Unmanned aerial vehicle quantum communication adaptive phase tracking and data post-selection method and system

Abstract

The application discloses a kind of unmanned plane quantum communication adaptive phase tracking and data post-selection method and system, belong to quantum communication and unmanned plane communication technical field, by extracting the link state representation quantity of current link, according to current link state adaptive update process noise covariance, and based on updated process noise covariance, Kalman recursion is carried out, obtains posteriori phase estimation value and posteriori estimation error covariance;Then based on posteriori estimation error covariance and link quality index, construct the screening criterion that changes with current estimation uncertainty dynamically, execute dynamic post-selection to data frame;Finally, information recovery processing is carried out to the data frame reserved.The application solves the problem that existing fixed parameter filtering is difficult to consider sudden response and steady accuracy under non-stationary channel, fixed threshold screening is easy to produce false rejection and false admission, improves the phase estimation accuracy and effective data utilization rate under non-stationary atmospheric channel.

Description

Technical Field

[0001] This invention relates to the fields of quantum communication and UAV communication technology, and in particular to a method and system for adaptive phase tracking and data post-selection in UAV quantum communication. Background Technology

[0002] The statements in this section are merely background information related to the present invention and do not necessarily constitute prior art.

[0003] In the scenario of quantum-secure direct communication of continuous variables in free space by UAVs, the optical signal propagates through a non-stationary atmospheric channel and is easily affected by factors such as atmospheric turbulence, wavefront distortion, platform vibration, attitude changes and relative motion. This results in significant phase random drift and transmittance fluctuations at the relay measurement node. These disturbances reduce the accuracy of phase compensation and the availability of effective data, and further affect the reliability of subsequent information recovery.

[0004] In measurement-device-independent scenarios, two continuous-variable quantum signals from the first and second transmitters are transmitted to a relay measurement node via their respective independent free-space links. Because the propagation environment, platform attitude, local turbulence intensity, and beam tracking state of the two links are typically inconsistent, the dual signals often exhibit inconsistencies and time-varying characteristics in phase shift, amplitude fluctuations, and transmittance changes upon arrival at the relay measurement node. When the relay measurement node performs joint measurements, the system not only needs to suppress measurement degradation caused by single-link disturbances but also needs to handle the time-varying relative phase relationship between the two signals, the coupling effects caused by asynchronous disturbances, and asymmetric link quality. The post-processing challenge in this scenario lies in achieving reliable information recovery based on publicly available measurement results from the relay end under conditions of independent non-stationary disturbances. Furthermore, since the relay measurement node is not a reliable basis for system security, the system cannot rely on the ideal stability of the measurement node itself to guarantee subsequent recovery quality. Instead, it needs to use digital signal processing after joint measurement to promptly perceive the current link state, robustly estimate phase drift, and effectively filter out low-quality data frames.

[0005] In existing technologies, standard Kalman filtering can be used to estimate and compensate for phase random drift. Some methods set the process noise covariance to a fixed constant, which is equivalent to assuming that the channel statistical characteristics are approximately stationary throughout the communication process. However, when UAVs are maneuvering at high speeds, experiencing wind shear, enhanced local turbulence, or sudden increases in beam tracking errors, link disturbances exhibit significant time-varying characteristics. Fixing the process noise covariance is difficult to adapt to both the stationary and abrupt phases simultaneously: when its value is small, the filter lags in responding to sudden disturbances, resulting in untimely phase tracking; when its value is large, the filtering result is prone to large estimation fluctuations in the stationary phase, thereby reducing the accuracy of phase estimation.

[0006] On the other hand, existing solutions often employ a fixed threshold data selection strategy, which involves pre-setting a single threshold and discarding data frames exceeding that threshold. This type of method uses a uniform threshold to process data under different link conditions, failing to consider the dynamic changes in estimation confidence and link quality over time. If a fixed threshold is still used, valid data may be mistakenly discarded when link quality is good, while low-quality data may be mistakenly accepted when link quality is poor. This reduces the valid data retention rate and affects the reliability of subsequent information recovery and transmission performance. Summary of the Invention

[0007] To address the shortcomings of existing technologies, this invention provides an adaptive phase tracking and post-selection method and system for UAV quantum communication. By extracting link state characteristics and adaptively updating the process noise covariance based on the current link state, it solves the problem that fixed parameter filtering struggles to balance burst response and steady-state accuracy in non-stationary channels. By constructing a selection criterion that dynamically changes with estimation uncertainty to perform post-selection of data frames, it solves the problem of false rejection and false acceptance caused by fixed threshold filtering. By sequentially link state perception, parameter update, phase estimation, and dynamic filtering to form a processing chain, phase tracking and post-selection work synergistically, thereby improving the phase estimation accuracy and effective data utilization in non-stationary atmospheric channels.

