A method for dynamic tracking of time-varying frequency offset in VDES satellite link

Abstract

The application relates to a time-varying frequency deviation dynamic tracking method in a VDES satellite link, comprising the following steps: acquiring a VDES satellite link baseband digital signal and performing coarse frequency deviation estimation to obtain a coarse frequency deviation estimation value; setting a processing window sliding along a sampling time, calculating a frequency deviation change rate according to the coarse frequency deviation estimation value in the window, and adaptively adjusting the window length according to the frequency deviation change rate; constructing a Kalman filter with a frequency deviation change rate and a frequency deviation optimal estimation value as a state vector, and establishing a frequency deviation estimation state transition matrix based on a uniform acceleration motion model; iteratively calculating by using the Kalman filter, outputting a frequency deviation posterior estimation value as a frequency deviation optimal estimation value at a corresponding time to realize time-varying frequency deviation dynamic tracking, and introducing fine frequency deviation estimation to optimize the frequency deviation prior estimation value at the current time during iteration. The application can realize fast and high-precision tracking of the time-varying frequency deviation of the VDE-SAT link, and improve the communication stability of the VDES satellite link.

Description

Technical Field

[0001] This invention relates to the field of ship communication technology, and in particular to a time-varying frequency offset dynamic tracking method in VDES satellite links. Background Technology

[0002] The VDES (Very High Frequency Data Exchange System) satellite link (VDE-SAT) serves as the core link for maritime communication, undertaking critical services such as navigation, collision avoidance, and information exchange between ships and satellites. During VDE-SAT communication, the high-speed relative motion between the ship and the satellite generates significant Doppler frequency offset, which exhibits distinct time-varying characteristics, meaning the offset value dynamically changes over time, ranging from -4kHz to +4kHz. Furthermore, channel fluctuations caused by complex weather conditions at sea further exacerbate this frequency offset instability, severely impacting signal synchronization accuracy, leading to increased bit error rate, and reduced communication quality.

[0003] Existing frequency offset estimation methods for VDES systems have significant drawbacks when dealing with time-varying frequency offsets in VDE-SAT links: First, most methods use a fixed-length sliding window for signal processing. When the satellite link frequency offset changes drastically, the fixed window cannot quickly capture the trend of frequency offset changes, leading to tracking lag. When the frequency offset is stable, the fixed window cannot fully utilize signal redundancy information, resulting in insufficient estimation accuracy. Second, in existing methods, coarse and fine frequency offset estimations are independent, lacking an effective data linkage mechanism. Fine frequency offset estimation requires searching for the optimal value over a large range, resulting in low estimation efficiency and difficulty in meeting the rapid tracking requirements of sudden time-varying frequency offsets in VDE-SAT links. Third, the phase tracking modules introduced in some methods have poor adaptability to time-varying frequency offsets in VDE-SAT links, especially in scenarios with sudden frequency offset changes, easily leading to phase ambiguity and tracking failure. Summary of the Invention

[0004] The technical problem to be solved by the present invention is to provide a dynamic tracking method for time-varying frequency offset in VDES satellite links, which can realize fast and high-precision tracking of time-varying frequency offset in VDE-SAT links and improve the communication stability of VDES satellite links.

[0005] The technical solution adopted by this invention to solve its technical problem is: to provide a time-varying frequency offset dynamic tracking method in a VDES satellite link, comprising: Receive VDES satellite link signals and preprocess them to obtain baseband digital signals; Coarse frequency offset estimation is performed on the baseband digital signal to obtain the coarse frequency offset estimate; Set a processing window that slides along the sampling time, calculate the frequency offset change rate based on the coarse frequency offset estimate within the window, and adaptively adjust the window length based on the frequency offset change rate; Construct a Kalman filter with the frequency offset change rate and the optimal frequency offset estimate as the state vector, and establish a frequency offset estimation state transition matrix based on a uniform acceleration motion model; The Kalman filter is used for iterative calculation, and the output frequency offset posterior estimate is used as the optimal frequency offset estimate at the corresponding time to achieve time-varying frequency offset dynamic tracking. During the iteration, a fine frequency offset estimate is introduced to optimize the frequency offset prior estimate of the current round. The frequency offset posterior estimate is calculated based on the optimized frequency offset prior estimate, and then the posterior error covariance matrix at the current time is updated for the next iteration calculation.

