Multi-carrier 4096 QAM (Quadrature Amplitude Modulation) signal coordination synchronization method and system

Abstract

The invention provides a multi-carrier 4096 QAM (Quadrature Amplitude Modulation) signal coordinated synchronization method and system, and relates to the technical field of digital communication. The method comprises the following steps: performing down-conversion and analog-to-digital conversion on a multi-carrier 4096 QAM signal to obtain a digital baseband signal; the method comprises the following steps: extracting a time-frequency spectrogram and a pilot symbol sequence of a multicarrier 4096QAM signal, and performing forward reasoning by taking the time-frequency spectrogram and the pilot symbol sequence as joint features to obtain a multi-dimensional estimation vector; performing time domain interpolation adjustment on the digital baseband signal to obtain a preliminary alignment time domain signal; pre-compensating the preliminary alignment time domain signal according to the preliminary estimation value of the multi-dimensional estimation vector to obtain a frequency offset compensation signal; performing state prediction on the phase track of the frequency offset compensation signal, and applying a reverse phase rotation on the frequency offset compensation signal according to an output value of state prediction to obtain a synchronized signal; and carrying out equalization and 4096QAM demodulation on the synchronized signal to obtain a data bit stream. According to the invention, the efficiency of coordination and synchronization of the multi-carrier 4096 QAM signal can be improved.

Description

Technical Field

[0001] This invention relates to the field of digital communication technology, and in particular to a method and system for coordinating and synchronizing multi-carrier 4096QAM signals. Background Technology

[0002] In existing technologies, the parameter estimation stage of multi-carrier 4096QAM signals does not combine the time-spectrum diagram and pilot symbol sequence as joint features for analysis. Relying solely on single features for parameter estimation fails to fully utilize the complementary information between different features, resulting in low estimation accuracy for key synchronization parameters such as symbol timing error and carrier frequency deviation. Furthermore, the parameter estimation process does not standardize the features or optimize the feature mapping relationship through nonlinear transformations, leading to poor accuracy and insufficient robustness in the generated estimation vectors. This makes it difficult to meet the high accuracy requirements of synchronization parameters for high-order modulation signals like 4096QAM, creating potential problems for subsequent synchronization adjustments.

[0003] Furthermore, existing technologies also have shortcomings in the synchronization correction and phase compensation stages of multi-carrier 4096QAM signals. During time-domain alignment, the interpolation direction and reference sampling point are not dynamically determined based on a precise estimate of the symbol timing error; instead, a fixed interpolation strategy is used to adjust the digital baseband signal, which easily leads to unreasonable sampling point selection and low alignment accuracy. During frequency offset compensation, the phase rotation sequence is not accurately calculated in conjunction with the sampling period; only coarse frequency adjustments are used to offset the deviation, failing to completely eliminate the impact of carrier frequency offset. In the phase correction stage, the phase trajectory of the frequency offset compensation signal is not dynamically predicted; only static phase adjustment or simple feedback mechanisms are used to correct the phase, which cannot adapt to local fluctuations and global changes in the phase trajectory. This results in phase deviations remaining in the synchronized signal, ultimately affecting the equalization and demodulation effects and making it difficult to output high-quality data bitstreams. Summary of the Invention

[0004] In view of this, the present invention provides a method and system for coordinating and synchronizing multi-carrier 4096QAM signals to solve the problems mentioned in the background art. The present invention can improve the efficiency of coordinating and synchronizing multi-carrier 4096QAM signals.

[0005] To achieve the above objectives, the technical solution provided by the present invention is as follows:

[0006] A method for coordinating and synchronizing multi-carrier 4096QAM signals includes the following steps:

[0007] Step 1: Perform down-conversion and analog-to-digital conversion on the received multi-carrier 4096QAM signal to obtain the digital baseband signal of the multi-carrier 4096QAM signal;

[0008] Step 2: Extract the time spectrum and pilot symbol sequence of the multi-carrier 4096QAM signal, and use the time spectrum and pilot symbol sequence as joint features for forward inference to obtain the multi-dimensional estimation vector of the multi-carrier 4096QAM signal;

[0009] Step 3: Based on the preliminary estimate of the symbol timing error in the multidimensional estimation vector, perform time-domain interpolation adjustment on the digital baseband signal to obtain the preliminary aligned time-domain signal of the multi-carrier 4096QAM signal;

[0010] Step 4: Based on the preliminary estimate of the carrier frequency deviation in the multidimensional estimation vector, perform frequency offset pre-compensation on the pre-aligned time-domain signal to obtain the frequency offset compensation signal of the multi-carrier 4096QAM signal.