[0008] On the one hand, an adaptive phase tracking and data post-selection method for UAV quantum communication is provided, including: The original analog photocurrent signal of the balanced zero-difference detection at the relay end is acquired, and the original analog photocurrent signal is preprocessed to obtain a discrete observation sequence. Based on the discrete observation sequence, a phase-drift discrete state-space model is established, which includes state equations and measurement equations. Based on the state space model, the link state representation of the current link is extracted. The process noise covariance is adaptively updated according to the current link state represented by the link state representation. Then, Kalman recursion is performed based on the updated process noise covariance to obtain the posterior phase estimate and the posterior estimation error covariance. Based on the posterior estimation error covariance and link quality index, a selection criterion that dynamically changes with the current estimation uncertainty is constructed, and dynamic post-selection is performed on data frames according to the selection criterion to obtain a set of retained data frames. Information recovery processing is performed on the retained data frame set to obtain the communication recovery result.

[0009] Furthermore, the preprocessing includes synchronization processing, DC removal processing, normalization processing, and frame segmentation processing.

[0010] Furthermore, the state equation is:

[0011] The measurement equation is:

[0012] in, for The true phase drift state at any given moment. for The true phase drift state at time -1 for The measured value at time, For process noise, To measure noise, For the observation matrix, Let be the state transition matrix.

[0013] Furthermore, the link state representation quantity is specifically as follows: ; in, These are non-negative weighting coefficients. For normalized mapping functions, For new information statistics Instantaneous transmittance, This is an indicator of the amplitude of new information.

[0014] Furthermore, the adaptive update of the process noise covariance is specifically as follows: Update the process noise covariance as follows: : ; in, This is the link state characterization quantity at time k, used to characterize the degree of change in the dual-link state and phase observation; , and These are the preset low process noise covariance, medium process noise covariance, and high process noise covariance, respectively, and satisfy the following conditions: < < ; and They are respectively the first preset threshold and the second preset threshold, and satisfy the following conditions: < .

[0015] Furthermore, based on the posterior estimation error covariance and link quality indicators, a selection criterion that dynamically changes with the current estimation uncertainty is constructed. Dynamic post-selection of data frames is then performed according to the selection criterion to obtain a set of retained data frames, specifically: A dynamic filtering threshold is constructed based on the posterior estimation error covariance, and the dynamic filtering threshold is positively correlated with the posterior estimation error covariance; the innovation of the current data frame is calculated, and the absolute value of the innovation is compared with the dynamic filtering threshold: if the absolute value of the innovation is less than or equal to the dynamic filtering threshold, the data frame is retained; if the absolute value of the innovation is greater than the dynamic filtering threshold, the data frame is discarded.

[0016] Furthermore, the dynamic post-selection also includes: A weighted fusion quality score is constructed, which includes instantaneous transmittance, residual phase quality index, and signal-to-noise ratio. The weighted fusion quality score is then compared with a preset threshold. If the weighted fusion quality score is greater than or equal to the preset threshold, the data frame is retained. If the weighted fusion quality score is less than the preset threshold, the data frame is discarded.

[0017] Furthermore, the information recovery process includes at least one of parameter negotiation processing, error correction processing, statistical feature calculation, security capacity estimation, or confidential information recovery.

[0018] On the other hand, an adaptive phase tracking and data post-selection system for UAV quantum communication is also provided, including: The data preprocessing module is configured to: acquire the raw analog photocurrent signal of the relay-end balanced zero-difference detection, and perform preprocessing on the raw analog photocurrent signal to obtain a discrete observation sequence; The state-space modeling module is configured to: establish a phase-drift discrete state-space model based on the discrete observation sequence, wherein the state-space model includes state equations and measurement equations; The adaptive filtering module is configured to: extract the link state representation of the current link based on the state space model; adaptively update the process noise covariance according to the current link state represented by the link state representation; and then perform Kalman recursion based on the updated process noise covariance to obtain the posterior phase estimate and the posterior estimation error covariance. The dynamic post-selection module is configured to: construct a selection criterion that dynamically changes with the current estimation uncertainty based on the posterior estimation error covariance and link quality index, and perform dynamic post-selection on data frames according to the selection criterion to obtain a set of retained data frames; The information recovery module is configured to perform information recovery processing on the retained data frame set to obtain the communication recovery result.

[0019] In another aspect, a computer-readable storage medium is also provided, on which a computer program is stored, which, when executed by a processor, performs the method described in the first aspect.

[0020] The above technical solution has the following advantages or beneficial effects: (1) This invention introduces a process noise covariance that can be dynamically updated with the channel state, enabling the phase tracking process to adaptively adjust the filtering parameters according to the current link disturbance level. In non-stationary scenarios such as strong disturbances, sudden fading, or enhanced platform vibration, it can improve the response speed to phase changes; when the link is relatively stable, it can reduce unnecessary estimation fluctuations, thereby improving the phase estimation accuracy and alleviating the problem that existing fixed process noise covariance methods cannot balance dynamic response performance and steady-state estimation accuracy.