[0006] Furthermore, during iteration, a fine frequency offset estimation is introduced to optimize the prior estimate of the frequency offset in this round, including: The frequency offset prior estimate predicted by the Kalman filter in this round is used as the initial value of the fine frequency offset estimate. Fine frequency offset estimation is performed within the set search step size, and the fine frequency offset estimate is used as the optimized frequency offset prior estimate.

[0007] Furthermore, the search step size is set to be adaptively adjusted based on the frequency offset rate of change.

[0008] Furthermore, the observation equation is expressed as in, for The observed value at time, For the observation matrix, for The observation noise vector at time step.

[0009] The time-varying frequency offset dynamic tracking method according to claim 4 is characterized in that the observation noise vector follows a function with a mean of 0 and a variance of . The distribution is Gaussian, and the variance is... The value is updated in real time based on the variance of the coarse frequency offset estimate within the processing window.

[0010] Furthermore, the state equation of the Kalman filter is expressed as follows: in, and They are respectively Time and The state vector at time t, Here is the state transition matrix. for The process noise vector at time step.

[0011] The time-varying frequency offset dynamic tracking method according to claim 6 is characterized in that the process noise vector is represented as: in, This is the process noise vector. and These represent the process noise variances of two different components, and are adaptively adjusted according to the signal-to-noise ratio.

[0012] Furthermore, the initial value of the optimal frequency offset estimate is set to the initial coarse frequency offset estimate, and the initial value of the frequency offset change rate is set to 0.

[0013] Furthermore, the window length is adaptively adjusted based on the frequency offset rate of change, including: When the frequency deviation rate of change is less than or equal to the first rate of change threshold, the window length is adjusted to the maximum. When the frequency deviation rate of change is greater than the first rate of change threshold and less than or equal to the second rate of change threshold, the window length is adjusted to the middle. When the frequency deviation rate of change is greater than the second rate of change threshold, the window length is adjusted to the minimum.

[0014] Furthermore, the first rate of change threshold is set to 500 Hz / s and the second rate of change threshold is set to 2000 Hz / s.

[0015] Furthermore, the maximum window length is set to 256 symbols, the intermediate window length to 128 symbols, and the minimum window length to 64 symbols.

[0016] Furthermore, the first derivative estimation method is used to calculate the frequency deviation rate.

[0017] Beneficial effects By adopting the above-mentioned technical solution, the present invention has the following advantages and positive effects compared with the prior art: This invention addresses the limitation of existing fixed windows, which cannot simultaneously balance satellite link tracking speed and estimation accuracy, by adaptively adjusting the sliding window length based on the frequency offset change rate. A large window is used to improve accuracy when the frequency offset is stable, while a small window is used to improve response speed when the frequency offset changes drastically, thus accurately adapting to the dynamic characteristics of time-varying frequency offset in VDE-SAT links. This invention introduces a Kalman filter module between coarse frequency offset estimation and fine frequency offset estimation. It uses the optimal estimate from the previous moment and the frequency offset change rate for joint prediction, providing a precise initial search range for fine frequency offset estimation. Compared with traditional large-range search, it can significantly improve estimation efficiency and effectively suppress the interference of satellite link channel noise on the estimation results. This invention uses a Kalman filter state update mechanism to feed back the fine frequency offset estimation results to the prediction module, forming a closed-loop tracking link of "prediction-estimation-update", which can effectively improve the stability and accuracy of time-varying frequency offset tracking and is suitable for dynamic frequency offset change scenarios of VDE-SAT link. The method of this invention has moderate computational complexity, and each module can be hardware integrated through FPGA. It uses fixed-point arithmetic instead of floating-point arithmetic, which can effectively reduce storage resource consumption and meet the lightweight deployment requirements of VDES satellite link terminals. Attached Figure Description

[0018] Figure 1 This is a flowchart of an embodiment of the present invention; Figure 2 This is a state transition flowchart of an embodiment of the present invention. Detailed Implementation

[0019] The present invention will be further illustrated below with reference to specific embodiments. It should be understood that these embodiments are for illustrative purposes only and are not intended to limit the scope of the invention. Furthermore, it should be understood that after reading the teachings of this invention, those skilled in the art can make various alterations or modifications to the invention, and these equivalent forms also fall within the scope defined by the appended claims.

