A 5G-TSN traffic scheduling dynamic gating method and device based on decoupling mechanism

Abstract

The application discloses a 5G-TSN traffic scheduling dynamic gating method and device based on a decoupling mechanism, and relates to the technical field of communication. The method obtains a clock bias observation value of a 5G air interface synchronization message; according to a crystal oscillator drift state transition model and a wireless channel observation model, the observation value is decomposed into a crystal oscillator low-frequency drift state quantity and a wireless time-varying measurement noise quantity; a measurement noise covariance at a current time is determined; a prior predictive value is obtained by predicting a current clock state according to the state transition model; the prior predictive value is corrected according to the measurement noise covariance to adjust a correction weight, and an optimal synchronization bias estimation value is obtained. The application realizes adaptive adjustment of Kalman gain by decoupling the crystal oscillator low-frequency drift and the wireless high-frequency noise, solves the problems that the existing gating scheme is difficult to distinguish between clock drift and instantaneous noise and is prone to excessive compensation, and improves the determinacy of key traffic transmission in a 5G-TSN fusion network.

Description

Technical Field

[0001] This application relates to the field of communication technology, and in particular to a 5G-TSN traffic scheduling dynamic gating method, apparatus, computer-readable storage medium and electronic device based on a decoupling mechanism. Background Technology

[0002] With the digital transformation of industrial manufacturing, introducing 5G systems as wireless bridges into Time-Sensitive Networking (TSN) has become an important way to achieve flexible deployment of industrial equipment.

[0003] When the 5G air interface is in an open electromagnetic environment, it is susceptible to multipath effects, shadowing fading and co-channel interference, which may trigger HARQ / ARQ retransmissions, causing random jitter in the originally relatively stable transmission delay.

[0004] Gated scheduling relies on a unified time base and the opening and closing times specified by the Gate Control List (GCL). When synchronization deviations or air interface delay jitter accumulate, the message arrival time may mismatch with the predetermined gated time slot, thereby increasing the risk of queuing, delay and packet loss.

[0005] Existing gating schemes typically employ fixed phase and fixed safety redundancy. If the safety redundancy is set too low, it will be difficult to absorb sudden delays and clock drift; if it is set too high, it will compress the effective time window available for transmission and reduce link utilization.

[0006] Some schemes directly adjust the gated phase based on a single synchronous observation, which can easily lead to misinterpreting high-frequency wireless measurement noise as low-frequency clock drift, resulting in frequent fluctuations in the compensation amount and even causing the noise variance estimation to diverge.

[0007] Therefore, a gating scheduling algorithm for dynamic changes in the 5G air interface is needed. Without changing the gating period and gate sequence, it should separate high-frequency measurement noise from crystal oscillator drift, estimate master-slave synchronization deviation, and dynamically correct the gating phase and safety redundancy before the next scheduling period arrives, so as to improve the determinism of critical traffic transmission in 5G-TSN converged networks.

[0008] In industrial scenarios where 5G and TSN are integrated, a key contradiction exists: the conflict between the uncertainty of 5G wireless links and the stringent time determinism of TSN gating scheduling. Existing gating scheduling methods often struggle to simultaneously distinguish between "long-term drift" and "instantaneous wireless measurement noise," and fail to adaptively adjust the gating phase based on the changing characteristics of both. Summary of the Invention

[0009] This application proposes a dynamic gating method and device for 5G-TSN traffic scheduling based on a decoupling mechanism, which solves the problems of existing gating schemes that are difficult to distinguish between clock drift and transient wireless noise, and are prone to overcompensation and filter divergence.

[0010] This application provides a dynamic gating method for 5G-TSN traffic scheduling based on a decoupling mechanism, including the following steps: Acquire the synchronization message transmitted over the 5G air interface and extract the clock deviation observation value at the current moment.

[0011] Based on a pre-configured crystal oscillator drift state transition model, the clock state at the current moment is predicted, and the wireless time-varying measurement noise is estimated based on the residual between the observed clock deviation and the predicted observed value.

[0012] Determine the measurement noise covariance at the current moment.

[0013] The clock state at the current moment is predicted based on the crystal oscillator drift state transition model to obtain a priori predicted value.

[0014] Based on the measurement noise covariance at the current moment, the correction weights of the prior prediction value are adjusted, and the prior prediction value is corrected to obtain the optimal synchronization deviation estimate at the current moment.

[0015] In this embodiment, the inherent low-frequency thermal noise and phase noise of the crystal oscillator are included in the process noise of the crystal oscillator drift state transition model. The statistical characteristics of the process noise are obtained in advance through hardware offline calibration tests and kept constant. Sudden jitter caused by multipath effect, shadow fading, co-channel interference and HARQ retransmission of 5G air interface is included in the measurement noise of the wireless channel observation model.

[0016] In this embodiment of the application, determining the measurement noise covariance at the current moment further includes: using a sliding residual window to statistically analyze the observation residuals at each historical moment, and determining the measurement noise covariance at the current moment from the wireless time-varying measurement noise quantity, wherein the observation residual is the difference between the observed clock deviation value and the predicted value predicted according to the crystal oscillator drift state transition model.