[0021] (2) This invention introduces a dynamic post-selection mechanism based on the covariance of posterior estimation errors, enabling the screening threshold to automatically change with the current estimation uncertainty. Simultaneously, it can combine instantaneous transmittance, residual phase quality indicators, and signal-to-noise ratio metrics for joint decision-making. Compared with existing fixed-threshold post-selection methods, this invention can more accurately distinguish between valid and low-quality data, reduce the erroneous rejection of valid data in high-quality link stages, and lower the probability of erroneous acceptance of low-quality data in poor-quality link stages, thereby improving the utilization rate of valid data and the reliability of subsequent information recovery.

[0022] (3) By constructing a unified state perception and adjustment framework compatible with single state representation quantities, composite state representation quantities, and different parameter update strategies, this invention avoids the excessive reliance of existing methods on single link state determination criteria and improves the consistency between state determination, parameter update, and data screening. Compared with schemes that rely solely on a single innovation quantity, a single transmittance index, or a single empirical rule, this invention can flexibly select state determination criteria and parameter adjustment methods according to different link conditions, different available state information, and different update requirements, thereby enhancing the method's adaptability, transferability, and robustness to complex dual-path non-stationary disturbances in measurement device-independent relay measurement scenarios.

[0023] (4) This invention organizes data preprocessing, state characterization construction, adaptive phase tracking, dynamic screening and decision-making, and information recovery processing into a continuous processing flow, enabling the phase estimation results and the data post-selection results to work synergistically. Compared with existing processing methods where phase compensation and post-selection are independent, this invention is beneficial for simultaneously improving phase estimation accuracy, effective data retention, and information recovery stability, thereby improving the availability of measurement results and the stability of subsequent message recovery. Because this invention places link state awareness before subsequent parameter updates and dynamic screening, it can ensure that each processing step has the input basis of the previous step. First, the current link conditions are determined through state identification, then the phase estimation process is changed through parameter updates, and then the output of estimation uncertainty supports dynamic post-selection, ultimately improving the quality of subsequent information recovery.

[0024] (5) The method flow of the present invention is clear and the functions are clearly divided, which makes it easy to implement in a digital signal processing device as a data preprocessing module, a state perception module, an adaptive filtering module, a dynamic post-selection module and an information recovery module. It has good engineering feasibility and system integration value.

[0025] (6) Without changing the basic system architecture of measurement device independence, this invention addresses the problem of inconsistent disturbances generated at relay measurement nodes in dual-path independent free-space links. It enhances the system's ability to perceive the current link state, the basis for parameter updates, and the targeting of data filtering, thereby improving the availability of publicly available Bell measurement results for Bob-side message recovery and the reliability of subsequent information recovery. Simultaneously, the method of this invention operates on the digital signal processing and data filtering stages after joint measurement by relay measurement nodes. It does not change the fundamental setting in the measurement device independence architecture that joint measurement is performed by relay measurement nodes and is not used as a trusted security basis. Therefore, it can improve the robustness of information recovery while maintaining the core security premise of measurement device independence (MDI). Attached Figure Description

[0026] The accompanying drawings, which form part of this invention, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an improper limitation of the invention.

[0027] Figure 1 This is a schematic diagram of the UAV quantum direct communication system architecture in Embodiment 1 of the present invention; Figure 2 This is a flowchart of the method in Embodiment 1 of the present invention; Figure 3 This is a time-domain comparison diagram of the standard Kalman filter and the adaptive Kalman filter of this invention under the dual-path non-stationary link condition in Embodiment 1 of this invention. Figure 3 (a) in the figure shows the true relative phase, the phase estimation results of the standard Kalman filter, and the phase estimation results of the adaptive Kalman filter; Figure 3 (b) in the figure shows the phase estimation error diagrams for the two filtering methods; Figure 3 (c) in the figure represents the reference process noise covariance and the adaptively updated process noise covariance plot; Figure 3 In the diagram, (d) represents the instantaneous transmittance of the Alice-Charlie link and the Bob-Charlie link, the smaller value of the instantaneous transmittance of the two links, and the adaptive Kalman filter trigger threshold. Detailed Implementation

[0028] To make the objectives, technical solutions, and advantages of the present invention clearer, the embodiments of the present invention will be described in further detail below with reference to the accompanying drawings. Those skilled in the art should understand that the specific embodiments described herein are for illustrative purposes only and are not intended to limit the scope of the invention.

[0029] It should be noted that the following detailed description is illustrative and intended to provide further explanation of the invention. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0030] Example 1 This embodiment provides an adaptive phase tracking and data post-selection method for UAV quantum communication, applicable to measurement device-independent continuous-variable quantum secure direct communication scenarios, such as... Figure 1 As shown, in this scenario, the first transmitter (Alice) and the second transmitter (Bob), as transmitting users, send quantum signals to the untrusted relay measurement terminal (Charlie) through their respective independent free-space atmospheric channels (optical signal channel A and optical signal channel B). Charlie inputs the two quantum signals into a 50:50 beam splitter for interference and performs balanced homodyne detection on the two output terminals to obtain the original photocurrent sequence. Based on this original photocurrent sequence, Charlie forms and publishes the Bell measurement record. The legitimate user side performs state characterization, phase tracking, post-selection, and information recovery, which helps improve the phase tracking stability, effective data utilization, and information recovery reliability under non-stationary channel conditions. The core processing logic of this invention is as follows: first, the current disturbance level and link quality status are obtained through link state perception; then, the process noise covariance is adaptively updated accordingly; and the posterior estimation error covariance is output based on the updated phase estimation result. Furthermore, dynamic data post-selection is performed in conjunction with relevant quality indicators. This forms a continuous processing chain of "state perception, parameter update, phase estimation, uncertainty characterization, dynamic filtering, and information recovery," enabling subsequent steps to be based on the state information or estimation results output by the preceding steps.

