Anti-jamming timing synchronization method and apparatus

Abstract

This application relates to the field of measurement and control signal processing technology, and provides an anti-interference timing synchronization method and apparatus. The method includes sampling the received signal for symbol decision time to obtain in-phase and quadrature branch signals; jointly processing the in-phase and quadrature branch signals according to time and spatial dimensions to obtain a joint timing deviation estimate; adjusting the local data clock using the joint timing deviation estimate to obtain an adjusted local data clock; and feeding back the adjusted local data clock to guide the next symbol decision time sampling. In harsh environments with incomplete carrier lock and inter-symbol self-noise, compared to traditional M&B detectors, this application can shorten the timing synchronization lock setup time and improve the tracking robustness and steady-state accuracy of the feedback timing synchronization loop.

Description

Technical Field

[0001] This application relates to the field of measurement and control signal processing technology, and in particular to an anti-interference timing synchronization method and device. Background Technology

[0002] In spaceborne telemetry and control transponders, independent clock deviations between the transmitting and receiving ends, as well as channel transmission delays, can lead to timing errors, severely impacting demodulation performance. In digital communication systems, achieving high-precision signal timing synchronization is a primary prerequisite for telemetry and control transponders to accurately extract telemetry and control information and maintain the stability of the telemetry and control link.

[0003] Currently, common feedback timing synchronization detectors in related technologies include the Gardner detector and the M&B detector. The Gardner detector requires two sampling points per symbol, resulting in double oversampling and significant hardware resource consumption. The M&B detector, based on symbol rate sampling, uses one sampling point per symbol, offering the advantage of low resource consumption and making it more suitable for resource-constrained spaceborne platforms. However, the M&B detector has inherent drawbacks: when adjacent symbols have the same polarity, it generates severe inter-symbol self-noise, easily leading to loop lockout; and by default, the I / Q branches are equally weighted, so when the carrier is not locked, constellation rotation can cause single-branch noise to dominate, slowing down convergence. Summary of the Invention

[0004] In view of this, this application aims to propose an anti-interference timing synchronization method, which can significantly shorten the lock establishment time of timing synchronization and improve the tracking robustness and steady-state accuracy of the feedback timing synchronization loop.

[0005] To achieve the above objectives, the technical solution of this application is implemented as follows:

[0006] An anti-interference timing synchronization method includes:

[0007] The received signal is sampled at the symbol decision time to obtain the in-phase branch signal and the quadrature branch signal. The in-phase branch signal and the quadrature branch signal are then jointly processed according to the time dimension and the spatial dimension to obtain the joint timing deviation estimate.

[0008] The local data clock is adjusted using the joint timing deviation estimate to obtain the adjusted local data clock;

[0009] Feedback is provided to the adjusted local data clock to guide the sampling of the next symbol decision time.

[0010] Furthermore, the step of sampling the received signal at the symbol decision time to obtain the in-phase branch signal and the quadrature branch signal includes:

[0011] Guided by the local data clock, the received signal after matched filtering is interpolated and filtered to obtain the complex sampled signal at the symbol decision time.

[0012] The complex sampled signal is decomposed into in-phase branch signals and quadrature branch signals.

[0013] Furthermore, the in-phase branch signal and the quadrature branch signal are jointly processed according to the time and spatial dimensions to obtain a joint timing deviation estimate, including:

[0014] The timing deviation of each branch signal is estimated by performing time-dimensional smoothing filtering on the signal.

[0015] Dynamic weights of each branch signal are calculated in the spatial dimension to obtain the dynamic weights of each branch.

[0016] The branch timing deviation estimates of each branch are weighted and fused using the dynamic weights of each branch to obtain the joint timing deviation estimate.

[0017] Furthermore, the time-dimensional smoothing filtering process for each branch signal includes:

[0018] Using the sampled values ​​of multiple symbol times contained in the in-phase branch signal and the quadrature branch signal, extract the sampled value of the current symbol and the sampled value of the previous adjacent symbol;

[0019] Based on the sampled value of the current symbol and the sampled value of the previous adjacent symbol, determine whether the polarity of the current symbol and the previous adjacent symbol are the same.

[0020] When the polarity is the same, the timing deviation estimate of the previous valid time is called as the branch timing deviation estimate of the current time.

[0021] When the polarities are different and the signal-to-noise ratio is greater than or equal to the preset threshold, the estimated branch timing deviation at the current moment is calculated based on the sampled signal of the current symbol and the pre-stored value is updated.

[0022] Even when the polarity is different but the signal-to-noise ratio is less than the preset threshold, the branch timing deviation estimate from the previous valid time step is still used.

[0023] Furthermore, determining whether the polarity of the current symbol is the same as that of the previous adjacent symbol includes:

[0024] Calculate the product of the sampled signal of the current symbol and the sampled signal of the previous adjacent symbol;

[0025] The sign of the product determines whether adjacent symbols have the same polarity; a positive product indicates that adjacent symbols have the same polarity, and a negative product indicates that adjacent symbols have different polarities.

[0026] Furthermore, the dynamic weight calculation of the spatial dimension for each branch signal includes:

[0027] Calculate the instantaneous power of the in-phase branch signal and the quadrature branch signal;

[0028] The instantaneous power is subjected to sliding window averaging filtering to obtain the smoothed power of each branch;

[0029] The dynamic weight of each branch is obtained based on the proportion of the smoothed power of each branch to the total smoothed power of the two branches.

[0030] The dynamic weight is positively correlated with the smoothing power of the branch signal; the branch with the higher the smoothing power receives the higher weight.

[0031] Furthermore, adjusting the local data clock using the joint timing bias estimate includes:

[0032] The joint timing deviation estimate is input into the loop filter for filtering to obtain the filter control quantity;

[0033] The filter control quantity is converted into a control word for a numerically controlled oscillator, which is used to adjust the period and phase of the local data sampling clock.

[0034] Furthermore, the loop filter is a second-order digital loop filter, including a proportional branch and an integral branch.