[0017] In this embodiment of the application, the step of using a sliding residual window to statistically analyze the observation residuals at each historical moment and determining the measurement noise covariance at the current moment from the wireless time-varying measurement noise further includes: storing the observation residual at the current moment into a sliding time window of length N; calculating the statistical covariance value of all historical observation residuals within the sliding time window; subtracting the uncertainty component of the prior prediction value from the statistical covariance value, and then taking a preset positive definite lower bound as the constraint lower limit to obtain the measurement noise covariance at the current moment.

[0018] In this embodiment, the length N of the sliding time window is configured jointly based on the coherence time of the 5G wireless channel and the thermal time constant of the crystal oscillator module. Specifically, the length N of the sliding time window is not less than the ratio of the coherence time to the sampling period, and not greater than the ratio of the thermal time constant to the sampling period.

[0019] In this embodiment of the application, the prediction of the clock state at the current moment further includes: calculating the prior prediction value of the synchronization deviation and the prior prediction value of the crystal oscillator drift at the current moment based on the optimal state estimate and the state transition matrix at the previous moment; and calculating the prior error covariance matrix at the current moment based on the error covariance matrix at the previous moment and the pre-acquired process noise covariance matrix, as the uncertainty of the prior prediction value.

[0020] In this embodiment of the application, adjusting the correction weight for the prior prediction value further includes: calculating the filter correction gain for the current time based on the uncertainty of the prior prediction value and the measurement noise covariance at the current time; when the measurement noise covariance at the current time increases, the filter correction gain decreases accordingly, thereby reducing the correction weight for the clock deviation observation at the current time; when the measurement noise covariance at the current time decreases, the filter correction gain increases accordingly, thereby increasing the correction weight for the clock deviation observation at the current time.

[0021] In this embodiment of the application, the following steps are also included: determining the dynamic redundancy duration for absorbing residual errors based on the uncertainty of the optimal synchronization deviation estimate and the prior prediction; and superimposing the optimal synchronization deviation estimate and the dynamic redundancy duration onto the reference time of the gating list to generate a phase-compensated gating list.

[0022] In this embodiment of the application, the phase-compensated gate list maintains the original gate control cycle, the opening duration of each queue, and the gate order of each queue, and only the time phase of the gate opening and closing times is shifted as a whole.

[0023] In this embodiment of the application, the dynamic redundancy duration is calculated based on the uncertainty of the prior prediction value. The greater the uncertainty of the prior prediction value, the longer the dynamic redundancy duration; the smaller the uncertainty of the prior prediction value, the shorter the dynamic redundancy duration.

[0024] In this embodiment of the application, determining the measurement noise covariance at the current moment includes: when the number of samples in the sliding residual window has not reached the preset length N, using a preset benchmark measurement noise covariance as the measurement noise covariance at the current moment.

[0025] In this embodiment, the method employs a two-stage operation mechanism: in the initial stage where the number of samples in the sliding residual window has not reached the preset length N, a fixed-parameter benchmark measurement noise covariance is used for filtering, and the amplification factor of the dynamic redundancy duration is temporarily increased to absorb the crystal oscillator drift error that was not fully decoupled during the initialization stage; in the steady-state stage after the number of samples in the sliding residual window has reached the preset length N, an adaptive determination mechanism for the measurement noise covariance based on sliding residual statistics is enabled, and the dynamic redundancy duration is restored to the standard configuration value.

[0026] This application also provides a 5G-TSN traffic scheduling dynamic gating device based on a decoupling mechanism, used to implement the method described in any of the above embodiments, including: The observation extraction module is used to acquire the synchronization message transmitted over the 5G air interface and extract the clock deviation observation value at the current moment. The sliding residual decoupling module is used to statistically analyze the observation residuals at each historical time using the sliding residual window, and to determine the measurement noise covariance at the current time from the clock deviation observation values. The state prediction module is used to predict the clock state at the current moment based on the crystal oscillator drift state transition model, and obtain the prior prediction value and its uncertainty. An adaptive filtering correction module is used to adjust the correction weights according to the measurement noise covariance at the current moment, and to correct the prior prediction value using the clock deviation observation value at the current moment, and output the optimal synchronization deviation estimate value. The gated phase dynamic reconstruction module is used to determine the dynamic redundancy duration based on the uncertainty of the optimal synchronization deviation estimate and the prior prediction, perform phase compensation on the reference time of the gated list, and send it to the gated execution unit.

[0027] This application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the method described in any embodiment of this application.

[0028] This application also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the method described in any embodiment of this application.

[0029] The above-described technical solutions adopted in the embodiments of this application can achieve the following beneficial effects: First, the present application establishes a dual time-scale estimation model that separates crystal oscillator drift from high-frequency measurement noise, classifying the inherent low-frequency drift of the crystal oscillator into the state transition process, and classifying burst jitter caused by 5G air interface multipath, fading and retransmission into time-varying measurement noise; the residual covariance is calculated online through a sliding residual window, the state prediction uncertainty is deducted and a positive definite lower bound is applied to obtain the dynamic measurement noise covariance. Through the above processing, the problems that the existing gating compensation methods are difficult to distinguish clock drift from instantaneous wireless noise, and are prone to over-compensation and filter divergence are solved, so that the Kalman gain can be adaptively adjusted with the air interface state, and the estimation accuracy is improved.

