, version number, and status marker. The formal calibration version area is used for clock synchronization, and the sub-item drift record area is used for subsequent calibration judgment and anomaly analysis. In the observation log branch, the current calibration cycle does not drive the evolution of the official version, but only generates observation logs. Observation logs include at least the cycle number, candidate real clock deviation value, link bidirectional propagation asymmetric drift log value, device internal latency drift log value, observation reason flag, and cumulative observation count. The observation reason flag preferentially uses a predefined enumeration set, and its sources may include situations such as the drift candidate quantity approaching the threshold, the operating state just changing, the boundary buffer participation ratio being too high, or the single update magnitude exceeding the limit. It is used to limit the priority path for continuing observation, entering freeze, or restoring candidate version confirmation in the next calibration cycle; the cumulative observation count... The number is used to reflect whether the current link is continuously on the edge of instability; if the current calibration cycle is in a suspended state and the number of consecutive suspensions reaches the preset upper limit, the value can be selected from 2 to 5 times, then the current state changes from observation to frozen state; the setting of this range is based on the fact that it should match the typical duration of the link edge fluctuation state and the judgment requirements for the observation state to the frozen state, corresponding to the maximum number of consecutive cycles that can be retained for observation without immediate freezing when the current link is in an edge fluctuation state; the real clock deviation candidate value and drift record value in the observation record enter the historical chain for subsequent calibration cycles to perform continuity judgment and recovery trend analysis; While in a frozen state, the current official calibration version remains unchanged, no new candidate versions are generated, and only the freeze reason and the current calibration cycle drift item are saved. Freeze reasons can include exceeding the limit for bidirectional propagation asymmetric drift, exceeding the limit for internal device delay drift, abnormal operating status, exceeding the limit for continuous observation, and incomplete interruption recovery, etc. The recording method uses a predefined enumeration set. The frozen state record should include the cycle number, the current official version number, the freeze reason, the drift type, the drift magnitude, the confidence level, and the freeze start flag. If the frozen state occurs for the first time in the current calibration cycle, the freeze start flag is recorded to initiate the recovery observation count. If the frozen state persists, the freeze duration cycle number is incremented. The freeze duration cycle number is calculated by adding the freeze start flag and the current cycle number. The freeze reason and drift record are stored together for future hold, recovery, and repeated calls. Regarding resource constraints, it is not required to retain all historical versions and all historical drift records in the long term at the current stage. To balance traceability and operational burden, it is advisable to retain the drift records of the most recent 5 to 20 official versions and the most recent 20 to 100 calibration cycles. For highly stable scenarios, it is advisable to retain 5 to 10 official versions and 20 to 50 drift record cycles. For highly volatile scenarios or links that frequently experience freeze recovery, it is advisable to retain 10 to 20 official versions and 50 to 100 drift record cycles. The number of official versions to retain is based on the needs of version rollback and trend analysis, while the drift record retention period is based on the needs of anomaly analysis and recovery trend determination. If resources are further limited, at least the current official version, the most recent version to be implemented, the most recent observation record, and drift records of the most recent cycles should be retained to ensure that subsequent processing has the basic conditions for anomaly recovery. Regarding time constraints, the update, observation, or freeze processes corresponding to the current calibration cycle should be completed within one calibration cycle after the end of that calibration cycle. For example, if the current calibration cycle is 1 second, the calibration update and drift recording actions should be completed within the next 1 second. The basis for setting this range is to ensure that the processing results of the current calibration cycle can be written before the next calibration cycle is called, so as to avoid subsequent clock synchronization calls lagging behind the version. If the processing times out, the results of the current calibration cycle will not enter the formal version update, but will be converted into observation records. The continuous confirmation count of the candidate version should be refreshed in a timely manner after the end of each calibration cycle, and should not be written in a concentrated manner across multiple calibration cycles, so as to maintain the real-time consistency of the state machine. When the current calibration cycle is in an updatable state, the status changes from pending determination to pending version generation; when the conditions are continuously confirmed to be met, the status changes from pending to officially effective; when the current calibration cycle is in a suspended state, the status changes from pending determination to observation and storage, and the number of observations is accumulated; when the current calibration cycle is in a frozen state, or the number of observations reaches the upper limit, or the update range exceeds the limit, the status changes to a frozen and maintained state; the frozen and maintained state has higher priority than the pending to effective state and the observation and storage state. After entering the frozen and maintained state, the current candidate version becomes invalid, and the original confirmation count is no longer retained; according to the above status transition relationship, the version record, observation record, and drift record of the current calibration cycle are updated accordingly.