[0031] Figure 2 This is a flowchart of a method according to an embodiment of the present invention. The method includes the following steps: S101: Acquire the raw analog photocurrent signal of the balanced zero-difference detection at the relay end, and perform preprocessing on the raw analog photocurrent signal to obtain a discrete observation sequence; S102: Based on discrete observation sequences, establish a phase drift discrete state-space model, which includes state equations and measurement equations; S103: Based on the state-space model, extract the link state representation of the current link, adaptively update the process noise covariance according to the current link state represented by the link state representation, and then perform Kalman recursion based on the updated process noise covariance to obtain the posterior phase estimate and the posterior estimation error covariance. S104: Based on the posterior estimation error covariance and link quality indicators, construct a selection criterion that dynamically changes with the current estimation uncertainty, and perform dynamic post-selection on data frames according to the selection criterion to obtain a set of retained data frames; S105: Perform information recovery processing on the retained data frame set to obtain the communication recovery result.

[0032] Specifically, in step S101, in order to solve the problem of phase tracking and data post-selection being mutually isolated and lacking a coordination mechanism in the prior art, this step first acquires the original analog photocurrent signal output by the balanced zero-difference detector at the relay measurement node, and performs preprocessing on the signal to form a unified and comparable discrete observation sequence, providing a consistent data input basis for subsequent link state sensing, phase estimation and dynamic screening.

[0033] In a measurement device-independent relay measurement scenario, two continuous-variable quantum signals from the first and second transmitters are transmitted to the relay measurement node via independent free-space links. The non-uniform disturbances of the two independent links introduce additional difficulties to the signal processing after joint measurement. Meanwhile, the relay measurement node does not serve as a trusted basis for system security. Therefore, the technical solution of this invention does not rely on changing the system role of the relay measurement node. Instead, without changing the architecture setting that the relay measurement node performs joint measurement and does not serve as a trusted security basis, it performs digital signal processing on the publicly available measurement results through the legitimate user side to track and compensate for the relative phase shift caused by the non-uniform disturbances of the two links, and completes data post-selection by combining link quality information.

[0034] Based on the above considerations, the relay measurement node performs balanced zero-difference detection on the received optical signal and outputs the original analog photocurrent signal. After sampling and analog-to-digital conversion, the original analog photocurrent signal is input into the digital signal processing module, which sequentially performs synchronization processing, DC removal processing, normalization processing, and framing processing to obtain the discrete observation sequence.

[0035] The original photocurrent sequence often contains problems such as inconsistent sampling scale, baseline offset and inconsistent time alignment. If it is used directly for state estimation and data discrimination, it is easy to introduce non-target disturbances into the subsequent processing. Therefore, the above preprocessing steps provide a consistent data basis for subsequent state determination and estimation recursion by forming a unified and comparable discrete observation sequence.

[0036] Let discrete time k represent the kth sampling time or the index of the kth data frame. Then, the preprocessed discrete observation sequence is used as the input for subsequent phase drift state estimation and data quality judgment, and is used in step S102 to establish the phase drift state space model.

[0037] In step S102, in order to address the problem that existing phase tracking and post-selection methods are heavily dependent on single-state representation quantities and lack the ability to uniformly adapt to different channel state determination criteria and parameter update methods, in step S102, based on the discrete observation sequence obtained in step S101, a discrete state space model of phase drift on the joint measurement side in the relay measurement node is established to describe the temporal evolution of phase state and measurement relationship.

[0038] Specifically, set Indicates the first The true phase drift state at any given moment. Indicates the first The measured value at time, The process noise is represented by a mean of 0 and a covariance of . The normal distribution The measurement noise is represented by a mean of 0 and a variance of 1. R The normal distribution Represents the process noise covariance. Indicates the variance of measurement noise. Let the observation matrix be represented, then the state equation and the measurement equation are respectively:

[0039]

[0040] Where N represents the normal distribution, F represents the state transition matrix, and H represents the observation matrix. For the phase random walk model, F can be taken as 1, and H can be taken as 1 when the measured value directly corresponds to the phase state.

[0041] This state-space model describes the temporal evolution of phase drift states between adjacent time points through state equations, and establishes a linear mapping relationship between phase states and observed values ​​through measurement equations, providing a unified mathematical foundation for subsequent link state sensing, adaptive update of process noise covariance, and Kalman recursion. Step S103 processes the state-space model and discrete observation sequences for link state sensing, process noise covariance update, and phase estimation recursion.