[0030] Second, under the condition that the gating period, the opening duration of each queue and the gating sequence remain unchanged, the present application introduces a dynamic gating phase compensation amount jointly determined by the synchronization deviation estimation value and the estimation covariance, and updates the opening and closing phases of the GCL before the arrival of the next scheduling cycle. The above gating method solves the problems that a fixed phase is difficult to track cross-domain synchronization deviation, and a fixed large redundancy compresses the effective transmission window, so that the gating time window can be dynamically adjusted with the message arrival time and estimation credibility, reducing the risks of time slot mismatch, queuing backlog and packet loss, and improving link utilization.

[0031] Third, aiming at the problem of insufficient residual samples in the initialization stage of adaptive estimation, the present application sets a two-stage initialization mechanism triggered by sample sufficiency: when k<N, the reference measurement noise parameter is adopted and the safety redundancy is amplified, and when k≥N, the residual decoupling estimation is enabled and the standard safety redundancy is restored; at the same time, the 5G channel coherence time and the crystal oscillator thermal time constant are used to constrain the window length N. The above mechanism solves the contradiction between unstable statistics in the cold start stage and insufficient tracking capability in the steady state stage, and takes into account both initialization robustness and steady state compensation accuracy. Brief Description of the Drawings

[0032] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and the descriptions thereof are used to explain the present application, and do not constitute an improper limitation on the present application. In the drawings: Figure 1 is an overall flowchart of the method embodiment of the present application; Figure 2 is a flowchart of an embodiment of the method of the present application including sliding residual decoupling; Figure 3 is a flowchart of an embodiment of the method of the present application using the number of sliding window samples; Figure 4 is a schematic diagram of GCL dynamic phase compensation; Figure 5 is a schematic diagram of two-stage initialization and safety redundancy adjustment; Figure 6 is an embodiment of the apparatus structure of the present application; Figure 7 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Detailed Implementation

[0033] To make the objectives, technical solutions, and advantages of this application clearer, the technical solutions of this application will be clearly and completely described below in conjunction with specific embodiments and corresponding drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0034] The core concept of this application can be summarized as a dynamic gating scheduling framework integrating decoupling, estimation, and compensation. This includes: dual-timescale state modeling (the basis of decoupling), which divides the system state into low-frequency crystal oscillator drift (included in the state transition process, considered a slowly varying physical quantity) and high-frequency wireless burst noise (included in measurement noise, considered a rapidly varying random quantity), providing a mathematical model basis for subsequent separation of the two types of interference; adaptive noise estimation via a sliding residual window, which, as a decoupling method, calculates the observation residual in real time and statistically analyzes its covariance within the sliding window. Based on the statistical results of the observation residual and the uncertainty of state prediction, the dynamic measurement noise covariance is estimated, and boundary constraints are introduced to ensure the physical rationality of the estimated value and prevent filter divergence; adaptive Kalman filtering correction, which uses the decoupled dynamic measurement noise covariance to adjust the Kalman gain in real time: channel difference → large measurement noise covariance → increase. Small gain → reduces trust in current noise data; good channel → small measurement noise covariance → large gain → quickly tracks the true deviation, thus outputting the optimal synchronization deviation estimate; dynamic GCL phase reconstruction (compensation execution) keeps the gating period and gate sequence unchanged, only compensating for the opening / closing phase: new phase = reference time + synchronization deviation estimate + dynamic safety redundancy, calculated from the estimated covariance, updated before the next scheduling cycle to ensure that the gating window is aligned with the actual arrival time of the message; a two-stage initialization mechanism ensures engineering robustness. In the cold start stage with insufficient samples, fixed parameter filtering and temporary amplification of safety redundancy are used to absorb initial uncertainties. In the steady-state stage with sufficient samples, adaptive filtering is switched to restore standard redundancy, balancing robustness and accuracy.

[0035] The technical solutions provided by the various embodiments of this application are described in detail below with reference to the accompanying drawings.

[0036] Figure 1 This is a general flowchart of an embodiment of the method of this application. Figure 1 As shown, this application provides a dynamic gating method for 5G-TSN traffic scheduling based on a decoupling mechanism, which includes the following steps: Step 110: Obtain the synchronization message of 5G air interface transmission and extract the clock deviation observation value at the current moment.

[0037] The clock deviation observation contains a mixture of low-frequency crystal oscillator drift components and burst noise components from the wireless channel.

[0038] In a 5G-TSN converged network, the 5G system acts as a wireless bridge connecting TSN devices to the TSN network. In this step, the network device receives synchronization messages from the synchronization master station via the 5G air interface and parses the current clock skew observation from the message. This observation reflects the time synchronization deviation between the master and slave devices. Because the 5G air interface operates in an open electromagnetic environment, it is susceptible to multipath effects, shadowing fading, and co-channel interference, and may trigger HARQ / ARQ retransmissions, causing random jitter in the originally relatively stable transmission delay. Therefore, the clock skew observation contains a mixture of low-frequency crystal oscillator drift components and burst noise components from the wireless channel.