[0021] S5. In the event of link malfunction, sudden environmental changes, or bidirectional observation mismatch, maintain the current calibration version and re-enter the next calibration cycle after recovery. Specifically, the implementation is as follows: After the aforementioned versioning update and drift sub-item recording are completed, the abnormal retention and recovery processing stage is entered. This stage is used to maintain the current formal calibration version when the link operation is abnormal, the environmental state changes suddenly, or bidirectional observation mismatch occurs, and to restore subsequent calibration processing according to the recovery conditions after the abnormality is resolved. The inputs at this stage include at least the following fields: formal calibration version, candidate version status, component drift records, and current calibration cycle running status. The formal calibration version is used to maintain the external reference during anomalies. The candidate version status is used to determine whether there are versions waiting to be implemented that need to be frozen or disabled. Component drift records are used to identify the source of the anomaly and its duration. The current calibration cycle running status is used to determine the anomaly type, anomaly intensity, and recovery conditions. Running anomalies may include unidirectional recording interruption, simultaneous loss of bidirectional recordings, continuous loss of time stamps, sudden changes in node probe response, and gating status anomalies. Sudden changes in environmental status may include the link temperature change rate exceeding the aforementioned environmental status anomaly judgment threshold, a sudden increase in link attenuation level, an abnormal increase in vibration level, or the start of maintenance switching. Bidirectional observation mismatch may include the difference between the number of forward and reverse records exceeding the aforementioned observation imbalance judgment ratio, the inability to form a sufficient number of corresponding observation groups within the pairing time window, and a significant disconnect between quantum correlation measurement records and bidirectional time transfer records. After identifying any of the above anomalies, the current calibration cycle will no longer enter the formal calibration update branch, but will maintain the current formal calibration version and switch the current running status to the freeze-hold state. Anomaly triggering employs a tiered judgment method; severe anomalies trigger a single-time freeze, while weak anomalies trigger a continuous cumulative freeze. Severe anomalies include at least the simultaneous loss of bidirectional records, continuous loss of time stamps reaching a preset limit, significantly exceeding the link temperature change rate limit, the start of maintenance switching, and complete disconnection between quantum correlation measurement records and bidirectional time transfer records. Weak anomalies include at least the short-term interruption of unidirectional recording, the difference between the number of forward and reverse records approaching the limit, short-term abrupt changes in node probe response, and short-term restrictions on gating states. Severe anomalies correspond to those where the current calibration cycle is insufficient to maintain the reliability of the formal calibration version, while weak anomalies correspond to those within the current calibration cycle. Anomalies that can still be observed but are not suitable for immediate formal updates; the preferred cumulative trigger count is 2 to 3 times; this range is set to avoid both the excessive amplification of a single weak anomaly into a freeze event and the prevention of the link remaining on the edge of instability for a long time when weak anomalies occur continuously, thus continuously generating unreliable candidate quantities; the minimum number of consecutive cycles for a weak anomaly to turn from an occasional event into a continuous anomaly trend is related to the calibration cycle length and the duration of the weakening of the candidate quantity's credibility by the weak anomaly; once a severe anomaly is identified, the current calibration cycle immediately enters a freeze state; when the continuous occurrence of weak anomalies reaches the cumulative threshold, the current calibration cycle also enters a freeze state; Once in freeze hold state, the current official calibration version continues to serve as the clock synchronization output reference, the current candidate version stops progressing, and the observed temporary version no longer accumulates confirmation counts; while in freeze hold state, data from subsequent calibration cycles continues to be collected, but only for anomaly observation, recovery judgment, and drift recording, so as to restore the continuity relationship after the freeze is lifted; The freeze duration can be set based on the number of calibration cycles or a fixed time. Based on the number of calibration cycles, the freeze duration can be selected from 2 to 10 calibration cycles; based on a fixed time, the freeze duration can be selected from 5 to 30 seconds. For scenarios with short links, fast environmental recovery, and good device stability, a shorter freeze duration can be used, such as 2 to 4 calibration cycles or 5 to 10 seconds. For scenarios with long links, high environmental sensitivity, significant maintenance and switching impact, or slow internal node drift recovery, a longer freeze duration can be used, such as 5 to 10 seconds. The freeze duration is set at 10 to 30 seconds per calibration cycle. This range is based on the fact that the environment and equipment typically require a buffer period to recover from an abnormal state to a stable state. If the freeze duration is too short, the update will be restarted before the anomaly has disappeared, which may cause the apparent deviation in the early recovery stage to be mistakenly written into the actual clock deviation. If the freeze duration is too long, it will reduce the response efficiency of clock synchronization calibration. To ensure parameter reproducibility, the freeze duration should be preset in advance during the initial deployment of the system in combination with the statistical characteristics of link recovery, and can be adjusted in subsequent maintenance based on historical anomaly recovery time. During the freeze period, the causes, duration, and individual drift changes of the anomaly are continuously recorded. The anomaly record structure includes at least the following fields: freeze status flag, freeze start period, anomaly type flag, anomaly intensity flag, current official version number, drift record reference flag during freeze, and recovery observation count. The anomaly type flag distinguishes between link operation anomalies, sudden environmental changes, and bidirectional observation mismatches, and can be further subdivided into subcategories such as unidirectional interruption, bidirectional interruption, temperature change, attenuation surge, detection anomaly, and gating anomaly. The anomaly intensity flag distinguishes between severe and weak anomalies. The drift record reference flag during freeze is used to associate the aforementioned bidirectional propagation asymmetric drift records and internal device delay drift records, so that the drift change trend before and after freeze can be directly read during recovery. The recovery observation count records the number of consecutive observation periods completed after the freeze is lifted to determine whether the recovery entry conditions are met. While in the frozen state, continuously check whether the cause of the anomaly has disappeared and whether the bidirectional observation has met the preset pairing conditions again. The cause of the anomaly disappearing means that the abnormal operating state or sudden change in the environmental state that triggered the freeze has returned to the allowable range, such as the bidirectional recording being restored to completeness, the time stamp being restored to continuity, the node detection response being restored to stability, the gating state being restored to normal, the temperature change