[0011] Step 5: Perform state prediction on the phase trajectory of the frequency offset compensation signal. Based on the output value of the state prediction, apply a reverse phase rotation to the frequency offset compensation signal to obtain the synchronized signal of the multi-carrier 4096QAM signal.

[0012] Step 6: Equalize and demodulate the synchronized signal using 4096QAM to obtain the data bit stream of the multi-carrier 4096QAM signal.

[0013] Furthermore, the multi-carrier 4096QAM signal received in step 1 is a radio frequency signal, and the specific method of step 1 is as follows:

[0014] Step 101: The received radio frequency signal is mixed with the local oscillator signal generated by the local oscillator through a mixer, thereby downconverting the radio frequency signal to an intermediate frequency signal.

[0015] Step 102: Use an anti-aliasing filter to filter the intermediate frequency signal, suppress out-of-band noise and image frequency interference, and obtain the filtered intermediate frequency signal;

[0016] Step 103: Sample and quantize the filtered intermediate frequency signal to obtain a digital intermediate frequency signal;

[0017] Step 104: Convert the digital intermediate frequency signal to the baseband frequency and separate the in-phase component and quadrature component to obtain the digital baseband signal of the multi-carrier 4096QAM signal.

[0018] Further, in step 2, the time-spectrum diagram and pilot symbol sequence of the multi-carrier 4096QAM signal are extracted, specifically as follows:

[0019] Step 201: Divide the digital baseband signal into continuous signal frames;

[0020] Step 202: Perform a short-time Fourier transform on each signal frame to obtain the time-frequency unit corresponding to the signal frame;

[0021] Step 203: Combine the time-frequency units in time order to form the time-spectrum diagram of the multi-carrier 4096QAM signal;

[0022] Step 204: Based on the pilot pattern predefined in the communication protocol, locate and extract pilot symbols from the frequency domain resources of the digital baseband signal to obtain the pilot symbol sequence of the multi-carrier 4096QAM signal.

[0023] Furthermore, in step 2, the time-spectrum diagram and pilot symbol sequence are used as joint features for forward inference to obtain the multi-dimensional estimation vector of the multi-carrier 4096QAM signal. Specifically, the method is as follows:

[0024] Step 205: Perform channel normalization on the time-frequency spectrum and phase unwinding on the pilot symbol sequence to obtain standardized time-frequency features and normalized pilot features;

[0025] Step 206: Concatenate the standardized time-frequency features with the normalized pilot features to obtain the joint feature vector of the multi-carrier 4096QAM signal;

[0026] Step 207: Perform a nonlinear transformation on the multi-carrier 4096QAM signal based on the joint eigenvector to obtain the multidimensional estimation vector of the multi-carrier 4096QAM signal.

[0027] Furthermore, the specific method of step 207 is as follows:

[0028] ;

[0029] In the formula, For multidimensional estimation vectors, It is the Sigmoid activation function. The hyperbolic tangent activation function is used. This is the dominant weight matrix in the nonlinear transformation. For joint feature vectors, This is the dominant bias vector in the nonlinear transformation. This is the gate weight matrix in the nonlinear transformation. This is the gated bias vector in the nonlinear transformation. For Hadama accumulation.

[0030] Furthermore, the specific method for step 3 is as follows:

[0031] Step 301: Extract the preliminary estimate of the symbol timing error from the multidimensional estimation vector;

[0032] Step 302: Based on the preliminary estimate of the symbol timing error and the symbol, determine the direction and reference sampling point of the time-domain interpolation operation;

[0033] Step 303: Based on the reference sampling points, the sampling sequence of the digital baseband signal is resampled to obtain interpolated sampling points, and the original sampling points are discarded.

[0034] Step 304: Recombine the interpolated sampling points with the original sampling points after selection in chronological order to obtain the preliminary aligned time domain signal of the multi-carrier 4096QAM signal.