[0035] Compared with related technologies, this application has the following advantages:

[0036] (1) The anti-interference timing synchronization method described in this application constructs a two-dimensional joint anti-interference architecture based on symbol rate sampling, encompassing both time and space dimensions. In the time dimension, the polarity state of adjacent symbols is identified through a polarity judgment factor. When the polarities are the same, the estimated value from the previous effective moment is used for zero-order smoothing, avoiding the control quantity abrupt changes, loop oscillations, and even lockout problems caused by traditional direct zeroing and hard truncation, effectively suppressing self-noise interference between symbols. In the space dimension, the inherent assumption of symmetrical equal weighting of the I / Q dual branches is broken, and a dynamic weight based on the confidence level of instantaneous power is introduced, allowing the high signal-to-noise ratio branch to dominate loop control and automatically shielding the pure noise pollution of inferior branches. After two-dimensional joint processing, the local clock is adjusted by loop filtering and feedback from a numerically controlled oscillator to form closed-loop control. This application ensures that the error control quantity received by the loop filter is always dominated by the branch with the highest current signal-to-noise ratio, thereby effectively accelerating the convergence speed of the phase-locked loop, shortening the timing synchronization lock-up establishment time of the receiver, and improving the tracking robustness and steady-state accuracy of the feedback timing synchronization loop in environments with low signal-to-noise ratio and residual phase deviation.

[0037] (2) By inheriting the advantage of symbol rate sampling based on the M&B algorithm, it eliminates the need for multiple oversampling as required by the traditional Gardner algorithm, reducing data throughput by half. Simultaneously, the introduced adaptive weights and polarity judgment logic can be implemented using low-complexity shift or absolute value approximation operations, saving multiplier and logic resources on the onboard FPGA. Based on this lower resource consumption, this application overcomes the inherent weakness of single-sampling detectors, which are susceptible to inter-symbol self-noise interference, through dynamic shielding in the spatial dimension and zero-order preservation in the temporal dimension. This effectively reduces normalized timing jitter during the steady-state tracking stage, achieving a balance between low resource overhead and high performance.

[0038] (3) By decomposing the joint processing into three sub-steps—time-dimensional smoothing filtering, spatial-dimensional dynamic weight calculation, and weighted fusion—modular decoupling of the two-dimensional joint processing architecture is achieved, enabling the inter-symbol self-noise suppression in the time domain and the signal-to-noise ratio weighted fusion in the spatial domain to be executed in parallel during the timing bias estimation process, thereby reducing the processing delay of the loop.

[0039] (4) By judging whether the polarity of the current symbol is the same as that of the previous adjacent symbol, if the polarity is the same, the estimated value of the previous effective time is called to maintain zero-order smoothness, avoiding the control quantity mutation and loop lockout problem caused by the traditional hard cut-off method, and maintaining the continuity of the closed-loop control quantity.

[0040] (5) The implementation method is simple, requiring only comparators and multipliers, and has low hardware overhead, making it suitable for implementation on spaceborne FPGAs.

[0041] (6) By calculating the instantaneous power of the I / Q signals and determining the dynamic weight of each branch according to the power ratio, the effective signal energy ratio of each branch can be quantified in real time, providing accurate weight coefficients for subsequent weighted fusion.

[0042] (7) By adjusting the local clock after the joint timing deviation estimate is successively processed through loop filtering and numerical control oscillator conversion, the closed-loop conversion of error quantity to clock control quantity is realized, which is the standard implementation method of feedback timing synchronization.

[0043] (8) By using a second-order digital loop filter, and by using a proportional branch to achieve instantaneous phase adjustment and an integral branch to achieve cumulative frequency deviation compensation, it can achieve zero steady-state error tracking of fixed frequency deviation and has the ability to track frequency ramp-up, making it suitable for high dynamic environments.

[0044] Another objective of this application is to provide an anti-interference timing synchronization device, comprising:

[0045] The sampling module is used to sample the received signal at the symbol decision time to obtain the in-phase branch signal and the quadrature branch signal;

[0046] A joint detection module, connected to the sampling module, is used to jointly process the in-phase branch signal and the quadrature branch signal according to the time dimension and the spatial dimension to obtain a joint timing deviation estimate.

[0047] The clock adjustment module is connected to both the joint detection module and the sampling module. It is used to adjust the local data clock using the joint timing deviation estimate to obtain the adjusted local data clock. The adjusted local data clock is then fed back to the sampling module to guide the sampling of the next symbol decision time.

[0048] Furthermore, the joint detection module includes:

[0049] The polarity determination submodule is used to extract the sampled value of the current symbol and the sampled value of the previous adjacent symbol by using the sampled values ​​of multiple symbol times contained in the in-phase branch signal and the quadrature branch signal, determine whether the polarity of the current symbol and the previous adjacent symbol are the same, and perform time dimension smoothing filtering on the branch timing deviation estimate based on the determination result.

[0050] The weight calculation submodule is used to calculate the dynamic weight of each branch based on the instantaneous power of the in-phase branch signal and the quadrature branch signal;

[0051] The weighted fusion submodule is connected to the polarity judgment submodule and the weight calculation submodule, respectively, and is used to sum the timing deviation estimates of the two branches after time-dimensional smoothing filtering according to their respective dynamic weights, and output the joint timing deviation estimate.

[0052] The anti-interference timing synchronization device described in this application achieves two-dimensional joint anti-interference under symbol rate sampling, and has the advantages of fast locking, strong robustness, high steady-state accuracy, and low hardware overhead, making it suitable for spaceborne platforms. Attached Figure Description

[0053] The accompanying drawings, which form part of this application, are used to provide a further understanding of this application. The illustrative embodiments and descriptions of this application are used to explain this application and do not constitute an undue limitation of this application. In the drawings:

[0054] Figure 1 This is a flowchart illustrating the anti-interference timing synchronization method described in the embodiments of this application.

[0055] Figure 2 This is a schematic diagram of the process for calculating the joint timing deviation estimate as described in the embodiments of this application.

[0056] Figure 3 This is a block diagram illustrating the principle of the feedback timing synchronization method described in the embodiments of this application.

[0057] Figure 4 This is a schematic diagram illustrating the QPSK signal constellation rotation and branch energy asymmetry phenomena described in the embodiments of this application.

[0058] Figure 5 This is a schematic diagram of the anti-interference timing synchronization device described in the embodiments of this application.

[0059] Figure 6 This is a schematic diagram of the joint detection module described in an embodiment of this application.

[0060] Explanation of reference numerals in the attached diagram: 410, Sampling module; 420, Joint detection module; 430, Clock adjustment module.

[0061] 421. Polarity determination submodule; 422. Weight calculation submodule; 423. Weighted fusion submodule.