[0020] The embodiments of the present invention relate to a dynamic tracking method for time-varying frequency offset in VDES satellite links, which aims to overcome the shortcomings of slow response, low accuracy and poor adaptability of time-varying frequency offset tracking in existing VDES satellite links (VDE-SAT), and achieve fast and high-precision tracking of time-varying frequency offset in VDE-SAT links, thereby improving the communication stability of VDES satellite links.

[0021] like Figure 1 As shown, the specific steps include: Step 1: Receive VDES satellite link signal, preprocess the signal to obtain baseband digital signal; the preprocessing includes filtering, down-conversion, sampling and analog-to-digital conversion; Step 2: Perform coarse frequency offset estimation on the baseband digital signal to obtain the coarse frequency offset estimate at the current time. Where k is the index at the current time; Step 3: Based on the coarse frequency offset estimate, calculate the frequency offset change rate using the sliding window first derivative estimation method. According to the frequency deviation rate Adaptively adjust the length of the sliding window; Step 4: Construct the state equation and observation equation of the Kalman filter module based on the time-varying frequency offset characteristics of the VDE-SAT link, and calculate the frequency offset change rate at the previous moment. The optimal estimate of frequency offset at the previous time step The input is given to the Kalman filter module, which uses Kalman filtering to predict the frequency offset at the current time. ; Step 5: Using frequency offset prediction values Using the initial value, fine frequency offset estimation is performed within a preset small search interval to obtain the fine frequency offset estimate at the current time. ; Step 6: Calculate the fine frequency offset estimate. The feedback is sent to the Kalman filter module to update the state equation of the Kalman filter, thus obtaining the optimal estimate of the frequency offset at the current time. Complete one time-varying frequency offset tracking; Step 7: Repeat steps 2-6 to achieve continuous dynamic tracking of the time-varying frequency offset of the VDES satellite link.

[0022] In step 3, the rate of change of frequency offset The calculation method is as follows: Where T is the time interval between two consecutive coarse frequency offset estimates, i.e. the sampling period, in seconds; This is the coarse frequency offset estimate before time N, in Hz; The unit is Hz / s, which is used to characterize the degree of drastic change in the frequency offset of the VDE-SAT link over time.

[0023] The sliding window length is adaptively adjusted based on the frequency offset change rate v(k), specifically including: Step 301: Preset two frequency offset change rate thresholds v1 and v2, where v1 < v2; adapt to VDE-SAT link scenario, preferably v1=500Hz / s, v2=2000Hz / s; Step 302: When v(k) ≤ v1, the frequency offset is determined to be stable, and the sliding window length is adjusted to the maximum window length Lmax to make full use of the redundant information of the VDE-SAT signal and improve the estimation accuracy; preferably, Lmax = 256 symbols to ensure the estimation accuracy under low frequency offset changes. Step 303: When v1 < v(k) ≤ v2, it is determined that the frequency offset is changing slowly, and the sliding window length is adjusted to the middle window length Lmid, where Lmax > Lmid > Lmin, to balance tracking speed and estimation accuracy; preferably, Lmid = 128 symbols to balance accuracy and response speed. Step 304: When v(k) > v2, it is determined that the frequency offset has changed drastically. The sliding window length is adjusted to the minimum window length Lmin to quickly capture the frequency offset change trend of the VDE-SAT link and improve the tracking response speed. Preferably, Lmin = 64 symbols to adapt to high frequency offset change scenarios.