[0039] Step 120: Based on the pre-configured crystal oscillator drift state transition model, predict the clock state at the current moment, and estimate the wireless time-varying measurement noise based on the residual between the observed clock deviation and the predicted observed value.

[0040] Based on the pre-configured crystal oscillator drift state transition model and the preset wireless channel observation model, the clock deviation observation value is decomposed into the crystal oscillator low-frequency drift state quantity and the wireless time-varying measurement noise quantity.

[0041] A discrete state space is constructed based on clock synchronization deviation and frequency drift rate, and a system state vector is defined. .in, This represents the absolute synchronization deviation of the system at the current moment. This represents the clock drift rate caused by the crystal oscillator frequency offset. Observation matrix That is, the observed value is only related to the synchronization deviation. Related to crystal oscillator drift rate The state transition process indirectly affects the observed values.

[0042] In this application, the system state is divided into two parts: low-frequency crystal oscillator drift ( The noise from the state transition process is considered a slowly varying physical quantity; high-frequency wireless burst noise (such as multipath and retransmission) is considered measurement noise and treated as a rapidly varying random quantity. Local hardware parameters are obtained through offline calibration testing, and the process noise covariance matrix of the system state transition is set. This matrix represents the inherent low-frequency thermal noise and phase noise of the crystal oscillator, which are input as constant parameters to the system and remain constant during system operation.

[0043] Step 130: Determine the measurement noise covariance at the current moment.

[0044] This step is used to determine the statistical characteristics of the measurement noise at the current moment, i.e., the measurement noise covariance. This covariance reflects the degree to which the current state of the 5G air interface channel affects the observed values. When channel conditions are good, the measurement noise covariance is small, and the reliability of the observed values ​​is high; when channel conditions deteriorate, the measurement noise covariance is large, and the reliability of the observed values ​​is low.

[0045] In one embodiment of this application, step 130 estimates the dynamic measurement noise covariance online using a sliding residual window. In another embodiment of this application, when there are insufficient samples during the initialization phase, step 130 directly uses a preset benchmark measurement noise covariance. The two specific implementations of step 130 will be described in detail in subsequent embodiments.

[0046] Step 140: Predict the clock state at the current moment based on the crystal oscillator drift state transition model to obtain the prior prediction value.

[0047] Utilizing the previous moment ( The optimal state estimate of ) With state transition matrix Combined with process noise Calculate the predicted prior state value at the current moment:

[0048] Simultaneously, update the error covariance matrix by taking the posterior error covariance matrix from the previous time step. With process noise matrix Combined, output the prior error covariance matrix at the current time step:

[0049] The prior error covariance matrix quantifies the theoretical uncertainty of the prior prediction results, providing a basis for subsequent gain calculations.

[0050] Step 150: Based on the measurement noise covariance at the current moment, adjust the correction weights of the prior prediction value, correct the prior prediction value, and obtain the optimal synchronization deviation estimate at the current moment.

[0051] The system receives the measurement noise covariance output in step 130. Calculate the current filter gain matrix:

[0052] When channel degradation leads to When it increases dramatically, the gain This automatically reduces the system's trust in the currently contaminated data. It is combined with the real-time observations obtained in step 110. With observation residuals It completes the correction of the prior state, outputs the posterior optimal state estimate at the current moment, and extracts the optimal master-slave synchronization deviation prediction value for the current period from it.

[0053] Next, the error covariance matrix is ​​updated to converge the uncertainty of the system and prepare for the sliding iteration of the next time step.

[0054] Step 160: Update the gated phase based on the synchronization deviation.

[0055] Based on the uncertainty of the optimal synchronization deviation estimate and the prior prediction, a dynamic redundancy duration for absorbing residual errors is determined; the optimal synchronization deviation estimate and the dynamic redundancy duration are superimposed on the base time of the gating list to generate a phase-compensated gating list, which is then sent out before the next scheduling cycle to control the opening and closing times of the gating units.

[0056] The extracted synchronization deviation estimate Original reference time and dynamic redundancy duration Superposition, to compensate for the phase of the gating list:

[0057] The reconstructed phase parameters are used as control commands and issued before the next scheduling cycle arrives. Gating opening / closing actions are performed according to the new timeline to reduce the risk of time slot mismatch, queue backlog and critical traffic packet loss, and improve the determinism of critical traffic transmission.

[0058] Figure 2 This is a flowchart illustrating an embodiment of the method of this application that includes sliding residual decoupling. In one embodiment, step 130 above, which determines the measurement noise covariance at the current moment, further includes the following steps: Step 131: Calculate the observation residuals at each historical moment using a sliding residual window, and determine the measurement noise covariance at the current moment from the wireless time-varying measurement noise. The observation residual is the difference between the observed clock deviation value and the predicted value obtained according to the crystal oscillator drift state transition model.

[0059] The burst interference caused by multipath effects and HARQ retransmissions in the 5G air interface is modeled as a first-order discrete-time Markov chain. Actual observations are then calculated. Deviation from filter prediction:

[0060] Store the current residual in a file of length [length]. A sliding time window, within which samples change over time. Continuously updated.

[0061] More specifically, step 131 also includes the following sub-steps: Step 131a: Store the observation residual at the current moment into a sliding time window of length N.