[0035] Furthermore, step 4 is specifically implemented as follows:

[0036] Step 401: Extract a preliminary estimate of the carrier frequency deviation from the multidimensional estimation vector;

[0037] Step 402: Determine the angular velocity of the multi-carrier 4096QAM signal based on the preliminary estimate of the carrier frequency deviation and the sampling period;

[0038] Step 403: The phase value of each element in the multi-carrier 4096QAM signal is proportional to the product of the angular velocity and used as the phase rotation sequence;

[0039] Step 404: In the time domain, the phase of the initially aligned time domain signal is reversed based on the phase rotation sequence to obtain the frequency offset compensation signal of the multi-carrier 4096QAM signal.

[0040] Furthermore, in step 5, the phase trajectory of the frequency offset compensation signal is predicted in the following way:

[0041] Step 501: Extract a sequence of historical phase trajectories containing at least one symbol period length from the memory cell of the nonlinear phase tracker;

[0042] Step 502: Capture the local fluctuation patterns and global change trends in the historical phase trajectory sequence to obtain a phase dynamic feature vector;

[0043] Step 503: The optimal estimated state vector of the previous moment is concatenated with the phase dynamic feature vector to obtain the enhanced state vector of the multi-carrier 4096QAM signal.

[0044] Step 504: Under the influence of phase dynamic characteristics, the enhanced state vector is nonlinearly evolved to obtain the predicted state vector of the multi-carrier 4096QAM signal.

[0045] Furthermore, the formula for calculating the predicted state vector is as follows:

[0046] ;

[0047] In the formula, for The predicted state vector at time t. This is a linear state transition function based on the classical state-space model. for The optimal estimated state vector at time t. As a time factor, It is a nonlinear mapping function. For phase dynamic eigenvectors, This is a historical phase trajectory sequence.

[0048] A multi-carrier 4096QAM signal coordination and synchronization system, comprising:

[0049] The signal processing module is used to perform down-conversion and analog-to-digital conversion on the received multi-carrier 4096QAM signal to obtain the digital baseband signal of the multi-carrier 4096QAM signal;

[0050] The multidimensional estimation vector extraction module is used to extract the time spectrum and pilot symbol sequence of the multi-carrier 4096QAM signal, and use the time spectrum and pilot symbol sequence as joint features for forward inference to obtain the multidimensional estimation vector of the multi-carrier 4096QAM signal;

[0051] The time-domain interpolation module is used to perform time-domain interpolation adjustment on the digital baseband signal based on the preliminary estimate of the symbol timing error in the multidimensional estimation vector, so as to obtain the preliminary aligned time-domain signal of the multi-carrier 4096QAM signal.

[0052] The frequency offset pre-compensation module is used to perform frequency offset pre-compensation on the initially aligned time domain signal based on the preliminary estimate of the carrier frequency offset in the multi-dimensional estimation vector, so as to obtain the frequency offset compensation signal of the multi-carrier 4096QAM signal.

[0053] The signal synchronization module is used to predict the phase trajectory of the frequency offset compensation signal. Based on the output value of the state prediction, a reverse phase rotation is applied to the frequency offset compensation signal to obtain the synchronized signal of the multi-carrier 4096QAM signal.

[0054] The signal conditioning module is used to equalize and demodulate the synchronized signal using 4096QAM to obtain a data bit stream of a multi-carrier 4096QAM signal.

[0055] Compared with the prior art, the present invention has the following beneficial effects:

[0056] 1. This invention provides a reliable foundation for multi-carrier 4096QAM signal synchronization through joint feature extraction and precise parameter estimation. The received signal undergoes down-conversion, filtering, and analog-to-digital conversion to generate a clean digital baseband signal. The signal's time-frequency spectrum and pilot symbol sequence are extracted, and a joint feature vector is constructed through normalization, phase unwrapping, and feature concatenation. A multi-dimensional estimation vector containing symbol timing error and carrier frequency deviation is obtained through nonlinear transformation, enabling accurate prediction of key synchronization parameters and providing a reliable basis for subsequent synchronization adjustments.