rate being reduced to below the aforementioned abnormal threshold, the attenuation level being restored, or the maintenance switch being completed. The bidirectional observation meeting the preset pairing conditions again means that the difference between the number of forward and reverse records in the current calibration cycle has fallen back to the aforementioned allowable range, and an observation corresponding group with no less than the minimum number of effective pairing groups is formed within the pairing time window, and the time proximity correspondence between the quantum correlation measurement record and the bidirectional time transfer record is re-established. If the above conditions have not yet been met, the frozen state is maintained and the freeze duration period is accumulated. If the above conditions have been met, the frozen state is switched to the recovery observation state. Once the observation period has ended, data for 1 to 3 consecutive calibration cycles is collected again, and the calibration process is re-executed according to the aforementioned time sequence, candidate quantity generation, and calibration judgment rules. For highly stable link scenarios, the observation period can be 1 to 2 cycles; for highly volatile scenarios or scenarios that have just experienced severe anomalies, the observation period can be 2 to 3 cycles. This range is set based on the fact that the number of observation cycles should meet the minimum number of consecutive observation cycles required for the restoration of the connection after the anomaly is resolved, while also taking into account the recovery speed and judgment stability, corresponding to the minimum number of consecutive verification cycles required for the return to a stable state from an abnormal state after the freeze is lifted. The results generated during the observation period are only used to restore the connection and re-establish the basis for candidate version confirmation. The system does not directly push for a switch to the official version. If the system is judged to be updatable for consecutive cycles in the recovery observation state, and there is no unacceptable break between the current real clock deviation candidate quantity and the official calibration version before freezing, then the system exits the freeze state and re-enters the next calibration cycle. An unacceptable break can be determined within the consecutive recovery observation cycle. It can be selected as a sudden change in the direction of change of the current real clock deviation candidate quantity relative to the official calibration version before freezing, and the change magnitude continuously exceeding the aforementioned threshold, or the link bidirectional propagation asymmetric drift candidate quantity and the device internal delay drift candidate quantity exceeding the limit again during the recovery observation period. If an anomaly occurs again during the recovery observation period, the recovery observation is immediately terminated, the freeze duration is extended, and the recovery observation count is restarted. When no available formal calibration version exists before freezing, the system does not directly enter the normal update process after recovery from the anomaly, but instead first enters the reconstruction state. The reconstruction state is applicable to scenarios where the initial baseline construction has not been completed before the anomaly occurs, the original formal version becomes invalid, or the system restarts after a long interruption. In the reconstruction state, the system still re-collects, processes, and judges data according to the aforementioned time sequence, source stratification, and calibration judgment rules. However, the newly generated candidate quantities do not depend on the formal version before freezing, but are determined based on the continuous reconstruction conditions to determine whether a new formal version is formed. The continuous reconstruction conditions are preferably set to meet the updateability judgment for 2 to 5 consecutive calibration cycles, with 2 to 3 cycles for high-stability links and 3 to 5 cycles for high-fluctuation links. The basis for setting this range is that the reconstruction process needs sufficient continuous stable cycles to rebuild a reliable calibration baseline, but should not last too long to avoid excessive reconstruction latency. A new formal calibration version is generated only when the continuous reconstruction conditions are met. Before that, only the reconstruction progress marker is output, and the formal update result is not output. During the abnormal period, constraints are applied to candidate versions and observation records. If a version awaiting activation exists when the freeze is triggered, it is immediately frozen and becomes invalid, and the original confirmation count is no longer retained. If an observation record has been formed before the freeze is triggered, it is retained for recovery trend analysis, but it is not directly continued as a version awaiting activation after the freeze is lifted. If the proportion of boundary buffer records remains high during the freeze period, or if the number of effective paired groups during the recovery observation period reaches the lower limit but is mainly supported by boundary buffer records, the current calibration cycle only retains the observation attributes and does not directly exit the freeze state. The current phase includes at least four states: normal update, frozen hold, recovery observation, and rebuild. In the normal update state, if an anomaly trigger condition is met, it transitions to the frozen hold state. The frozen hold state remains unchanged until the cause of the anomaly is resolved; after the cause is resolved, it transitions to the recovery observation state. In the recovery observation state, if continuous observation meets the update conditions and the continuity is normal, it returns to the normal update state. If another anomaly occurs, it returns to the frozen hold state. If no usable official version exists, it transitions to the rebuild state. In the rebuild state, if continuous rebuild conditions are met, a new official version is generated and it returns to the normal update state. If another anomaly occurs, it transitions to the frozen hold state. In terms of output structure, while maintaining the current official calibration version as the output benchmark, it should include at least the following fields: freeze status flag, freeze start period, anomaly type flag, recovery observation count, recovery entry status flag, and current official version number. Among them, the freeze status flag indicates whether the current link is in a state where updates are prohibited; the freeze start period indicates the freeze start point; the anomaly type flag indicates the anomaly category; the recovery observation count indicates the continuous recovery observation status; the recovery entry status flag indicates whether the current link is in a state of freeze maintenance, recovery observation, or reconstruction; and the current official version number indicates the currently available official calibration version. If necessary, it may also include fields such as a freeze reason summary and a drift change trend flag during the freeze period. Regarding resource constraints, the current priority is to retain the complete recovery chain of the most recent freeze event and the summary records of the last 5 to 20 anomalies. The complete recovery chain should include at least the freeze start period, freeze reason, drift record index during freeze, recovery observation period record, and final recovery result. The basis for setting this range is that the number of anomaly summary records to be retained should cover the recent anomaly statistical trend of the link, but should not be too large to avoid increasing the long-term storage burden. Regarding time constraints, the freeze state switch after an anomaly is triggered should be completed before the end of the current calibration period or at the latest before the start of the next calibration period. The recovery observation count and reconstruction count should be refreshed in a timely manner at the end of each calibration period.