[0062] Step 131b: Calculate the statistical covariance of all historical observation residuals within the sliding time window.

[0063] Calculate the statistical covariance of the residual sequence within the current window. .

[0064] Step 131c: Subtract the uncertainty component of the prior prediction from the statistical covariance value, and then take a preset positive definite lower bound as the constraint lower limit to obtain the measurement noise covariance at the current time.

[0065] To prevent filter failure due to wireless interference, the algorithm adds a boundary constraint mechanism to calculate adaptive measurement noise variance:

[0066] Through the above processing, the decoupled estimation of crystal oscillator low-frequency drift and 5G air interface time-varying measurement noise was successfully achieved, and the measurement noise covariance dynamically changing with the 5G channel was obtained. .

[0067] In one embodiment, the length N of the sliding time window is configured jointly based on the coherence time of the 5G wireless channel and the thermal time constant of the crystal oscillator module.

[0068] This application constrains the sliding window length N based on the channel coherence time and the clock drift rate:

[0069] in, For the coherence time of the 5G wireless channel, The sampling period is This is the thermal time constant of the crystal oscillator module. The coherence time characterizes the length of time the wireless channel remains in a constant state. If the window length is less than the ratio of the coherence time to the sampling period, it will be insufficient to capture the complete channel change cycle. The thermal time constant characterizes the time scale of the crystal oscillator frequency drift. If the window length is greater than the ratio of the thermal time constant to the sampling period, the long-term drift trend of the crystal oscillator may be mixed into the noise estimation, affecting the decoupling accuracy.

[0070] In one embodiment, step 140 above, which predicts the clock state at the current moment, further includes the following steps: Step 141: Based on the optimal state estimate and state transition matrix of the previous moment, calculate the prior prediction of the synchronization deviation and the prior prediction of the crystal oscillator drift at the current moment.

[0071] Step 142: Simultaneously, based on the error covariance matrix of the previous moment and the pre-acquired process noise covariance matrix, calculate the prior error covariance matrix of the current moment, which serves as the uncertainty of the prior prediction value.

[0072] The statistical characteristics of the process noise are obtained in advance through hardware offline calibration tests and kept constant; sudden jitter caused by multipath effect, shadow fading, co-channel interference and HARQ retransmission in 5G air interface is included in the measurement noise of the wireless channel observation model.

[0073] This application only addresses the time-varying noise covariance of high-frequency measurements. Dynamic estimation is performed, while the process noise covariance... It belongs to a known physical constant and remains constant during system operation.

[0074] In one embodiment, step 150 above, "adjusting the correction weights for the prior predictions," further includes the following steps: Step 151: Calculate the filter correction gain at the current time based on the uncertainty of the prior prediction value and the measurement noise covariance at the current time.

[0075] Calculate the current filter gain matrix:

[0076] Step 152: When the measurement noise covariance at the current moment increases, the filter correction gain decreases accordingly, thereby reducing the correction weight of the clock deviation observation at the current moment.

[0077] When channel degradation leads to When it increases dramatically, the gain Automatically reduces the system's trust in currently contaminated data.

[0078] Step 153: When the measurement noise covariance at the current moment decreases, the filter correction gain increases accordingly, thereby increasing the correction weight of the clock deviation observation at the current moment.

[0079] When the channel quality is good, the measurement noise covariance is small, the Kalman gain increases, and the system can quickly track the actual synchronization deviation changes.

[0080] By utilizing the decoupled dynamics Real-time adjustment of Kalman gain: Channel difference → Larger gain means lower gain, which reduces confidence in the current noisy data; a better channel means better gain. Small gain → large gain → fast tracking of the true deviation, thus outputting the optimal synchronization deviation estimate.

[0081] In one embodiment, in step 160, after the phase-compensated gating list is generated, the phase-compensated gating list keeps the original gating period, the gating duration of each queue and the gating sequence of each queue unchanged, and only integrally shifts the time phases of the gating start time and the gating end time.

[0082] Figure 3 is a flowchart of an embodiment using the number of sliding window samples in the method of the present application.

[0083] As Figure 3 shown, this embodiment illustrates the switching judgment logic between the initialization phase and the steady-state phase of the system. At each time step k, the system executes: Step 310: Acquire the number of accumulated observation residual samples in the current sliding window, which is denoted as m.

[0084] Step 320: Determine whether the current number of samples m has reached the preset window length N.

[0085] If m<N, it indicates that the system is in the initialization phase (cold start phase), and the number of historical residual samples accumulated in the sliding window is insufficient to perform reliable statistical covariance estimation. At this time, step 330 is executed. Step 330: Perform filtering by using the reference measurement noise covariance with fixed parameters, that is, directly use the preset reference measurement noise covariance as the measurement noise covariance at the current moment , and temporarily increase the amplification factor of the dynamic redundancy duration (for example, amplify times, ) to absorb the crystal oscillator drift error that is not completely decoupled in the initialization phase. This step corresponds to the first phase of the two-phase initialization mechanism of the present application.