[0042] In step S103, to address the problem that existing standard Kalman filters struggle to balance the accuracy of sudden disturbance response and steady-state estimation due to the fixed setting of process noise covariance, and to solve the lack of unified adaptability of existing methods to different link state perception bases and parameter update methods, this step, based on the state-space model established in step S102, the current observations, and the Kalman prior prediction results, extracts the link state representation of the current link. Based on the represented current link state, the process noise covariance is then analyzed. An adaptive update is performed, and then Kalman recursion is executed based on the updated process noise covariance to obtain the posterior phase estimate and the posterior estimation error covariance. Under non-stationary link conditions, the disturbance intensity and phase change rate are not the same in different stages. If a fixed process noise covariance is always used, it is difficult to balance the response capability in the strong disturbance stage with the estimation smoothness in the stationary stage. Therefore, it is necessary to first sense the current link state and then adjust the filtering parameters accordingly to keep the estimation process consistent with the current disturbance level.

[0043] Specifically, set express The prior phase estimate at time t is used to define the innovation as: ; In length of Calculate the innovation statistic within a sliding window :

[0044] in, i Let be the innovation value at time i, representing the difference between the actual observation value at time i and the predicted observation value obtained based on prior phase estimation.

[0045] Furthermore, combined with instantaneous transmittance Link quality metrics Construct link state representation quantities:

[0046] in, These are non-negative weighting coefficients. This is the normalized mapping function. The above link state representation quantities are weighted and fused from the innovation statistics, instantaneous transmittance and link quality indicators to represent the current link state from three aspects: phase perturbation, link fading and observation quality, avoiding the representation distortion that may occur under complex perturbation scenarios with a single indicator.

[0047] Q k Based on link state characterization With preset threshold as well as The relationship determines the current link state, and the process noise covariance is updated as follows. :

[0048] in, , For a preset threshold and < , , and These represent the low, medium, and high process noise covariances corresponding to relatively stable link status, moderate fluctuation, and large fluctuation, respectively, and satisfy the following conditions: The physical meaning of this update rule is that when the link state representation is small, it indicates that the current link is relatively stable, and a smaller process noise covariance is adopted. This can make the filtering results smoother and reduce steady-state estimation fluctuations; when the link state characterization is in a medium range, it indicates that there is a certain disturbance in the link, and a medium process noise covariance is used. A compromise is made: when the link state representation is large, it indicates that the current link is in a state of strong disturbance, and a larger process noise covariance is adopted. This improves the filter's response speed to phase changes, thereby enabling adaptive tracking.

[0049] The above three-stage switching rule has a simple structure and only requires two preset thresholds to cover three typical link states. In engineering implementation, it has the advantages of low computational load and timely response, and is especially suitable for scenarios where the computing resources of the UAV platform are limited.

[0050] In one implementation, the link state representation quantity Defined as the normalized square of innovation, i.e. = ,in, This represents the innovation quantity between pilot phase observations and predictions. To innovate covariance. Based on degrees of freedom The chi-square distribution can be preset with a first threshold. Second threshold =3.84. When When the current phase observation deviation is within one standard deviation, it indicates that the link is in a relatively stable state; when 1.00 At 3.84, it indicates that there is a general disturbance in the link; when At 3.84, it indicates that the current innovation exceeds the 95% confidence range of normal random fluctuations, and the link enters a state of strong disturbance.

[0051] After completion After the update, execute the Kalman recursion:

[0052] in, Let K be the Kalman gain matrix at time k. It is the identity matrix that matches the phase state dimension.

[0053] This step outputs the posterior phase estimate. and posterior estimation error covariance ,in Used for phase compensation The output is sent to step S104 as one of the criteria for dynamic post-selection.

[0054] In other implementations, the link state perception criterion can be any one of the following: normalized innovation statistics, dual-link transmittance obtained from pilot estimation, pilot signal-to-noise ratio estimate, phase estimation error covariance, or phase perturbation index constructed based on pilot observations. Alternatively, it can be a composite state representation formed by normalizing, weighted fusion, joint scoring, or other combinations of the above at least two indicators. The parameter update method can be segmented switching, continuous adjustment, or a combination of both. Through the above settings, different state determination criteria, different state representation forms, and different parameter update methods can all be integrated into the subsequent phase tracking and dynamic post-selection process under a unified processing framework, thereby improving the method's ability to uniformly adapt to different link state perception methods and different update strategies.

[0055] Step S104: To address the issue that existing fixed-threshold selection methods cannot adaptively adjust the selection rules based on phase estimation confidence and time-varying link quality characteristics, this step is based on the posterior estimation error covariance output in step S103. Based on link quality indicators, dynamic post-selection is performed on data frames to obtain a set of retained data frames. After phase compensation, the availability of different data frames may still vary significantly. If a uniform static threshold is continued to be used, it will be difficult to distinguish between data with reliable estimation and good link quality and data with unstable estimation or link degradation. Therefore, the selection rules need to be changed synchronously with the current estimation uncertainty and link quality to improve the consistency between data discrimination and the current state.