[0022] The technical solution of this embodiment takes the quantum measurement link between two quantum measurement stations as an example. The system continuously collects bidirectional time transfer records, quantum correlation measurement records, and link operation status records according to a fixed calibration cycle, and completes the processing under a unified time reference. Then, based on the correspondence between forward and reverse observations, the total time deviation is classified into three levels: real clock deviation, bidirectional propagation asymmetric drift of the link, and internal device delay drift. After that, the calibration of the current calibration cycle is determined by combining the succession relationship of adjacent calibration cycles and the current node status and environmental status. If the current calibration cycle meets the update conditions, a candidate calibration version is generated first, and after continuous confirmation, it is converted into a formal calibration version, while simultaneously recording the historical information of bidirectional propagation asymmetric drift of the link and internal device delay drift. If the current calibration cycle experiences recording interruption, temperature change, gating abnormality, or bidirectional observation mismatch, the current formal calibration version remains unchanged and enters a frozen state. After the abnormality disappears, the system first enters the recovery observation process, then restores normal calibration according to the succession relationship, and enters the reconstruction process if necessary, thereby maintaining the consistency and stability of the quantum measurement link time and frequency reference under continuous operation and environmental change conditions.

[0023] It should be noted that this invention can be deployed on the device itself to realize embedded applications, or it can run on a PC or other terminal with a user interface, thereby meeting various hardware environments and usage requirements.