[0086] If m ≥ N, it indicates that the system has entered the steady-state phase, and sufficient samples are accumulated in the sliding window to support reliable statistical estimation. At this time, step 340 is executed. Step 340: Enable the adaptive measurement noise covariance determination mechanism based on sliding residual statistics — that is, use the sliding residual window to count the observation residuals at each historical moment, calculate the statistical covariance of the residual sequence in the window, deduct the state prediction uncertainty and impose a positive definite lower bound constraint, so as to obtain the dynamic measurement noise covariance at the current moment , and restore the dynamic redundancy duration to the standard configuration value at the same time.

[0087] Step 350: Based on the measurement noise covariance determined in step 330 or step 340, perform the prediction and correction of Kalman filtering (that is, execute step 140 and step 150) to obtain the optimal synchronization deviation estimation value at the current moment, and then execute step 160 to update the gating phase.

[0088] Step 360: The time step advances to k+1, then returns to step 310 to proceed to the next cycle of judgment and processing.

[0089] Through the above process, the system adaptively selects the filtering strategy at each time step based on the sufficiency of the current sliding window samples, which solves the contradiction between the instability of statistics in the cold start stage and the insufficient tracking ability in the steady state stage, and balances the initialization robustness and steady-state compensation accuracy.

[0090] Figure 4 This is a schematic diagram of GCL dynamic phase compensation. Figure 4 As shown, this application introduces a dynamic gating phase compensation amount determined by the synchronization deviation estimate and the estimated covariance while keeping the gating cycle, the opening duration of each queue and the gate sequence unchanged, and updates the opening and closing phases of the GCL before the next scheduling cycle arrives.

[0091] The above gating method solves the problems of fixed phase difficulty in tracking cross-domain synchronization deviation and fixed large redundancy compression of effective transmission window. It enables the gating time window to be dynamically adjusted according to the message arrival time and estimated reliability, reducing the risk of time slot mismatch, queuing backlog and packet loss.

[0092] Keeping the gating cycle and gate sequence unchanged, compensation is only made for the opening / closing phase: New phase = base time + estimated synchronization deviation + dynamic safety redundancy (calculated from the estimated covariance). Updated before the next scheduling cycle to ensure that the gating window is aligned with the actual arrival time of the message.

[0093] In one embodiment, the dynamic redundancy duration is calculated based on the uncertainty of the prior prediction value. The greater the uncertainty of the prior prediction value, the longer the dynamic redundancy duration; the smaller the uncertainty of the prior prediction value, the shorter the dynamic redundancy duration.

[0094] Specifically, the prior error covariance matrix output in step 140 is extracted. Calculate the dynamic safety redundancy used to absorb residual errors. . The diagonal elements in the graph reflect the variance of the current prediction. The larger the variance, the less reliable the prediction is, and therefore a longer safety redundancy period is needed to cover the possible range of bias.

[0095] In one embodiment, step 130 (determining the measurement noise covariance at the current moment) includes: Step 132: When the number of samples in the sliding residual window has not reached the preset length N, the preset benchmark measurement noise covariance is used as the measurement noise covariance at the current moment.

[0096] This embodiment is applicable to the system cold start phase or scenarios with insufficient sliding window samples. At this stage, since the number of historical residual samples accumulated in the sliding window is insufficient for reliable statistical estimation, directly using the preset reference measurement noise covariance as the measurement noise covariance at the current moment can avoid the problem of inaccurate estimation caused by insufficient samples.

[0097] Figure 5 is a schematic diagram of two-stage initialization and safety redundancy adjustment. In one embodiment, the method adopts a two-stage operation mechanism: First stage (k<N): In the initial stage where the number of samples in the sliding residual window has not reached the preset length N, filtering is performed using the reference measurement noise covariance with fixed parameters, and the amplification factor of the dynamic redundancy duration is temporarily increased to absorb the crystal oscillator drift error that is not completely decoupled in the initialization phase.

[0098] To ensure the robustness of the system during the initialization phase (i.e., the time step ), the present application designs a two-stage initialization mechanism. In the first stage ( ), conventional Kalman filtering is performed, and the gating safety redundancy is temporarily amplified by times ( ) to absorb the undecoupled monotonic drift error of the crystal oscillator.

[0099] Second stage (k≥N): In the steady-state phase after the number of samples in the sliding residual window reaches the preset length N, the adaptive determination mechanism for measurement noise covariance based on sliding residual statistics is enabled, and the dynamic redundancy duration is restored to the standard configuration value.

[0100] In the second stage ( ), when the sliding window samples are sufficient, the system enters the residual-decoupled adaptive Kalman filtering mechanism, and restores the safety redundancy to the standard set value.

[0101] For example, Figure 3 is a flowchart of an embodiment of the method of the present application using the number of sliding window samples. As shown in Figure 3 , the system checks at each time step whether the number of current window samples reaches the preset length N. If not, the first-stage logic is executed (cold start phase: filtering with fixed parameters, and temporarily amplifying the safety redundancy to absorb initial uncertainties); if it has reached, the system switches to the second-stage logic (steady-state phase: switching to adaptive filtering, restoring the standard redundancy, taking both robustness and accuracy into account).

[0102] Through the above two-stage initialization mechanism, the contradiction between unstable statistics in the cold start phase and insufficient tracking capability in the steady-state phase is resolved, and both initialization robustness and steady-state compensation accuracy are achieved.