[0024] In step 4, the state equations and observation equations of the Kalman filter module are constructed as follows: Step 401: Define the state vector of the Kalman filter: Where X(k) is the state vector at time k. Let v(k) be the optimal frequency offset of the VDE-SAT link at time k, and v(k) be the frequency offset change rate at time k.

[0025] Step 402 defines the state equation of the Kalman filter: Where A is the state transition matrix, constructed based on the uniform acceleration motion model of the VDE-SAT link frequency offset, and its expression is: in, Let W(k-1) be the sampling period, and W(k-1) be the process noise vector at time k-1, which follows a Gaussian distribution with mean 0 and variance Q. Let Q be the process noise variance matrix, adapted to the VDE-SAT link scenario. Preferably, the process noise variance matrix Q is set as follows: Among them, qf and qv are adaptively adjusted according to the signal-to-noise ratio. Step 403 defines the observation equation for the Kalman filter: Where Z(k) is the observed value at time k, i.e., the fine frequency offset estimate at the current time. H is the observation matrix, expressed as: V(k) is the observation noise vector at time k, which follows a Gaussian distribution with mean 0 and variance R, adapted to the characteristics of the VDE-SAT link channel. Preferably, The estimation is updated in real time based on the sliding window variance, specifically determined by the coarse frequency offset estimate within the processing window at the current moment. The coarse frequency offset estimate is the raw result of the baseband signal estimation, without processing such as small-range search and filtering correction, thus fully preserving the original noise characteristics such as channel noise and interference, and reflecting the true noise characteristics of the link.

[0026] Step 402 uses a Kalman filter to estimate the prior state, resulting in the prior state estimation vector at the current time. It can be represented as: in , is the posterior state vector of the previous time step. This is the posterior estimate of the rate of change of frequency offset at the previous time step. Therefore, the prior estimate of the frequency offset is calculated as follows: = +T That is, the frequency offset prediction value at the current moment is: = +T Next, using the frequency offset prediction value Starting from a predetermined point, fine frequency offset estimation is performed within a set interval. The preset small-range search interval is: Wherein, Δf is a preset search step size, adapted to the frequency offset range of the VDE-SAT link, with a value range of 10Hz to 50Hz. Preferably, Δf is as follows: Compared to the large-scale search of traditional methods (such as ±1kHz), this search range can significantly improve the efficiency of fine frequency offset estimation and is suitable for the fast tracking requirements of VDE-SAT links.

[0027] In step 6, the fine frequency offset estimate is... Feedback is sent to the Kalman filter module, which then updates the Kalman filter state, specifically including: Step 601: At this point, the prior error covariance matrix... It can be represented as: in Let $\mathbf{a}$ be the posterior error covariance matrix of the previous time step, with the initial value set to $\mathbf{a}$. Preferred , Let A be the transpose of the state transition matrix A.

[0028] Step 602: The Kalman gain can be expressed as: in, The transpose of the observation matrix H, and the frequency offset dimension gain. , for The frequency offset dimension of the diagonal elements.

[0029] Step 603: Update the posterior state vector estimate as follows: After expansion, the optimal estimate of the frequency offset at the current time is obtained: Step 604: Update the posterior error covariance at the current time step as follows: Where I is a 2-order identity matrix The updated It will be used as the input for the next moment.

[0030] The state transition flowchart of the entire time-varying frequency deviation dynamic tracking process is as follows: Figure 2 As shown, before performing iterative calculations, the initial value of the optimal frequency offset estimate can be set to the initial coarse frequency offset estimate, and the initial value of the frequency offset change rate can be set to 0. The coarse frequency offset estimate is the only data available during the system cold start, which allows the filter to converge quickly. At the same time, during the cold start, the ship is stationary, and the relative motion between the ship and the satellite is stable, so the frequency offset change rate should be 0.