[0057] 2. This invention significantly improves the accuracy and stability of multi-carrier 4096QAM signal synchronization by employing step-by-step synchronization optimization and dynamic phase correction. Based on the timing error and frequency offset estimates in the multi-dimensional estimation vector, time-domain interpolation alignment and frequency offset pre-compensation are performed on the digital baseband signal to initially eliminate synchronization deviations. By extracting historical phase trajectory sequences, capturing dynamic features, and constructing an enhanced state vector, the phase trajectory of the frequency offset compensation signal is accurately predicted. Reverse phase rotation is applied to achieve phase calibration. Finally, a high-quality data bitstream is output after equalization and demodulation, comprehensively ensuring the efficiency and reliability of signal synchronization. Attached Figure Description

[0058] Figure 1 This is a flowchart illustrating a multi-carrier 4096QAM signal coordination and synchronization method according to an embodiment of the present invention.

[0059] Figure 2 This is a schematic diagram of a multi-carrier 4096QAM signal coordination and synchronization system provided in an embodiment of the present invention. Detailed Implementation

[0060] The present invention will now be described in further detail with reference to the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0061] A multi-carrier 4096QAM signal coordination and synchronization method, comprising:

[0062] S1. Perform down-conversion and analog-to-digital conversion on the received multi-carrier 4096QAM signal to obtain the digital baseband signal of the multi-carrier 4096QAM signal;

[0063] S2. Extract the time spectrum and pilot symbol sequence of the multi-carrier 4096QAM signal, and use the time spectrum and pilot symbol sequence as joint features for forward inference to obtain the multi-dimensional estimation vector of the multi-carrier 4096QAM signal;

[0064] S3. Based on the preliminary estimate of the symbol timing error in the multidimensional estimation vector, the digital baseband signal is time-domain interpolated and adjusted to obtain the preliminary aligned time-domain signal of the multi-carrier 4096QAM signal.

[0065] S4. Based on the preliminary estimate of the carrier frequency deviation in the multidimensional estimation vector, perform frequency offset pre-compensation on the pre-aligned time domain signal to obtain the frequency offset compensation signal of the multi-carrier 4096QAM signal.

[0066] S5. Perform state prediction on the phase trajectory of the frequency offset compensation signal. Based on the output value of the state prediction, apply a reverse phase rotation to the frequency offset compensation signal to obtain the synchronized signal of the multi-carrier 4096QAM signal.

[0067] S6. Equalize and demodulate the synchronized signal with 4096QAM to obtain the data bit stream of the multi-carrier 4096QAM signal.

[0068] In a preferred embodiment, the received multi-carrier 4096QAM signal undergoes down-conversion and analog-to-digital conversion to obtain a digital baseband signal of the multi-carrier 4096QAM signal, including:

[0069] The received radio frequency signal is mixed with the local oscillator signal generated by the local oscillator through a mixer, and the radio frequency signal is down-converted to an intermediate frequency signal.

[0070] An anti-aliasing filter is used to filter the intermediate frequency signal, suppressing out-of-band noise and image frequency interference, to obtain the filtered intermediate frequency signal;

[0071] The filtered intermediate frequency signal is sampled and quantized to obtain a digital intermediate frequency signal;

[0072] The digital intermediate frequency signal is converted into the baseband frequency, and the in-phase and quadrature components are separated to obtain the digital baseband signal of the multi-carrier 4096QAM signal.

[0073] In a preferred embodiment, the time-spectrum diagram and pilot symbol sequence of the multi-carrier 4096QAM signal are extracted, including:

[0074] Divide the digital baseband signal into consecutive signal frames;

[0075] Perform a short-time Fourier transform on each signal frame to obtain the time-frequency unit corresponding to the signal frame;

[0076] The time-frequency units are combined in time order to form the time-spectrum diagram of a multi-carrier 4096QAM signal;

[0077] Based on the pilot pattern predefined by the communication protocol, pilot symbols are located and extracted from the frequency domain resources of the digital baseband signal to obtain the pilot symbol sequence of the multi-carrier 4096QAM signal.

[0078] In a preferred embodiment, the time-spectrum graph and pilot symbol sequence are used as joint features for forward inference to obtain a multidimensional estimation vector of the multi-carrier 4096QAM signal, including:

[0079] Channel normalization is performed on the time-frequency spectrum, and phase unwrapping is performed on the pilot symbol sequence to obtain standardized time-frequency features and normalized pilot features;

[0080] By concatenating the standardized time-frequency features with the normalized pilot features, a joint feature vector of the multi-carrier 4096QAM signal is obtained.