[0103] Figure 6This is an embodiment of the device structure described in this application. This application also provides a 5G-TSN traffic scheduling dynamic gating device based on a decoupling mechanism, used to implement the method described in any of the above embodiments. The device includes: The observation extraction module 51 is used to acquire the synchronization message transmitted over the 5G air interface and extract the clock deviation observation value at the current moment. This module performs the function of step 110.

[0104] The sliding residual decoupling module 52 is used to statistically analyze the observation residuals at each historical time using a sliding residual window, and to determine the measurement noise covariance at the current time from the clock deviation observations. This module performs the functions of steps 130 and 131-131c.

[0105] The state prediction module 53 is used to predict the clock state at the current moment based on the crystal oscillator drift state transition model, and obtain the prior prediction value and its uncertainty. This module performs the functions of steps 140 and 141-142.

[0106] The adaptive filtering correction module 54 is used to adjust the correction weights according to the measurement noise covariance at the current moment, and to correct the prior prediction value using the clock deviation observation value at the current moment, outputting the optimal synchronization deviation estimate value. This module performs the functions of steps 150 and 151-153.

[0107] The gated phase dynamic reconstruction module 55 is used to determine the dynamic redundancy duration based on the uncertainty of the optimal synchronization deviation estimate and the prior prediction, perform phase compensation on the reference time of the gated list, and send it to the gated execution unit. This module performs the function of step 160.

[0108] It should be noted that the above modules can be software modules, hardware modules, or a combination of software and hardware. The modules can communicate with each other via a bus or other means.

[0109] It should be noted that the execution subject of each step of the method provided in the above method embodiments can be the same device, or the method can be executed by different devices. For example, the execution subject of steps 110 and 120 can be device 1, and the execution subject of step 130 can be device 2; or the execution subject of step 110 can be device 1, and the execution subject of steps 120 and 130 can be device 2; and so on.

[0110] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0111] Therefore, this application also proposes a computer-readable storage medium having a computer program stored thereon that, when executed by a processor, implements the methods described in any embodiment of this application.

[0112] Furthermore, this application also proposes an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the method described in any embodiment of this application.

[0113] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0114] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0115] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0116] In a typical configuration, a computing device includes one or more processors (CPUs), input / output interfaces, network interfaces, and memory. Memory may include non-persistent storage in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.

[0117] Figure 7 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. The electronic device 600 shown is merely an example and should not impose any limitations on the function and scope of use of the embodiments of this application. It includes: one or more processors 620; and a storage device 610 for storing one or more programs, which, when run by the one or more processors 620, enable the one or more processors 620 to implement the methods provided in the embodiments of this application.

[0118] The electronic device 600 also includes an input device 630 and an output device 640; the processor 620, storage device 610, input device 630 and output device 640 in the electronic device can be connected by a bus or other means, as shown in the figure, which is connected by a bus 650.

[0119] Storage device 610, as a computer-readable storage medium, can be used to store software programs, computer-executable programs, and module units, such as program instructions corresponding to the methods in the embodiments of this application. Storage device 610 may primarily include a program storage area and a data storage area. The program storage area may store the operating system and at least one application program required for a function; the data storage area may store data created based on terminal usage. Furthermore, storage device 610 may include high-speed random access memory and non-volatile memory, such as at least one disk storage device, flash memory device, or other non-volatile solid-state storage device. In some instances, storage device 610 may further include memory remotely located relative to processor 620, and these remote memories can be connected via a network. Examples of such networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof.

[0120] Input device 630 can be used to receive input digital, character, or voice information, and to generate key signal inputs related to user settings and function control of the electronic device. Output device 640 may include electronic devices such as a display screen and a speaker.

[0121] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0122] Those skilled in the art will understand that, unless otherwise stated, the singular forms “a,” “an,” “the,” and “the” used herein may also include the plural forms. It should be understood that when a device or component is “connected” to another device or component, it may be directly connected to the other device or component, or there may be an intermediary device or component. Furthermore, the term “connection” as used herein may include partially wireless connections as well as partially wired connections. The character “ / ” generally indicates that the preceding and following objects are in an “or” relationship.

[0123] The above description is merely an embodiment of this application and is not intended to limit the scope of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of the claims of this application.

Claims

1. A dynamic gating method for 5G-TSN traffic scheduling based on a decoupling mechanism, characterized in that, Includes the following steps: Acquire the synchronization message transmitted over the 5G air interface and extract the clock deviation observation value at the current moment; Based on the pre-configured crystal oscillator drift state transition model, the clock state at the current moment is predicted, and the wireless time-varying measurement noise is estimated based on the residual between the observed clock deviation and the predicted observed value. Determine the measurement noise covariance at the current moment; The clock state at the current moment is predicted based on the crystal oscillator drift state transition model to obtain the prior prediction value; Based on the measurement noise covariance at the current moment, the correction weights of the prior prediction value are adjusted, and the prior prediction value is corrected to obtain the optimal synchronization deviation estimate at the current moment.