[0056] Specifically, let the dynamic screening threshold be... The main implementation method based on the posterior estimation error covariance is: ; in, This is the threshold adjustment coefficient. Using the information gating method, then for the... Each data frame executes the following decision: If , retain the data frame; if The data frame is discarded. The physical meaning of this decision method is as follows: when the uncertainty of the predicted observation is large or small, it indicates that the current phase estimate has a high degree of confidence. At this time, a smaller screening threshold can effectively remove data frames with abnormal innovation residuals. When the covariance of the posterior estimation error is large, it indicates that the current estimation uncertainty is high. It is necessary to appropriately relax the screening threshold to avoid misjudging the fluctuation of innovation residuals caused by link disturbances as unusable data, thereby retaining more potentially valuable data frames under the condition of high estimation uncertainty.

[0057] As an extended implementation, instantaneous transmittance can also be incorporated. New information amplitude indicator and signal-to-noise ratio characterization Construct a joint quality score: ; in, These are non-negative weighting coefficients. Let be the mapping function constructed from the signal-to-noise ratio (SNR) characteristic. .according to With preset threshold Relationship execution judgment: If If The data frame is then discarded. This extended implementation introduces multi-dimensional link quality metrics, enabling a more comprehensive evaluation of the availability of each data frame, particularly suitable for scenarios with asymmetrical quality between two links. When only one link in the dual-link system experiences a strong disturbance at a given moment, a decision based on a single metric might render the entire frame unusable. However, the multi-dimensional joint quality score can more accurately reflect the actual availability of the frame, thereby improving the accuracy and robustness of dynamic post-selection. This step outputs a set of retained data frames for subsequent information recovery processing.

[0058] The weighting coefficients and decision thresholds are determined based on simulation or calibration results, taking into account the impact of each indicator on information recovery performance and data frame retention rate. In one specific embodiment, the following is taken: , , , The decision threshold is taken as follows =0.50.

[0059] This step outputs a set of retained data frames for subsequent information recovery processing.

[0060] In step S105, to unify the effects of preceding link state awareness, parameter update, phase estimation, and dynamic data filtering into the final communication recovery result, this step performs subsequent information recovery processing on the data frames retained in step S104 to obtain the communication recovery result. The purpose of phase estimation and dynamic post-selection is not merely to improve intermediate processing results, but to provide more stable and reliable data input for subsequent recovery stages. Therefore, information recovery, as the final link in the entire processing chain, requires the phase compensation result output in step S103 and the set of retained data frames output in step S104 as input. Through at least one operation among parameter negotiation processing, error correction processing, or confidential information recovery, the improvement effects of preceding steps on phase estimation accuracy and effective data utilization are transformed into the final communication recovery result. Further statistical feature calculations or security capacity estimation can be performed to further transform the improvement effects of preceding steps on phase estimation accuracy and effective data utilization into the final communication recovery result. This step outputs the information recovery result to complete subsequent communication processing.

[0061] Connecting steps S101 to S105 sequentially forms a complete joint measurement post-processing flow for measurement device-independent relay measurement scenarios. Specifically, step S101 preprocesses the original analog photocurrent signal into a unified discrete observation sequence, providing a consistent data input basis for subsequent processing; step S102 establishes a phase drift discrete state-space model based on this observation sequence, providing a mathematical framework for state estimation; step S103 adaptively updates the process noise covariance based on the link state characterization and performs Kalman recursion, outputting the posterior phase estimate. Used for phase compensation, and also outputs the posterior estimation error covariance. Used to characterize estimation uncertainty; step S104 is based on innovation. The prediction uncertainty corresponding to the new information and the link quality indicators are dynamically selected after execution, and the set of retained data frames is output; step S105 performs information recovery processing on the retained data frames and outputs the final communication recovery result.

[0062] The above steps are interconnected and progressive, with the output of the previous step serving as the input for the next. Together, they form a joint measurement post-processing chain that includes "state awareness, parameter update, phase estimation, uncertainty characterization, dynamic filtering, and information recovery." This allows phase tracking and data post-selection to work synergistically within the same processing framework, thereby effectively improving the accuracy of phase estimation, effective data utilization, and reliability of information recovery in non-stationary atmospheric channels while maintaining the core safety premise of measurement equipment independence.

[0063] To verify the effectiveness of the adaptive Kalman phase tracking method described in this invention under dual-path non-stationary free-space link conditions, this embodiment constructs a numerical simulation sequence of 150,000 pilot pulses and selects the first 1,500 pilot pulses to display the time-domain results. Pilot pulses 1 to 499 and 1001 to 1500 are set as the relatively stationary phase of the link, while pilot pulses 500 to 1000 are set as the strongly disturbed phase. The relative phase state is generated according to a random walk model, and the reference process noise covariance of the stationary phase is set to 0.001 rad. 2 The reference process noise covariance during the strong disturbance phase is set to 0.06 rad. 2 The measurement noise variance was set to 0.1 rad. 2 .