[0081] A nonlinear transformation is performed on the multi-carrier 4096QAM signal based on the joint eigenvector to obtain the multidimensional estimation vector of the multi-carrier 4096QAM signal.

[0082] In a preferred embodiment, a nonlinear transformation is performed on the multi-carrier 4096QAM signal based on the joint eigenvector to obtain a multidimensional estimation vector of the multi-carrier 4096QAM signal. Specifically, the method is as follows:

[0083] ;

[0084] In the formula, For multidimensional estimation vectors, It is the Sigmoid activation function. The hyperbolic tangent activation function is used. This is the dominant weight matrix in the nonlinear transformation. For joint feature vectors, This is the dominant bias vector in the nonlinear transformation. This is the gate weight matrix in the nonlinear transformation. This is the gated bias vector in the nonlinear transformation. For Hadama accumulation.

[0085] In a preferred embodiment, based on a preliminary estimate of the symbol timing error in the multidimensional estimation vector, the digital baseband signal is subjected to time-domain interpolation adjustment to obtain a preliminary aligned time-domain signal of the multi-carrier 4096QAM signal, including:

[0086] A preliminary estimate of the symbol timing error is extracted from the multidimensional estimation vector;

[0087] Based on the preliminary estimate of the symbol timing error and the sign, determine the direction and reference sampling point of the time-domain interpolation operation;

[0088] Based on the reference sampling points, the sampling sequence of the digital baseband signal is resampled to obtain interpolated sampling points and to discard the original sampling points.

[0089] The interpolated sampling points are recombined with the original sampling points after selection in chronological order to obtain the preliminary aligned time-domain signal of the multi-carrier 4096QAM signal.

[0090] In a preferred embodiment, based on the preliminary estimate of the carrier frequency deviation in the multidimensional estimation vector, frequency offset pre-compensation is performed on the initially aligned time-domain signal to obtain the frequency offset compensation signal of the multi-carrier 4096QAM signal, including:

[0091] A preliminary estimate of the carrier frequency deviation is extracted from the multidimensional estimation vector;

[0092] Based on the preliminary estimate of the carrier frequency deviation and the sampling period, the angular velocity of the multi-carrier 4096QAM signal is determined.

[0093] The phase rotation sequence is obtained by multiplying the phase value of each element in the multi-carrier 4096QAM signal by the angular velocity.

[0094] In the time domain, the phase of the initially aligned time-domain signal is reversed based on the phase rotation sequence to obtain the frequency offset compensation signal of the multi-carrier 4096QAM signal.

[0095] In a preferred embodiment, state prediction of the phase trajectory of the frequency offset compensation signal includes:

[0096] Extract a sequence of historical phase trajectories containing at least one symbol period length from the memory cells of the nonlinear phase tracker;

[0097] By capturing local fluctuation patterns and global change trends in historical phase trajectory sequences, a phase dynamic feature vector is obtained;

[0098] The optimal estimated state vector from the previous moment is concatenated with the phase dynamic feature vector to obtain the enhanced state vector of the multi-carrier 4096QAM signal.

[0099] The enhanced state vector is nonlinearly evolved under the influence of phase dynamic characteristics to obtain the predicted state vector of the multi-carrier 4096QAM signal.

[0100] In a preferred embodiment, the formula for calculating the predicted state vector is as follows:

[0101] ;

[0102] In the formula, for The predicted state vector at time t. This is a linear state transition function based on the classical state-space model. for The optimal estimated state vector at time t. As a time factor, It is a nonlinear mapping function. For phase dynamic eigenvectors, This is a historical phase trajectory sequence.