[0024] The above embodiments can be implemented in whole or in part by software, hardware, firmware, or any other combination. When implemented in software, the above embodiments can be implemented in whole or in part by a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, the processes or functions of the embodiments of this application are implemented in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted wirelessly or wiredly from one website, computer, server, or data center to another website, computer, server, or data center. Wired methods include optical fiber, twisted pair, coaxial cable, etc. Wireless methods include infrared, microwave, etc. Available media include any available media that can be accessed by a computer or data storage devices such as servers and data centers that contain one or more sets of available media. Available media can be magnetic media (floppy disks, hard disks, magnetic tapes), optical media (DVDs), or semiconductor media. Semiconductor media can be solid-state drives.

[0025] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0026] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for clock synchronization calibration of a quantum measurement link, characterized in that, include: S1. Collect bidirectional time transfer records, quantum correlation measurement records, and link operation status records of the quantum measurement link during the current calibration cycle, and perform time-series organization based on a unified time reference; S2. Based on the correspondence of the bidirectional observation records, the total time deviation within the current calibration cycle is classified and judged in layers to generate candidate quantities of real clock deviation, bidirectional propagation asymmetric drift of the link, and internal delay drift of the device. S3. Based on the deviation succession relationship between adjacent calibration cycles and the current operating status, perform calibration and determination on the candidate quantities of real clock deviation, link bidirectional propagation asymmetric drift, and device internal delay drift. S4. Perform clock synchronization calibration update based on the calibration judgment result, and record the bidirectional propagation asymmetric drift of the link and the internal delay drift of the device; S5. In case of abnormal link operation, sudden change in environmental status, or bidirectional observation mismatch, maintain the current calibration version and re-enter the next calibration cycle after recovery.

2. The quantum measurement link clock synchronization calibration method according to claim 1, characterized in that, Collect bidirectional time transfer records, quantum correlation measurement records, and link operation status records of the quantum measurement link during the current calibration cycle, including: The current calibration period is established according to a fixed-length window, and multiple source records from the quantum measurement link are uniformly accessed, periodically merged, and field-identified. Multi-source records must include at least a record number, record type, time stamp, direction stamp, node identifier, period identifier, and validity stamp. Record types must be distinguished into at least forward time transfer records, reverse time transfer records, quantum correlation hit records, node running status records, and environmental status records. Integrity verification and validity identification are performed on each type of record.

3. The quantum measurement link clock synchronization calibration method according to claim 1, characterized in that, Time series processing based on a unified time base includes: The data sets merged into the current calibration period are time-mapped according to a unified time base, and then sorted by time markers after being merged by period identifier; Establish candidate correspondence indexes for forward time transfer records, reverse time transfer records, and quantum correlation measurement records that are close in time location; Based on the periodic set and boundary buffer, the boundary records are assigned affiliation, and abnormal records are cached and diverted to form a periodic validity mark. The results of the data processing are used to create a structured dataset, which includes bidirectional time transfer sequences, quantum correlation measurement sequences, link state sequences, anomaly record sequences, and period validity markers.