[0064] During the relatively stable phase of the link, the instantaneous transmittance of the Alice-Charlie and Bob-Charlie links fluctuates randomly around 0.95. During the strong disturbance phase, the instantaneous transmittance of the two links is generated according to independent log-normal distributions and is limited to the range of 0 to 1. The smaller of the instantaneous transmittance of the two links is used as the link disturbance trigger value, and the adaptive Kalman filter trigger threshold is set to 0.15. When the instantaneous transmittance of either link falls below the trigger threshold, the filter enters the disturbance state. The standard Kalman filter uses a fixed process noise covariance of 0.001 rad. 2 The adaptive Kalman filter uses a sliding window of length 50 to calculate the innovation power and adaptively updates the process noise covariance based on the innovation power and the instantaneous transmittance of the two links, with the update range limited to 0.001 rad. 2 up to 0.08 rad 2 .

[0065] Figure 3 Time-domain comparison results of standard Kalman filtering and the adaptive Kalman filtering of this invention under dual-path non-stationary link conditions are presented. Figure 3 (a) shows the true relative phase, the standard Kalman filter phase estimation results, and the adaptive Kalman filter phase estimation results; Figure 3 (b) shows the phase estimation errors corresponding to the two filtering methods; Figure 3 (c) shows the reference process noise covariance and the adaptively updated process noise covariance; Figure 3 (d) shows the instantaneous transmittance of the Alice-Charlie link, the Bob-Charlie link, the smaller value of the instantaneous transmittance of the two links, and the adaptive Kalman filter trigger threshold.

[0066] Depend on Figure 3As shown in (a), during the relatively stable link phase, both the standard Kalman filter and the adaptive Kalman filter can track the true relative phase well. However, after entering the strong disturbance phase, the amplitude and speed of the change in the true relative phase increase significantly. The standard Kalman filter, due to its use of a fixed process noise covariance, is relatively slow to respond to rapid phase changes. In contrast, the adaptive Kalman filter of this invention can adjust the process noise covariance in a timely manner according to the current link state, and its estimation results maintain better consistency with the true relative phase.

[0067] Depend on Figure 3 As shown in (b), during the strong disturbance phase, the phase estimation error of the standard Kalman filter increases significantly and exhibits more error peaks, while the overall estimation error of the adaptive Kalman filter of this invention is smaller, indicating that the noise covariance of the adaptive update process can improve the filter's ability to track sudden phase changes. After the link returns to a stable state, the estimation errors of both methods gradually decrease.

[0068] Depend on Figure 3 As shown in (c), during the relatively stable phase of the link, the adaptively updated process noise covariance remains at a low level, which helps to suppress unnecessary estimation fluctuations. When the link enters a strong disturbance phase, the adaptively updated process noise covariance increases accordingly, thereby improving the filter's response speed to phase state changes. When the disturbance ends, the process noise covariance gradually recovers to a low level, indicating that the method of the present invention can dynamically adjust the filtering parameters according to the link disturbance state.

[0069] Depend on Figure 3 As shown in (d), during the strong disturbance phase, the instantaneous transmittance of both the Alice-Charlie and Bob-Charlie links exhibits significant independent random fluctuations, with the smaller values ​​of the two instantaneous transmittances repeatedly falling below the preset trigger threshold. At this point, the adaptive Kalman filter enters the disturbance state and increases the process noise covariance; once the instantaneous transmittance of both links recovers and the innovation power decreases, the filter exits the disturbance state. These results demonstrate that the present invention can comprehensively utilize the transmittance and innovation statistics of both links to identify the current link disturbance state and adjust the process noise covariance accordingly, thus balancing the estimation accuracy during the stable phase and the dynamic response capability during the strong disturbance phase.

[0070] Example 2 This embodiment provides an adaptive phase tracking and data post-selection system for UAV quantum communication, including: The data preprocessing module is configured to: acquire the raw analog photocurrent signal of the relay-end balanced zero-difference detection, and perform preprocessing on the raw analog photocurrent signal to obtain a discrete observation sequence; The state-space modeling module is configured to: establish a phase-drift discrete state-space model based on discrete observation sequences. The state-space model includes state equations and measurement equations. The adaptive filtering module is configured to: extract the link state representation of the current link based on the state space model, adaptively update the process noise covariance according to the current link state represented by the link state representation, and then perform Kalman recursion based on the updated process noise covariance to obtain the posterior phase estimate and the posterior estimation error covariance. The dynamic post-selection module is configured to: construct a selection criterion that dynamically changes with the current estimation uncertainty based on the posterior estimation error covariance and link quality indicators, and perform dynamic post-selection on data frames according to the selection criterion to obtain a set of retained data frames; The information recovery module is configured to perform information recovery processing on the retained data frame set to obtain the communication recovery result.

[0071] It should be noted that each module in this embodiment corresponds one-to-one with each step in Embodiment 1, and their specific implementation process is the same, so it will not be repeated here.