[0062] in, Figure 4 The markings 1, 2, 3, and 4 in the diagram represent the positions of the four QPSK constellation points after the actual overall rotation, under the adverse condition of residual phase deviation θ when the carrier is not fully locked. Detailed Implementation

[0063] To make the technical solution and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0064] It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other.

[0065] Furthermore, it should be noted that in the description of this application, if terms such as "upper," "lower," "inner," or "outer" appear, indicating orientation or positional relationship, these are based on the orientation or positional relationship shown in the accompanying drawings and are only for the convenience of describing this application and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation on this application. In addition, if terms such as "first" or "second" appear, they are also used for descriptive purposes only and should not be construed as indicating or implying relative importance.

[0066] Furthermore, in the description of this application, unless otherwise expressly defined, the terms "installation," "connection," "joining," and "connector" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection between two components. Those skilled in the art can understand the specific meaning of the above terms in this application in light of the specific circumstances.

[0067] In this application, the terms "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., refer to a specific feature, structure, material, or characteristic described in connection with that embodiment or example, which is included in at least one embodiment or example of this application. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0068] The present application will now be described in detail through exemplary embodiments. However, it should be understood that, without further description, elements, structures, and features in one embodiment may be advantageously incorporated into other embodiments.

[0069] It should be noted that this application describes QPSK modulation as an example, but the method is not limited to this. For constant envelope or quasi-constant envelope modulations such as BPSK, 8PSK, and π / 4-QPSK, the energy projection characteristics of the I / Q branches still hold, and this method can be directly applied. For higher-order QAM modulations such as 16QAM and 64QAM, the signal amplitude is no longer constant. In this case, it is recommended to introduce amplitude normalization processing for dynamic weight calculation, or replace the instantaneous power with the sliding window average power, to reduce the impact of amplitude fluctuations on weight stability.

[0070] An embodiment of the first aspect of this application provides an anti-interference timing synchronization method. This method is based on symbol rate sampling, performing matched filtering and interpolation sampling on the received signal to obtain symbol sampling sequences of in-phase and quadrature branch signals. Then, it determines whether adjacent symbols have the same polarity; if the polarities are the same, it calls the previous valid estimate for zero-order preserving smoothing to suppress inter-symbol self-noise. Based on instantaneous power, it calculates the confidence weight of each branch, dynamically weights and fuses the errors of the two branch signals, adaptively shielding noise pollution from inferior branches to obtain a joint estimate. The joint estimate is then filtered through a loop and fed back to adjust the local data clock, achieving closed-loop timing synchronization. This shortens the locking setup time for timing synchronization and improves the tracking robustness and steady-state accuracy of the feedback timing synchronization loop, even in harsh environments with incomplete carrier lock and strong self-noise.

[0071] Among related technologies, timing synchronization techniques are mainly divided into forward timing estimation and feedback timing synchronization. While forward timing estimation can use the phase information of the signal spectrum to estimate timing deviation, it requires resampling at a multiple of the symbol rate, resulting in high implementation complexity and low accuracy, making it unsuitable for resource-constrained aerospace telemetry and control systems.

[0072] In contrast, feedback timing synchronization methods are more suitable for aerospace telemetry and control scenarios. The working principle of feedback timing synchronization is similar to a phase-locked loop (PLL), which uses feedback error information to continuously adjust the local data bit sampling pulses, gradually bringing the local clock closer to and locking onto the timing information of the received signal. However, existing feedback timing synchronization detectors face serious limitations in complex aerospace telemetry and control scenarios.

[0073] Currently, common feedback timing synchronization detectors in related technologies, such as Gardner and M&B detectors, have two major drawbacks: First, when adjacent symbols have the same polarity, the M&B detector will generate severe self-noise interference. Traditional hard truncation methods will cause abrupt changes in the loop control quantity, which can easily lead to loop lockout under low signal-to-noise ratio. Second, existing detectors are generally based on the assumption of symmetrical and equal weights of in-phase and quadrature branches (i.e., I / Q dual branches). When the carrier is not fully locked, the constellation diagram rotation will cause one branch to be occupied by pure noise. Equal weight superposition will seriously reduce the convergence speed.

[0074] Furthermore, traditional zero-crossing estimation schemes such as the Gardner detector rely heavily on the physical zero-crossing point in the transition region between adjacent symbols for polarity determination and error extraction. This mechanism detects polarity reversal by extracting data sampling points from two adjacent symbols and using the amplitude difference in the formula: when the polarities are opposite, the waveform crosses the zero axis, the amplitude difference is large, and the system uses the transition sampling point exactly between the two points to extract timing errors; when the polarities are the same, the waveform does not cross the zero axis, the amplitude difference approaches zero, and self-noise is suppressed by setting it to zero through mathematical operations. Due to the existence of the above physical mechanism, zero-crossing detection algorithms require the system to operate at two or more times the symbol rate oversampling in terms of hardware architecture, that is, each symbol requires at least two accurate sampling points: the data point and the zero-crossing point. This directly doubles the analog-to-digital conversion throughput of the receiver front-end, the data bus width of the back-end, and the computational resources of the interpolation filter, significantly increasing the hardware overhead and system power consumption of the onboard signal processing pipeline.

[0075] In view of this, in order to overcome the shortcomings of related technologies, the anti-interference timing synchronization method in this embodiment combines... Figure 1 The overall design includes the following steps S110~S130.

[0076] Step S110: Sample the received signal for symbol decision time to obtain the in-phase branch signal and the quadrature branch signal; and perform joint processing of the in-phase branch signal and the quadrature branch signal according to the time dimension and the spatial dimension to obtain the joint timing deviation estimate.

[0077] In some embodiments, refer to Figures 2-3 In step S110, the received signal is sampled at the symbol decision time to obtain the in-phase branch signal and the quadrature branch signal, and then jointly processed. Specifically, it may include the following steps S111 to S114.

[0078] Step S111: Establish the received signal model, and perform matched filtering and interpolation filtering on the received signal to obtain the complex sampled signal. The complex sampled signal is decomposed into in-phase branch signals and quadrature branch signals.

[0079] In step S111, under the guidance of the local data clock, the received signal after matched filtering is interpolated to obtain a complex sampled signal for the symbol decision time, which is used for subsequent timing deviation estimation. In this embodiment, the interpolation filter uses a cubic Lagrange interpolation implemented with a Farrow structure, and the sampled value for the optimal decision time is calculated in real time based on the phase deviation of the local data clock. The Farrow structure has the advantages of fixed coefficients and ease of hardware implementation, making it suitable for spaceborne FPGA applications.