2. The method as described in claim 1, characterized in that, The inherent low-frequency thermal noise and phase noise of the crystal oscillator are included in the process noise of the crystal oscillator drift state transition model. The statistical characteristics of the process noise are obtained in advance through hardware offline calibration tests and kept constant. Sudden jitter caused by multipath effect, shadow fading, co-channel interference and HARQ retransmission in 5G air interface is included in the measurement noise of the wireless channel observation model.

3. The method as described in claim 1, characterized in that, Determining the measurement noise covariance at the current moment further includes: The observation residuals at each historical moment are statistically analyzed using a sliding residual window. The measurement noise covariance at the current moment is determined from the wireless time-varying measurement noise. The observation residual is the difference between the observed clock deviation value and the predicted value obtained based on the crystal oscillator drift state transition model.

4. The method as described in claim 3, characterized in that, The step of using a sliding residual window to statistically analyze the observation residuals at each historical time point, and determining the measurement noise covariance at the current time from the wireless time-varying measurement noise, further includes: Store the observation residual at the current moment into a sliding time window of length N; Calculate the statistical covariance of all historical observation residuals within the sliding time window; The measurement noise covariance at the current moment is obtained by subtracting the uncertainty component of the prior prediction from the statistical covariance value and then taking a preset positive definite lower bound as the constraint lower limit.

5. The method as described in claim 4, characterized in that, The length N of the sliding time window is not less than the ratio of the coherence time to the sampling period, and not greater than the ratio of the thermal time constant to the sampling period.

6. The method as described in claim 1, characterized in that, The prediction of the current clock state further includes: Based on the optimal state estimate and state transition matrix of the previous moment, calculate the prior prediction of the synchronization deviation and the prior prediction of the crystal oscillator drift at the current moment. Based on the error covariance matrix of the previous moment and the pre-acquired process noise covariance matrix, the prior error covariance matrix of the current moment is calculated, which is used as the uncertainty of the prior prediction value.

7. The method as described in claim 1, characterized in that, The adjustment of the correction weights for the prior predictions further includes: The filter correction gain at the current moment is calculated based on the uncertainty of the prior prediction and the measurement noise covariance at the current moment. When the measurement noise covariance at the current moment increases, the filter correction gain decreases accordingly, thereby reducing the correction weight of the clock deviation observation at the current moment; When the measurement noise covariance at the current moment decreases, the filter correction gain increases accordingly, thereby increasing the correction weight for the clock deviation observation at the current moment.

8. The method as described in claim 1, characterized in that, It also includes the following steps: Based on the uncertainty of the optimal synchronization deviation estimate and the prior prediction, the dynamic redundancy duration for absorbing residual errors is determined. The optimal synchronization deviation estimate and the dynamic redundancy duration are superimposed on the base time of the gating list to generate a phase-compensated gating list.

9. The method as described in claim 8, characterized in that, The phase-compensated gating list retains the original gating cycle, the opening duration of each queue, and the gate order of each queue, with only the time phase of the gating opening and closing times being shifted overall.

10. The method as described in claim 8, characterized in that, The dynamic redundancy duration is calculated based on the uncertainty of the prior prediction value. The greater the uncertainty of the prior prediction value, the longer the dynamic redundancy duration; the smaller the uncertainty of the prior prediction value, the shorter the dynamic redundancy duration.

11. The method as described in claim 1, characterized in that, Determining the measurement noise covariance at the current moment includes: when the number of samples in the sliding residual window has not reached the preset length N, using a preset benchmark measurement noise covariance as the measurement noise covariance at the current moment.

12. The method as described in claim 1, characterized in that, In the initial stage when the number of samples in the sliding residual window has not reached the preset length N, the reference measurement noise covariance with fixed parameters is used for filtering, and the amplification factor of the dynamic redundancy duration is temporarily increased to absorb the crystal oscillator drift error that is not completely decoupled in the initialization stage. In the steady state phase after the number of samples in the sliding residual window reaches the preset length N, an adaptive determination mechanism for measurement noise covariance based on sliding residual statistics is activated, and the dynamic redundancy duration is restored to the standard configuration value.

13. A 5G-TSN traffic scheduling dynamic gating device based on a decoupling mechanism, used to implement the method described in any one of claims 1 to 12, characterized in that, include: The observation extraction module is used to acquire the synchronization message transmitted over the 5G air interface and extract the clock deviation observation value at the current moment. The sliding residual decoupling module is used to statistically analyze the observation residuals at each historical time using the sliding residual window, and to determine the measurement noise covariance at the current time from the clock deviation observation values. The state prediction module is used to predict the clock state at the current moment based on the crystal oscillator drift state transition model, and obtain the prior prediction value and its uncertainty. An adaptive filtering correction module is used to adjust the correction weights according to the measurement noise covariance at the current moment, and to correct the prior prediction value using the clock deviation observation value at the current moment, and output the optimal synchronization deviation estimate value. The gated phase dynamic reconstruction module is used to determine the dynamic redundancy duration based on the uncertainty of the optimal synchronization deviation estimate and the prior prediction, perform phase compensation on the reference time of the gated list, and send it to the gated execution unit.

14. 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 method as described in any one of claims 1 to 12.

15. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the method as described in any one of claims 1 to 12.

Images