[0072] Example 3 This embodiment also provides a computer-readable storage medium for storing computer instructions, which, when executed by a processor, complete the method described in Embodiment 1.

[0073] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. An adaptive phase tracking and data post-selection method for UAV quantum communication, characterized in that, include: The original analog photocurrent signal of the balanced zero-difference detection at the relay end is acquired, and the original analog photocurrent signal is preprocessed to obtain a discrete observation sequence. Based on the discrete observation sequence, a phase-drift discrete state-space model is established, which includes state equations and measurement equations. Based on the state space model, the link state representation of the current link is extracted. The process noise covariance is adaptively updated according to the current link state represented by the link state representation. Then, Kalman recursion is performed based on the updated process noise covariance to obtain the posterior phase estimate and the posterior estimation error covariance. Based on the posterior estimation error covariance and link quality index, a selection criterion that dynamically changes with the current estimation uncertainty is constructed, and dynamic post-selection is performed on data frames according to the selection criterion to obtain a set of retained data frames. Information recovery processing is performed on the retained data frame set to obtain the communication recovery result; The link state representation quantity is specifically as follows: ； in, These are non-negative weighting coefficients. For normalized mapping functions, For new information statistics Instantaneous transmittance, This is an indicator of the magnitude of the new information. The adaptive update of the process noise covariance is specifically as follows: Update the process noise covariance as follows: : ； in, This is the link state characterization quantity at time k, used to characterize the degree of change in the dual-link state and phase observation; , and These are the preset low process noise covariance, medium process noise covariance, and high process noise covariance, respectively, and satisfy the following conditions: < < ; and They are respectively the first preset threshold and the second preset threshold, and satisfy the following conditions: < ; Based on the posterior estimation error covariance and link quality indicators, a selection criterion that dynamically changes with the current estimation uncertainty is constructed. Dynamic post-selection of data frames is then performed according to this selection criterion to obtain a set of retained data frames, specifically: A dynamic filtering threshold is constructed based on the posterior estimation error covariance, and the dynamic filtering threshold is positively correlated with the posterior estimation error covariance; the innovation of the current data frame is calculated, and the absolute value of the innovation is compared with the dynamic filtering threshold: if the absolute value of the innovation is less than or equal to the dynamic filtering threshold, the data frame is retained; if the absolute value of the innovation is greater than the dynamic filtering threshold, the data frame is discarded. Based on the link state characterization, the noise covariance of the process is adaptively updated and Kalman recursion is performed to output the posterior phase estimate. Used for phase compensation, and also outputs the posterior estimation error covariance. Used to characterize estimation uncertainty.

2. The UAV quantum communication adaptive phase tracking and data post-selection method according to claim 1, characterized in that, The preprocessing includes synchronization processing, DC removal processing, normalization processing, and frame segmentation processing.

3. The UAV quantum communication adaptive phase tracking and data post-selection method according to claim 1, characterized in that, The state equation is: The measurement equation is: in, for The true phase drift state at any given moment. for The true phase drift state at time -1 for The measured value at time, For process noise, To measure noise, For the observation matrix, Let be the state transition matrix.

4. The UAV quantum communication adaptive phase tracking and data post-selection method according to claim 1, characterized in that, The dynamic post-selection also includes: A weighted fusion quality score is constructed, which includes instantaneous transmittance, residual phase quality index, and signal-to-noise ratio. The weighted fusion quality score is then compared with a preset threshold. If the weighted fusion quality score is greater than or equal to the preset threshold, the data frame is retained. If the weighted fusion quality score is less than the preset threshold, the data frame is discarded.

5. The UAV quantum communication adaptive phase tracking and data post-selection method according to claim 1, characterized in that, The information recovery process includes at least one of parameter negotiation processing, error correction processing, statistical feature calculation, security capacity estimation, or confidential information recovery.

6. A UAV quantum communication adaptive phase tracking and data post-selection system employing the method described in claim 1, characterized in that, include: The data preprocessing module is configured to: acquire the raw analog photocurrent signal of the relay-end balanced zero-difference detection, and perform preprocessing on the raw analog photocurrent signal to obtain a discrete observation sequence; The state-space modeling module is configured to: establish a phase-drift discrete state-space model based on the discrete observation sequence, wherein the state-space model includes state equations and measurement equations; The adaptive filtering module is configured to: extract the link state representation of the current link based on the state space model; adaptively update the process noise covariance according to the current link state represented by the link state representation; and then perform Kalman recursion based on the updated process noise covariance to obtain the posterior phase estimate and the posterior estimation error covariance. The dynamic post-selection module is configured to: construct a selection criterion that dynamically changes with the current estimation uncertainty based on the posterior estimation error covariance and link quality index, and perform dynamic post-selection on data frames according to the selection criterion to obtain a set of retained data frames; The information recovery module is configured to perform information recovery processing on the retained data frame set to obtain the communication recovery result.

7. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the steps in the UAV quantum communication adaptive phase tracking and data post-selection method as described in any one of claims 1-5.

Images