[0031] Example 1: A method for dynamic tracking of time-varying frequency offset in VDES satellite links This embodiment targets the VDES satellite link downlink scenario, in which the relative speed between the ship and the satellite can reach 3 km / s, the Doppler frequency offset range is -4 kHz to +4 kHz, and the frequency offset changes dynamically with time. The specific implementation steps are as follows: S1: Signal preprocessing: Receive the QPSK modulated signal from the VDE-SAT downlink, filter out noise outside the frequency band through a bandpass filter (filtered frequency band is 156.025MHz~162.025MHz), reduce the signal frequency to intermediate frequency (70MHz) through downconversion, and then obtain the baseband digital signal through ADC sampling (sampling rate 50KHz) and analog-to-digital conversion; S2: Coarse Frequency Offset Estimation: The DFT phase difference algorithm is used to perform coarse frequency offset estimation on the baseband digital signal to obtain the coarse frequency offset estimate at the current time. The estimated values ​​for the first 128 time points in the cache are used, and the time interval between two adjacent coarse estimates is T=1ms (i.e. 0.001s). S3: Dynamic Window Adjustment: The frequency offset change rate v(k) is calculated according to the formula. Based on the typical relative motion speed of the VDES satellite link (maximum ship speed 30 knots), the preset threshold v1 = 500Hz / s is derived. Considering the maximum possible frequency offset change rate under extreme weather conditions, the preset v2 = 2000Hz / s is determined. The maximum window length Lmax = 256 symbols, the intermediate window length Lmid = 128 symbols, and the minimum window length Lmin = 64 symbols. If v(k) = 300Hz / s ≤ v1, the sliding window length is adjusted to 256 symbols; if v(k) = 1500Hz / s (v1 < v(k) ≤ v2), it is adjusted to 128 symbols; if v(k) = 3000Hz / s > v2, it is adjusted to 64 symbols. S4: Kalman Filter Prediction: Construct the state equation and observation equation for the Kalman filter, where the state transition matrix A = [[1, 0.001], [0, 1]], the observation matrix H = [[1, 0]], the process noise variance Q = diag([10000, 100]), and the observation noise variance R = 2500; The optimal estimate of frequency offset at the previous time step The input is to the Kalman filter module, and the prediction steps of the Kalman filter are as follows: calculate the prior state estimate and the prior error covariance to obtain the predicted frequency offset value at the current time. ; S5: Fine Frequency Offset Estimation: A preset search step size Δf is used. If v(k) = 300Hz / s ≤ v1, Δf is set to 30Hz; if v(k) = 1500Hz / s (v1 < v(k) ≤ v2), Δf is set to 100Hz; if v(k) = 3000Hz / s > v2, Δf is set to 200Hz. The Quinn algorithm is used to perform fine frequency offset estimation within this interval to obtain the fine frequency offset estimate. ; S6: Optimal estimate update: [The remaining text appears to be incomplete and requires further context.] The update steps for the input Kalman filter module are as follows: calculate the Kalman gain, posterior state estimate, and posterior error covariance; update the posterior estimate of the state equation; and obtain the optimal frequency offset estimate at the current time. Complete this follow-up; S7: Repeat steps S2-S6 to achieve continuous dynamic tracking of the time-varying frequency offset of the VDE-SAT link.

[0032] Example 2: A time-varying frequency offset dynamic tracking device in a VDES satellite link The device in this embodiment corresponds to the method in Embodiment 1. It is hardware integrated through an FPGA (Field Programmable Gate Array) to meet the lightweight requirements of VDE-SAT link terminals. The specific hardware resource allocation and performance parameters are as follows: Signal preprocessing module: It occupies 12% of the FPGA's logic resources, and uses a pipelined structure to implement filtering and down-conversion operations, with a processing delay of ≤50ns; Coarse frequency offset estimation module: It occupies 18% of the FPGA's logic resources, is implemented based on a parallel DFT architecture, has a computing speed of ≥100Msps, and is adapted to the high-speed processing requirements of VDE-SAT link QPSK signals. Dynamic window adjustment module: It occupies 5% of the FPGA's logic resources and uses comparators and multiplexers to realize threshold judgment and window length switching, with a response delay of ≤10ns; Kalman filter prediction module: It occupies 25% of the FPGA's logic resources, uses fixed-point arithmetic instead of floating-point arithmetic, reduces storage resource consumption, and has an arithmetic accuracy error of ≤0.1Hz; Fine frequency offset estimation module: It occupies 20% of the FPGA's logic resources and is implemented using a parallel interpolation architecture, which improves the search efficiency by 60% compared to traditional methods; Optimal estimate update module: occupies 10% of the FPGA's logic resources to implement data feedback and status updates, with a delay of ≤20ns.