[0080] Specifically, in some exemplary implementations, signal models are constructed for the transmitting baseband modulated signal and the receiving white Gaussian noise channel. The transmitting baseband modulated signal s(t) is shown in Equation 1:

[0081] (Formula 1).

[0082] Where t is a continuous-time variable, i is a discrete transmission symbol sequence index, and a i For transmitting complex signal sequences, T is the symbol period and τ is the transmission delay. The energy of each symbol signal is E S Unity power pulse shaping filter h T (t) represents a root-raised cosine Nyquist filter with a roll-off factor of β. In a specific implementation, the transmitter can be configured to use QPSK modulation, with a symbol period T set to 0.2 microseconds, corresponding to a symbol rate of 5 Mbps, and a pulse shaping filter h. T (t) A root-raised cosine quister filter with a roll-off factor β=0.5 is used.

[0083] More specifically, the receiver passes through a matched filter h R The output signal r(t) after (t) is shown in Formula 2:

[0084] (Formula 2).

[0085] Where e is the natural constant and j is the imaginary unit. Representing the residual carrier phase, we have , This is complex Gaussian white noise, where the real and imaginary parts have zero mean and zero variance. The noise power is , It is a Nyquist pulse that satisfies the condition of no inter-symbol interference.

[0086] More specifically, the complex sampled signal x is obtained by interpolating and filtering the matched filter output signal r(t). k Let the decision time be denoted as . , express The estimated value is then the complex sampled signal x at the decision time. k As shown in Formula 3:

[0087] (Formula 3).

[0088] in, The sampling time noise is represented by the time deviation. , , Complex sampled signal x k It can be represented as two orthogonal components. and , respectively represent complex sampled signals x k The real and imaginary parts.

[0089] The in-phase and quadrature branch signals obtained in step S111 contain timing deviation information, but they are susceptible to interference from data correlation self-noise generated when adjacent symbols have the same polarity and branch energy asymmetry caused by constellation diagram rotation when the carrier is not locked. Therefore, steps S112 and S113 are also required to perform anti-interference processing in the time and spatial dimensions.

[0090] Additionally, it should be noted that adjacent symbols are the sampled values ​​of the kth and (k-1th)th symbols after interpolation at the receiving end, not the original symbols at the sending end.

[0091] Step S112: Perform time-dimension smoothing filtering on the signals of each branch to obtain the estimated timing deviation of each branch.

[0092] In step S112, the sampled values ​​of multiple symbols contained in the in-phase and quadrature branch signals are used to extract the sampled value of the current symbol and the sampled value of the previous adjacent symbol. Based on the sampled values ​​of the current symbol and the previous adjacent symbol, it is determined whether the polarity of the current symbol and the previous adjacent symbol are the same. If the polarity is the same, the pre-stored branch timing deviation estimate from the previous valid time is used as the branch timing deviation estimate for the current time. If the polarity is different and the signal-to-noise ratio is greater than or equal to a preset threshold, the branch timing deviation estimate for the current time is calculated based on the sampled signal of the current symbol, and the pre-stored value is updated. If the polarity is different but the signal-to-noise ratio is less than the preset threshold, the branch timing deviation estimate from the previous valid time is still used. That is, it is necessary to use the in-phase and quadrature branch signals to calculate the branch timing deviation estimate for each branch at the current time.

[0093] Specifically, based on the timing detection principle of symbol rate sampling, the initial branch timing deviation estimates of the in-phase branch and the quadrature branch at the current time are calculated respectively. The initial timing deviation estimates reflect the original error information without self-noise suppression.

[0094] More specifically, the estimated branch timing deviation of the in-phase branch signal. As shown in Formula 4:

[0095] (Formula 4);

[0096] Initial branch timing deviation estimate of orthogonal branch signals As shown in Formula 5:

[0097] (Formula 5).

[0098] in, This represents the initial branch timing deviation estimate of the in-phase branch signal at the current time. This represents the initial branch timing deviation estimate of the orthogonal branch signal at the current time. The real parts of the sampled signals of the current k-th symbol and the previous (k-1)-th symbol are extracted, i.e., I-path, and denoted as Ik and Ik respectively. k and I k-1 Extract the imaginary parts of the sampled signals of the current k-th symbol and the previous (k-1)-th symbol, i.e., the Q-path, denoted as Qk. k and Q k-1 . and These represent the instantaneous power of the in-phase branch signal and the quadrature branch signal at the current symbol time, respectively.

[0099] More specifically, a hard decision is made on the sampled signal using the sign function sign(·) to determine whether the polarity of adjacent transmitted symbols has changed. At this time, the polarity determination factor α of the in-phase branch signal... I Polarity determination factor α of orthogonal branch signals Q As shown in formulas 6 and 7:

[0100] (Formula 6);

[0101] (Formula 7).

[0102] Here, `sign(·)` is the sign function, which outputs 1 when the input is greater than 0, 0 when the input is equal to 0, and -1 when the input is less than 0. Specifically, the definition of the sign function `sign(·)` is as follows:

[0103] .

[0104] When the product of adjacent symbols is less than or equal to 0, they are determined to have different polarities (valid moment); when the product of adjacent symbols is greater than 0, they are determined to have the same polarity (invalid moment).

[0105] It should be noted that, as an optional implementation, the polarity is determined by calculating the sign of the product of the sampled signal of the current symbol and the sampled signal of the previous adjacent symbol; a positive product indicates that the adjacent symbols have the same polarity, and a negative product indicates that the adjacent symbols have different polarities. Specifically, if the adjacent symbols have the same polarity, the polarity determination factor is 1, indicating that the current moment is susceptible to inter-symbol self-noise interference, and the timing deviation estimate of the previous valid moment is used as the branch timing deviation estimate of the current moment. If the adjacent symbols have different polarities, the polarity determination factor is 0, indicating that the timing estimate is valid at this moment, and the branch timing deviation estimate of the current moment is calculated based on the sampled signal of the current symbol. It is understood that those skilled in the art can also use other equivalent methods to determine the polarity, such as directly comparing the sign bits of the two sampled values, and these methods all fall within the protection scope of this application.