[0103] A multi-carrier 4096QAM signal coordination and synchronization system, comprising:

[0104] The signal processing module is used to perform down-conversion and analog-to-digital conversion on the received multi-carrier 4096QAM signal to obtain the digital baseband signal of the multi-carrier 4096QAM signal;

[0105] The multidimensional estimation vector extraction module is used to extract the time spectrum and pilot symbol sequence of the multi-carrier 4096QAM signal, and use the time spectrum and pilot symbol sequence as joint features for forward inference to obtain the multidimensional estimation vector of the multi-carrier 4096QAM signal;

[0106] The time-domain interpolation module is used to perform time-domain interpolation adjustment on the digital baseband signal based on the preliminary estimate of the symbol timing error in the multidimensional estimation vector, so as to obtain the preliminary aligned time-domain signal of the multi-carrier 4096QAM signal.

[0107] The frequency offset pre-compensation module is used to perform frequency offset pre-compensation on the initially aligned time domain signal based on the preliminary estimate of the carrier frequency offset in the multi-dimensional estimation vector, so as to obtain the frequency offset compensation signal of the multi-carrier 4096QAM signal.

[0108] The signal synchronization module is used to predict the phase trajectory of the frequency offset compensation signal. Based on the output value of the state prediction, a reverse phase rotation is applied to the frequency offset compensation signal to obtain the synchronized signal of the multi-carrier 4096QAM signal.

[0109] The signal conditioning module is used to equalize and demodulate the synchronized signal using 4096QAM to obtain a data bit stream of a multi-carrier 4096QAM signal.

[0110] Here are more specific examples:

[0111] A multi-carrier 4096QAM signal coordination and synchronization method is disclosed. The execution subject of this method includes, but is not limited to, at least one of the following: a server, a terminal, or other electronic devices that can be configured to execute the method. In other words, the method can be executed by software or hardware installed on a terminal device or a server device. The server includes, but is not limited to, a single server, a server cluster, a cloud server, or a cloud server cluster. The server can be a standalone server or a cloud server providing basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, content delivery networks (CDNs), and big data and artificial intelligence platforms.

[0112] Figure 1 The flowchart shown is for this method, which includes the following steps:

[0113] S1. Perform down-conversion and analog-to-digital conversion on the received multi-carrier 4096QAM signal to obtain the digital baseband signal of the multi-carrier 4096QAM signal;

[0114] In this embodiment, the received multi-carrier 4096QAM signal undergoes down-conversion and analog-to-digital conversion to obtain the digital baseband signal of the multi-carrier 4096QAM signal, including:

[0115] The received radio frequency signal is mixed with the local oscillator signal generated by the local oscillator through a mixer, and the radio frequency signal is down-converted to an intermediate frequency signal.

[0116] An anti-aliasing filter is used to filter the intermediate frequency signal, suppressing out-of-band noise and image frequency interference, to obtain the filtered intermediate frequency signal;

[0117] The filtered intermediate frequency signal is sampled and quantized to obtain a digital intermediate frequency signal;

[0118] The digital intermediate frequency signal is converted into the baseband frequency, and the in-phase and quadrature components are separated to obtain the digital baseband signal of the multi-carrier 4096QAM signal.

[0119] Specifically, the received radio frequency (RF) signal is mixed with a local oscillator (LO) signal generated by a local oscillator using a mixer. When down-converting the RF signal to an intermediate frequency (IF) signal, the local oscillator generates a fixed-frequency LO signal. This LO signal has a fixed frequency difference from the received RF signal, and this difference is exactly equal to the target IF signal frequency. The RF signal and the LO signal are simultaneously input into the mixer. The mixer interacts with the two signals through signal superposition and frequency synthesis to generate a mixed signal containing multiple frequency components. The signal component whose frequency equals the difference between the RF signal and the LO signal is selected from the mixed signal; this component is the down-converted IF signal, ensuring that the signal frequency is reduced to a range suitable for subsequent processing.

[0120] Furthermore, an anti-aliasing filter is used to filter the intermediate frequency (IF) signal, suppressing out-of-band noise and image frequency interference. The resulting filtered IF signal is then input into the anti-aliasing filter, which has a fixed passband range that strictly matches the frequency range of the IF signal. The filter fully preserves the effective signal components within the passband range, allowing them to pass smoothly. It attenuates and suppresses out-of-band noise and image frequency interference generated during mixing, preventing them from passing through the filter. After processing, the effective signal is preserved while the interference components are significantly weakened, ultimately yielding the filtered IF signal.