Claims

1. A time-varying frequency offset dynamic tracking method in a VDES satellite link, characterized in that, include: Receive VDES satellite link signals and preprocess them to obtain baseband digital signals; Coarse frequency offset estimation is performed on the baseband digital signal to obtain the coarse frequency offset estimate; Set a processing window that slides along the sampling time, calculate the frequency offset change rate based on the coarse frequency offset estimate within the window, and adaptively adjust the window length based on the frequency offset change rate; Construct a Kalman filter with the frequency offset change rate and the optimal frequency offset estimate as the state vector, and establish a frequency offset estimation state transition matrix based on a uniform acceleration motion model; The Kalman filter is used for iterative calculation, and the output frequency offset posterior estimate is used as the optimal frequency offset estimate at the corresponding time to achieve time-varying frequency offset dynamic tracking. During the iteration, a fine frequency offset estimate is introduced to optimize the frequency offset prior estimate of the current round. The frequency offset posterior estimate is calculated based on the optimized frequency offset prior estimate, and then the posterior error covariance matrix at the current time is updated for the next iteration calculation.

2. The time-varying frequency deviation dynamic tracking method according to claim 1, characterized in that, During iteration, a fine frequency offset estimation is introduced to optimize the prior estimate of the frequency offset in this round, including: The frequency offset prior estimate predicted by the Kalman filter in this round is used as the initial value of the fine frequency offset estimate. Fine frequency offset estimation is performed within the set search step size, and the fine frequency offset estimate is used as the optimized frequency offset prior estimate.

3. The time-varying frequency deviation dynamic tracking method according to claim 2, characterized in that, The search step size is set to be adaptively adjusted based on the frequency offset rate of change.

4. The time-varying frequency deviation dynamic tracking method according to claim 2, characterized in that, the observation... The equation is expressed as in, for The observed value at time, For the observation matrix, for The observation noise vector at time step.

5. The time-varying frequency deviation dynamic tracking method according to claim 4, characterized in that, The observed noise vector follows a pattern with a mean of 0 and a variance of 0. The distribution is Gaussian, and the variance is... The value is updated in real time based on the variance of the coarse frequency offset estimate within the processing window.

6. The time-varying frequency deviation dynamic tracking method according to claim 1, characterized in that, The state equation of the Kalman filter is expressed as follows: in, and They are respectively Time and The state vector at time t, Here is the state transition matrix. for The process noise vector at time step.

7. The time-varying frequency deviation dynamic tracking method according to claim 6, characterized in that, The process noise vector is represented as in, This is the process noise vector. and These represent the process noise variances of two different components, and are adaptively adjusted according to the signal-to-noise ratio.

8. The time-varying frequency deviation dynamic tracking method according to claim 1, characterized in that, The initial value of the optimal frequency offset estimate is set to the initial coarse frequency offset estimate, and the initial value of the frequency offset change rate is set to 0.

9. The time-varying frequency deviation dynamic tracking method according to claim 1, characterized in that, The window length is adaptively adjusted based on the frequency offset rate of change, including: When the frequency deviation rate of change is less than or equal to the first rate of change threshold, the window length is adjusted to the maximum. When the frequency deviation rate of change is greater than the first rate of change threshold and less than or equal to the second rate of change threshold, the window length is adjusted to the middle. When the frequency deviation rate of change is greater than the second rate of change threshold, the window length is adjusted to the minimum.

10. The time-varying frequency deviation dynamic tracking method according to claim 1, characterized in that, The first derivative estimation method is used to calculate the frequency deviation rate.

Images