[0106] Furthermore, the above processing method is called "zero-order preserving smoothing," which means that when adjacent symbols have the same polarity, the estimated value of the previous effective time step is kept unchanged to avoid abrupt changes in the control quantity. The effective time step is the moment when adjacent symbols have different polarities and the signal-to-noise ratio (SNR) is greater than or equal to a preset threshold SNRth, for example, the preset threshold SNRth is 6dB. The selection of the preset threshold SNRth is determined according to the modulation scheme and the target bit error rate, and the value range is 3dB to 10dB. When the SNR is lower than this threshold, even if adjacent symbols have different polarities, the reliability of the timing deviation estimate is also low. In this case, it is advisable to keep the estimated value of the previous effective time step unchanged to avoid introducing additional noise due to misestimation under low SNR. Here, this threshold corresponds to the lowest SNR condition for the receiver to reliably perform symbol decision. Below this threshold, the confidence of the symbol decision result decreases significantly, and the reliability of the timing deviation estimate is low, making it unsuitable for loop updates.

[0107] In a specific embodiment, the signal-to-noise ratio corresponding to the preset threshold SNRth can be estimated in real time using the following sliding window averaging method:

[0108] Calculate the instantaneous power P at the current k-th symbol time. k As shown in Formula 8:

[0109] (Formula 8);

[0110] Among them, I k and Q k These are the sampled signal values ​​for the in-phase branch and the quadrature branch, respectively.

[0111] Calculate the average power of the signal within a sliding window of length N. As shown in Formula 9:

[0112] (Formula 9).

[0113] in, This represents the average power of the signal after smoothing through a sliding window of length N at the k-th symbol time. It is represented as the combined instantaneous power of the I / Q branch corresponding to the i-th symbol within the window.

[0114] Noise power is estimated using the difference between adjacent symbols. As shown in Formula 10:

[0115] (Formula 10).

[0116] in, Let represent the channel equivalent noise average power after averaging over an N-point sliding window at time k, and let i represent the i-th sampled symbol within the window. Represented as the sampled signal value of the in-phase branch at the i-th symbol time; Represented as the in-phase branch sample value of the (i-1)th adjacent symbol, Represented as adjacent sign difference of in-phase branches, Represented as the sampled signal value of the orthogonal branch at the i-th symbol time; Represented as the sampled value of the orthogonal branch of the (i-1)th adjacent symbol. This is represented as the adjacent symbol difference of orthogonal branches.

[0117] Signal-to-noise ratio estimate at the current time As shown in Formula 11:

[0118] (Formula 11).

[0119] When SNRest(k) is greater than or equal to SNRth, the current time is determined to be a valid time, and the currently calculated timing deviation estimate is adopted; otherwise, the estimate of the previous valid time remains unchanged. In this embodiment, the sliding window length N can be 16, 32, or 64, and the specific value is determined based on the trade-off between symbol rate and channel change rate. The larger the value of N, the smoother the signal-to-noise ratio estimation, but the slower the response to channel changes; the smaller the value of N, the faster the response, but the larger the estimation jitter. That is, the optimal value is determined through a limited number of trials based on the actual application scenario. The above algorithm can be implemented using adders, multipliers, and shift registers in an FPGA, without the need for dividers or floating-point units, making it suitable for the hardware resource constraints of spaceborne platforms.

[0120] After calculating the estimated branch timing deviation values ​​for each branch in step S112, the raw error information without anti-interference processing is reflected. However, the estimated branch timing deviation values ​​are severely contaminated by self-noise when adjacent symbols have the same polarity, and cannot be used directly. Therefore, the polarity judgment factor of adjacent symbols is calculated simultaneously to identify whether the estimated branch timing deviation value at the current moment is valid. In addition, to address the issue of branch energy asymmetry when the carrier is not locked, step S113 is also required to calculate the confidence weight of each branch based on instantaneous power.

[0121] Step S113: Calculate the dynamic weight of each branch signal in the spatial dimension to obtain the dynamic weight of each branch.

[0122] In step S113, the instantaneous power of the in-phase branch signal and the quadrature branch signal is calculated; the instantaneous power is subjected to sliding window averaging filtering to obtain the smoothed power of each branch; and the dynamic weight of each branch is obtained according to the proportion of the smoothed power of each branch to the total smoothed power of the two branches. The dynamic weight is positively correlated with the smoothed power of the branch signal, and the branch with the greater smoothed power receives a higher weight.

[0123] like Figure 4 As shown, the hollow circles in the figure represent the ideal QPSK constellation point positions without phase deviation (e.g., at 45°, 135°, 225°, and 315°). Points 1, 2, 3, and 4 represent the actual QPSK constellation point positions after overall rotation under adverse conditions where the carrier is not fully locked and a residual phase deviation θ exists. In some exemplary embodiments, when the QPSK constellation diagram rotates and a constellation point moves closer to the coordinate axis due to carrier incomplete locking, most of the effective signal energy of the transmitted symbol is projected onto the in-phase branch, exhibiting a larger amplitude; while the projection component on the quadrature branch is extremely small. At this time, the sampled signal values ​​of the quadrature branch are almost entirely dominated by channel Gaussian white noise and inter-symbol interference, resulting in a very low signal-to-noise ratio. This embodiment utilizes this observation to introduce dynamic weighting processing of the spatial dimension.

[0124] Specifically, this embodiment introduces a dynamic confidence weight based on instantaneous power. Taking into account the energy projection characteristics of constant envelope modulation signals in orthogonal space, it first extracts the instantaneous power of the in-phase branch signal and the quadrature branch signal at the current symbol time in real time. and Considering that in extremely low signal-to-noise ratio environments such as aerospace telemetry and control, the instantaneous power of a single symbol is severely contaminated by Gaussian white noise in the channel, resulting in drastic random fluctuations. Directly using the instantaneous power of a single symbol to calculate the weights would cause frequent jumps in the dynamic weights between symbols, introducing additional modulation noise and deteriorating the tracking jitter performance of the loop. Therefore, to obtain a stable and reliable energy assessment benchmark, a sliding window averaging process is introduced after extracting the instantaneous power.

[0125] Specifically, the sliding window length is set to L, and the moving average power of the in-phase branch and the quadrature branch within the current window is calculated respectively. and The calculation is shown in Formulas 12 and 13:

[0126] (Formula 12);

[0127] (Formula 13).