4. The quantum measurement link clock synchronization calibration method according to claim 1, characterized in that, Based on the correspondence between the two-way observation records, the total time deviation within the current calibration period is stratified and classified, including: The bidirectional time transfer sequence is divided into a forward record set and a reverse record set according to the direction marking, and the valid records in the forward record set, reverse record set and quantum correlation measurement sequence are merged according to a unified time reference and period identifier; Observation correspondence groups are established based on temporal proximity, and a total time deviation characterization value is generated for each observation correspondence group; Using the total time deviation as input, the system sequentially performs directional symmetry discrimination, node consistency discrimination, and intra-period continuity discrimination.

5. The quantum measurement link clock synchronization calibration method according to claim 1, characterized in that, Generate candidate quantities for real clock skew, link bidirectional propagation asymmetric drift, and device internal delay drift, including: The deviation components after hierarchical discrimination are organized into candidate quantities to form a candidate quantity structure with real clock deviation candidate quantity, link bidirectional propagation asymmetric drift candidate quantity, device internal delay drift candidate quantity, local disturbance mark, candidate quantity source mark, candidate quantity confidence level and current calibration cycle discrimination mark; Local disturbance markers correspond to deviation components that do not form a continuous relationship, while candidate quantity confidence levels and current calibration cycle discrimination markers are used to characterize the usage conditions of candidate quantities.

6. The quantum measurement link clock synchronization calibration method according to claim 1, characterized in that, Based on the deviation succession relationship between adjacent calibration cycles and the current operating status, calibration determination is performed on the candidate quantities of real clock deviation, link bidirectional propagation asymmetric drift, and device internal delay drift, including: A continuous judgment window is established around the current calibration cycle. The candidate quantities of the real clock deviation are judged in accordance with the direction of change, the magnitude of change, and the degree of continuous stability. The candidate quantities of bidirectional propagation asymmetric drift of the link and the candidate quantities of internal delay drift of the device are constrained and judged in combination with the link operation status. Then, according to the state machine rules, the current calibration cycle is transferred to the pending update state, the observation and storage state, or the frozen and held state, and the cases of record interruption, high boundary buffer participation, and missing state records are judged to be downgraded, temporarily suspended, or rebuilt.

7. The quantum measurement link clock synchronization calibration method according to claim 1, characterized in that, Perform clock synchronization calibration updates based on the calibration determination results, including: Based on the current calibration cycle's determination status, a calibration update record is generated, and the actual clock deviation determination value is written into the candidate version area; Candidate versions are assigned version statuses such as pending, officially effective, under observation, and frozen. Candidate versions are then processed for official status, extended confirmation, or invalidation according to continuous confirmation conditions, single update magnitude constraints, and status transition rules.

8. The quantum measurement link clock synchronization calibration method according to claim 1, characterized in that, Record the bidirectional propagation asymmetric drift and internal device delay drift, including: The asymmetric drift of bidirectional propagation of the link and the internal delay drift of the device are recorded separately to form a drift record area that is independent of the formal calibration version area and is associated with the period number, version number and status flag. Observation records are generated in the suspended state, and frozen records are generated in the frozen state. The drift history for each calibration cycle is organized and updated based on drift type, drift magnitude, confidence level, associated operating status, observation reason, freeze reason, number of observations, and freeze start marker.

9. The quantum measurement link clock synchronization calibration method according to claim 1, characterized in that, In the event of link malfunction, sudden environmental changes, or bidirectional observation mismatch, maintain the current calibration version and re-enter the next calibration cycle after recovery, including: After identifying link operation abnormalities, sudden changes in environmental state, or bidirectional observation mismatch, the current state is switched to the freeze hold state according to the rule that a severe abnormality triggers a freeze once or a weak abnormality triggers a freeze continuously. The current formal calibration version is maintained, the candidate version is stopped, and the abnormality type, freeze start information, and drift change information are recorded. Once the cause of the anomaly disappears and the pairing conditions are met again in both directions, the current state will be switched to the recovery observation state, and the recovery determination will be made based on the continuous observation conditions and the succession relationship. If no official calibration version is available, the current state is switched to the reconstruction state, and a new official calibration version is generated based on the continuous reconstruction conditions.

Images