[0128] Where k represents the discrete-time index of the current symbol; m represents the historical time offset within the sliding window (i.e., the summation index variable), and its value ranges from 0 to L-1; I k-m With Q k-m and represent the sampled signal values ​​of the in-phase branch and the quadrature branch at the k-th symbol time, respectively.

[0129] Furthermore, in this embodiment, the sliding window length L can be an integer between 4 and 16, with a preferred value of 8. The selection of the window length L requires a trade-off between smoothing effect and dynamic response speed: a larger L results in smoother power estimation and stronger noise immunity, but reduces the tracking response to rapid channel changes; a smaller L results in faster response, but larger power estimation jitter. In the initial reception phase, when the number of received symbols is less than the window length L, a partial window averaging method can be used (i.e., the window length is the current number of available symbols), or the power estimate from the previous moment can be used until the window is full. This embodiment uses a partial window averaging method to ensure that an effective power estimate is obtained in the initial phase.

[0130] Subsequently, the average power after smoothing and filtering is normalized, and the dynamic weight ω of the confidence score of the in-phase branch signal at the current time is calculated. I (k) and dynamic weights ω of orthogonal branch signal confidence Q (k) As shown in Equations 14 and 15:

[0131] (Formula 14);

[0132] (Equation 15).

[0133] Where, ω I (k) represents the dynamic weight of the confidence level of the in-phase branch signal at the current time, ω Q (k) represents the dynamic weight of the confidence level of the orthogonal branch signal at the current time. and These represent the moving average power of the in-phase branch and the orthogonal branch within the current window, respectively.

[0134] The aforementioned weight calculation process based on sliding window averaging is particularly suitable for deep space telemetry and control scenarios with extremely low signal-to-noise ratios and strong background noise. Under these harsh conditions, energy integration over multiple symbol periods can effectively filter out high-frequency noise spikes and prevent abrupt changes in loop control quantities. Simultaneously, by rationally configuring the window length L, an optimal engineering balance can be achieved between smoothing and denoising capabilities and dynamic response delay with extremely low onboard FPGA hardware overhead, thereby meeting the stringent requirements of onboard transponders for high steady-state synchronization accuracy.

[0135] Among them, the confidence dynamic weight ω I (k) and ω Q(k) is used to quantify the proportion of effective signal energy in each branch. The larger the weight, the higher the reliability of the timing estimate of that branch. It is also used to characterize the relative proportion of effective signal energy contained in each branch when there is a residual bias in the carrier phase or the constellation diagram is rotated. When the instantaneous power of a branch is significantly greater than that of another branch, the confidence weight of that branch approaches 1, while the confidence weight of the other branch approaches 0.

[0136] In step S113, the polarity judgment factor in the time dimension and the confidence weight in the spatial dimension are obtained. Error abrupt changes when polarities are the same in the time dimension can cause loop oscillations or even loss of lock. In the spatial dimension, noise from inferior branches can contaminate the loop control quantity, leading to slow convergence. Processing these two factors independently cannot effectively address the harsh environment of carrier lock-in and strong self-noise. Therefore, after obtaining the joint timing deviation estimate in step S113, step S114 is still required to complete the two-dimensional joint anti-interference processing.

[0137] Step S114: Use the dynamic weights of each branch to perform weighted fusion of the timing deviation estimates of each branch to obtain the joint timing deviation estimate.

[0138] In step S114, the branch timing deviation estimate from step S112, the polarity judgment factor, and the confidence dynamic weight from step S113 are combined, and the deviation estimate is determined based on the principles of smoothing filtering and two-dimensional spatial weighting. As shown in Formula 16:

[0139] (Formula 16).

[0140] in, Expressed as the joint bias estimate, ω I (k) and ω Q (k) represents the dynamic confidence weights of the in-phase branch signal and the quadrature branch signal at the current time, respectively. and Let α represent the estimated branch timing deviations of the in-phase branch signal and the quadrature branch signal at the current time, respectively. I and α Q These represent the polarity determination factors for in-phase and quadrature branch signals, respectively. and These represent the estimated timing deviations of the in-phase and quadrature branch signals at the previous valid time, respectively. Initially, and All values ​​are set to 0. Specifically, during receiver startup or the initial stage of burst communication, since there is no estimate of the previous valid time, we can wait for the first pair of adjacent symbols with different polarities to appear, and then directly use the branch timing deviation estimate calculated at the current time as the valid estimate, and initialize it accordingly. and .

[0141] It should be noted that the timing deviation estimate of each branch at the previous effective time is smoothed to zero order; this is achieved through dynamic weight ω. I (k) and ω Q (k) The two error paths are adaptively combined to automatically suppress inferior branches with low instantaneous effective energy and high noise ratio, and output a high-precision joint timing deviation estimate. .

[0142] In addition, the errors of the two branches after time-dimensional processing are multiplied by their respective spatial confidence dynamic weights ω. I (k) and ω Q (k), and then sum them to obtain the final joint timing deviation estimate. The role of dynamic weight is: when the carrier is not locked, the effective signal energy is mainly projected onto a certain branch, and the weight of that branch is automatically increased, while the weight of the other branch dominated by noise is automatically decreased, so as to achieve adaptive shielding of inferior branches.

[0143] After calculating the estimated joint timing deviation in step S110, although it has undergone two-dimensional anti-interference processing, it still contains high-frequency noise components. If it is directly used to adjust the local clock, it will cause loop jitter. Therefore, after obtaining the estimated joint timing deviation in step S110, step S120 still needs to be executed to convert the error into a clock control quantity to achieve closed-loop regulation.

[0144] Step S120: Adjust the local data clock using the joint timing deviation estimate to obtain the adjusted local data clock.

[0145] In step S120, the joint timing deviation estimate is input into a loop filter for filtering to obtain a filter control quantity; the filter control quantity is then converted into a control word for a digitally controlled oscillator to adjust the period and phase of the local data sampling clock. The loop filter is a second-order digital loop filter, including proportional and integral branches.

[0146] Specifically, the joint timing deviation estimate output in step S110 is... The signal is fed into a second-order digital loop filter for smoothing, which filters out high-frequency noise and smooths tracking jitter. The output control quantity v(k) of the loop filter in the discrete-time domain is shown in Equation 17:

[0147] (Equation 17).

[0148] Where v(k) is the loop filter output control quantity at time k, v(k-1) is the loop filter output at the previous time, C1 is the proportional branch gain coefficient, and C2 is the integral branch gain coefficient. At receiver startup, the loop filter output v(k) is initialized to 0, and the cumulative term of the integral branch is initialized to 0. After the first joint timing error estimate e(0) is input, the loop begins the acquisition phase.

[0149] It should be noted that the gain coefficients C1 and C2 are configured according to the preset loop damping factor and loop noise bandwidth. In this embodiment, the damping factor ζ = 0.707 and the loop noise bandwidth Bn = 0.01 / T (T is the symbol period) are taken. The corresponding gain coefficients C1 and C2 are shown in Formulas 18 and 19.

[0150] (Formula 18);

[0151] (Equation 19).

[0152] The above gain is calculated by substituting the damping factor ζ=0.707 and the loop bandwidth Bn=0.01 / T into the second-order loop formula, and we can take C1≈0.028 and C2≈0.0004.

[0153] After obtaining the adjusted local data clock in step S120, there is still a residual deviation between the adjusted local clock and the actual timing. If we stop here, we can only perform a one-time timing estimation and adjustment, and the residual deviation cannot be detected and corrected. Therefore, after obtaining the adjusted local data clock in step S120, we still need to execute step S130 to gradually eliminate the timing deviation and continuously track the time-varying drift of the clock.

[0154] Step S130: Feedback is provided to the adjusted local data clock to guide the sampling of the next symbol decision time.

[0155] In step S130, the loop filter control quantity v(k) output in step S120 is converted into a timing frequency control word for the numerically controlled oscillator. Based on the timing frequency control word, the period and phase of the local data bit sampling pulse are compensated and adjusted in real time. The updated local data clock, as a feedback signal, is input to the interpolation filter in step S110 to guide signal interpolation and optimal decision-time sampling for the next symbol period.

[0156] The loop filter output control quantity v(k) is converted into the timing frequency control word Δf(k) of the numerically controlled oscillator (NCO) through a linear mapping, and the conversion relationship is shown in Equation 20:

[0157] Δf(k)=K NCO×v(k) (Formula 20).

[0158] Among them, K NCO This is the gain coefficient of the numerically controlled oscillator, determined by the oscillator bit width, system clock frequency, and symbol period, and is a constant. The adjusted control word is used for real-time compensation and adjustment of the period and phase of the local data bit sampling pulse.

[0159] Thus, through continuous feedback and iteration in this process, the local data clock gradually approaches and locks onto the true timing information of the received signal, achieving high-precision closed-loop timing synchronization locking.

[0160] Continue to combine Figures 1-4 As a preferred implementation, when the received signal has a large carrier frequency offset, such as greater than 5% of the symbol rate, the timing synchronization performance will be affected by the phase rotation introduced by the frequency offset. In this embodiment, the dynamic weight is based on instantaneous power and is not sensitive to phase rotation, and can still effectively distinguish between effective branches and noisy branches. However, it is recommended that the acquisition bandwidth of the timing synchronization loop be designed to be greater than twice the expected frequency offset.

[0161] Continue to combine Figures 1-4 As a preferred implementation, in burst communication, it is not possible to rely on historical statistical information to initialize the "previous effective moment estimate" to 0, and to accelerate convergence by using known sequences in the burst lead phase, or to preset a set of empirical weights until the first effective moment occurs.

[0162] Furthermore, it is worth noting that traditional single-sampling algorithms are prone to generating severe self-noise when polarities are the same, and performance deteriorates due to constellation diagram rotation when the carrier is not locked. This embodiment overcomes the lock-out problem by maintaining the continuity and momentum of the closed-loop control quantity through zero-order time-dimensional maintenance; simultaneously, it uses spatial-dimensional dynamic confidence weighting to real-time filter out inferior branches occupied by pure noise. In low signal-to-noise ratio and carrier-locked environments, this embodiment not only surpasses the traditional M&B algorithm in loop acquisition convergence speed and lock-out robustness, but also outperforms the oversampled Gardner detector, which requires twice the hardware resources, in steady-state tracking accuracy. Moreover, this application successfully breaks the technical convention that high-precision synchronization necessarily relies on high sampling rates and high hardware overhead, achieving a performance leap of high robustness and high precision under extremely low resource boundaries through two-dimensional joint digital purification of time and space.

[0163] An embodiment of the second aspect of this application provides an anti-interference timing synchronization device, such as... Figure 5 As shown, the anti-interference timing synchronization device includes a sampling module 410, a joint detection module 420, and a clock adjustment module 430.

[0164] The sampling module 410 is used to sample the received signal at the symbol decision time to obtain the in-phase branch signal and the quadrature branch signal. The joint detection module 420 is connected to the sampling module 410 and is used to jointly process the in-phase branch signal and the quadrature branch signal according to the time and spatial dimensions to obtain a joint timing deviation estimate. The clock adjustment module 430 is connected to both the sampling module 410 and the joint detection module 420, and is used to adjust the local data clock using the joint timing deviation estimate to obtain the adjusted local data clock; and feeds back the adjusted local data clock to the sampling module 410 to guide the next symbol decision time sampling.

[0165] In addition, such as Figure 6 As shown, the aforementioned joint detection module 420 includes a polarity judgment submodule 421, a weight calculation submodule 422, and a weighted fusion submodule 423.

[0166] Specifically, the polarity determination submodule 421 is used to extract the sampled value of the current symbol and the sampled value of the previous adjacent symbol from the sampled values ​​of multiple symbols contained in the in-phase branch signal and the quadrature branch signal, determine whether the polarity of the current symbol and the previous adjacent symbol are the same, and perform time-dimensional smoothing filtering on the branch timing deviation estimate based on the determination result. The weight calculation submodule 422 is used to calculate the dynamic weight of each branch based on the instantaneous power of the in-phase branch signal and the quadrature branch signal. The weighted fusion submodule 423 is connected to the polarity determination submodule 421 and the weight calculation submodule 422 respectively, and is used to perform weighted summation of the two branch timing deviation estimates after time-dimensional smoothing filtering according to their respective dynamic weights, and output the joint timing deviation estimate.

[0167] Among them, the polarity judgment submodule 421 and the weight calculation submodule 422 are executed in parallel, and there is no data dependency between them; the weighted fusion submodule 423 is executed after waiting for the output results of the two.

[0168] More specifically, in the implementation of the anti-interference timing synchronization device of this embodiment, the above-mentioned modules can be existing module products with data transmission, storage or computing functions.

[0169] In practical applications, the specific implementation process of the functions of each module in the anti-interference timing synchronization device of this embodiment can be found in the relevant descriptions in the above method embodiments, and will not be repeated here.

[0170] The anti-interference timing synchronization device in this embodiment acquires two signals through a sampling module. A joint detection module suppresses self-noise abrupt changes in the time dimension using polarity judgment and zero-order hold, and uses instantaneous power weighting to shield against contamination from inferior branches in the spatial dimension. Finally, a clock adjustment module provides closed-loop feedback. This two-dimensional joint anti-interference mechanism, implemented under symbol rate sampling, offers advantages such as fast locking, strong robustness, high steady-state accuracy, and low hardware overhead, making it suitable for spaceborne platforms.

[0171] The above descriptions are merely some embodiments of this application and are not intended to limit this application. The technical features or structures in the foregoing different embodiments can be arbitrarily combined to form other specific technical solutions as needed. For those skilled in the art, this application can have various modifications and variations. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of protection of the claims of this application.

Claims

1. An anti-interference timing synchronization method, characterized in that: The received signal is sampled at the symbol decision time to obtain the in-phase branch signal and the quadrature branch signal. The in-phase branch signal and the quadrature branch signal are then jointly processed according to the time dimension and the spatial dimension to obtain the joint timing deviation estimate. The local data clock is adjusted using the joint timing deviation estimate to obtain the adjusted local data clock; Feedback is provided to the adjusted local data clock to guide the sampling of the next symbol decision time.

2. The interference-free timing synchronization method according to claim 1, wherein, The step of sampling the received signal at the symbol decision time to obtain the in-phase branch signal and the quadrature branch signal includes: Guided by the local data clock, the received signal after matched filtering is interpolated and filtered to obtain the complex sampled signal at the symbol decision time. The complex sampled signal is decomposed into in-phase branch signals and quadrature branch signals.

3. The interference-robust timing synchronization method of claim 1, wherein The in-phase branch signal and the quadrature branch signal are jointly processed according to the time and spatial dimensions to obtain a joint timing deviation estimate, including: The timing deviation of each branch signal is estimated by performing time-dimensional smoothing filtering on the signal. Dynamic weights of each branch signal are calculated in the spatial dimension to obtain the dynamic weights of each branch. The timing deviation estimates of each branch are weighted and fused using the dynamic weights of each branch to obtain the joint timing deviation estimate.

4. The anti-interference timing synchronization method according to claim 3, characterized in that, The time-dimensional smoothing filtering process for each branch signal includes: Using the sampled values ​​of multiple symbol times contained in the in-phase branch signal and the quadrature branch signal, extract the sampled value of the current symbol and the sampled value of the previous adjacent symbol; Based on the sampled value of the current symbol and the sampled value of the previous adjacent symbol, determine whether the polarity of the current symbol and the previous adjacent symbol are the same. When the polarity is the same, the branch timing deviation estimate of the previous valid time is called as the branch timing deviation estimate of the current time. When the polarities are different and the signal-to-noise ratio is greater than or equal to the preset threshold, the estimated branch timing deviation at the current moment is calculated based on the sampled signal of the current symbol and the pre-stored value is updated. Even when the polarity is different but the signal-to-noise ratio is less than the preset threshold, the branch timing deviation estimate from the previous valid time step is still used.

5. The anti-interference timing synchronization method according to claim 4, characterized in that, The determination of whether the polarity of the current symbol is the same as that of the previous adjacent symbol includes: Calculate the product of the sampled signal of the current symbol and the sampled signal of the previous adjacent symbol; The sign of the product determines whether adjacent symbols have the same polarity; a positive product indicates that adjacent symbols have the same polarity, and a negative product indicates that adjacent symbols have different polarities.

6. The anti-interference timing synchronization method according to claim 3, characterized in that, Dynamic weighting of each branch signal in the spatial dimension is calculated, including: Calculate the instantaneous power of the in-phase branch signal and the quadrature branch signal; The instantaneous power is subjected to sliding window averaging filtering to obtain the smoothed power of each branch; The dynamic weight of each branch is obtained based on the proportion of the smoothed power of each branch to the total smoothed power of the two branches. The dynamic weight is positively correlated with the smoothing power of the branch signal; the branch with the higher the smoothing power receives the higher weight.

7. The anti-interference timing synchronization method according to claim 1, characterized in that, Adjusting the local data clock using the joint timing bias estimate includes: The joint timing deviation estimate is input into the loop filter for filtering to obtain the filter control quantity; The filter control quantity is converted into a control word for a numerically controlled oscillator, which is used to adjust the period and phase of the local data sampling clock.

8. The anti-interference timing synchronization method according to claim 7, characterized in that: The loop filter is a second-order digital loop filter, including a proportional branch and an integral branch.

9. An anti-interference timing synchronization device, characterized in that, include: The sampling module is used to sample the received signal at the symbol decision time to obtain the in-phase branch signal and the quadrature branch signal; A joint detection module, connected to the sampling module, is used to jointly process the in-phase branch signal and the quadrature branch signal according to the time dimension and the spatial dimension to obtain a joint timing deviation estimate. The clock adjustment module is connected to both the joint detection module and the sampling module. It is used to adjust the local data clock using the joint timing deviation estimate to obtain the adjusted local data clock. The adjusted local data clock is then fed back to the sampling module to guide the sampling of the next symbol decision time.

10. The anti-interference timing synchronization device according to claim 9, characterized in that, The joint detection module includes: The polarity determination submodule is used to extract the sampled value of the current symbol and the sampled value of the previous adjacent symbol by using the sampled values ​​of multiple symbol times contained in the in-phase branch signal and the quadrature branch signal, determine whether the polarity of the current symbol and the previous adjacent symbol are the same, and perform time dimension smoothing filtering on the branch timing deviation estimate based on the determination result. The weight calculation submodule is used to calculate the dynamic weight of each branch based on the instantaneous power of the in-phase branch signal and the quadrature branch signal; The weighted fusion submodule is connected to the polarity judgment submodule and the weight calculation submodule, respectively, and is used to sum the timing deviation estimates of the two branches after time-dimensional smoothing filtering according to their respective dynamic weights, and output the joint timing deviation estimate.

Images