[0121] Furthermore, the filtered intermediate frequency (IF) signal is sampled and quantized to obtain a digital IF signal. A fixed sampling frequency is used to periodically sample the filtered IF signal, acquiring the signal amplitude at a single instant each time. Continuous sampling forms a series of discrete signal amplitude data. For each acquired signal amplitude data point, it is classified according to a preset quantization level, converting the continuously changing analog amplitude into discrete digital codes. Each digital code uniquely corresponds to a signal amplitude range. By sampling, the continuous-time IF signal is converted into a discrete-time signal, and then quantization converts the analog amplitude signal into a digital signal, ultimately forming a digital IF signal containing only digital codes.

[0122] Furthermore, when converting the digital intermediate frequency (IF) signal to the baseband frequency and separating the in-phase and quadrature components to obtain the digital baseband signal of the multi-carrier signal, the IF signal undergoes frequency shifting. By adjusting the signal's frequency parameters, the frequency of the IF signal is reduced to the baseband frequency range, achieving frequency conversion from IF to baseband. During the frequency conversion process, phase decomposition is performed simultaneously on the signal. Based on the signal's phase characteristics, the signal is split into two mutually perpendicular components: one in-phase and the other quadrature. These two components retain all the information of the original IF signal and are independent of each other, ultimately yielding a digital baseband signal containing both in-phase and quadrature components.

[0123] In this embodiment, the received radio frequency (RF) signal is mixed with the local oscillator (LO) signal generated by the local oscillator using a mixer, downconverting the RF signal to an intermediate frequency (IF) signal. This reduces the signal frequency to a range suitable for subsequent processing. The LO signal and the RF signal maintain a fixed frequency difference, ensuring the stability of the downconverted IF signal and avoiding increased processing complexity due to excessively high frequencies. This lays the foundation for subsequent filtering, sampling, and other operations, improving the overall feasibility of signal processing.

[0124] An anti-aliasing filter is used to filter the intermediate frequency (IF) signal. The filter's passband range is strictly matched to the IF signal's frequency range, ensuring the complete preservation of the effective components in the IF signal while significantly attenuating out-of-band noise and image frequency interference generated during mixing. The filtered IF signal removes irrelevant interference components, significantly improving signal purity and preventing interference signals from affecting subsequent sampling and quantization accuracy, thus ensuring the accuracy of signal processing.

[0125] The filtered intermediate frequency (IF) signal is periodically sampled to obtain discrete signal amplitude data. Then, the continuous analog amplitudes are converted into discrete digital codes according to a preset quantization level. This process converts the analog IF signal into a digital IF signal, enabling the signal to be recognized and processed by digital processing equipment. At the same time, by using a fixed sampling frequency and quantization level, it ensures that the digital signal can accurately reproduce the characteristics of the original analog signal, providing a high-quality digital signal source for subsequent baseband conversion.

[0126] The frequency of the digital intermediate frequency (IF) signal is shifted to the baseband frequency range, completing the frequency conversion from IF to baseband. Simultaneously, based on the signal's phase characteristics, mutually perpendicular in-phase and quadrature components are separated. These two components completely preserve all the information of the original digital IF signal and are independent of each other. The resulting digital baseband signal meets the input requirements of subsequent processing stages such as synchronization, equalization, and demodulation in multi-carrier signals, ensuring that subsequent processing can accurately extract effective data from the signal and improving the coherence and reliability of the overall signal processing flow.

[0127] S2. Extract the time spectrum and pilot symbol sequence of the multi-carrier 4096QAM signal, and use the time spectrum and pilot symbol sequence as joint features for forward inference to obtain the multi-dimensional estimation vector of the multi-carrier 4096QAM signal;

[0128] In this embodiment, the time-spectrum diagram and pilot symbol sequence of the multi-carrier 4096QAM signal are extracted, including:

[0129] Divide the digital baseband signal into consecutive signal frames;

[0130] Perform a short-time Fourier transform on each signal frame to obtain the time-frequency unit corresponding to the signal frame;

[0131] The time-frequency units are combined in time order to form the time-spectrum diagram of a multi-carrier 4096QAM signal;

[0132] Based on the pilot pattern predefined by the communication protocol, pilot symbols are located and extracted from the frequency domain resources of the digital baseband signal to obtain the pilot symbol sequence of the multi-carrier 4096QAM signal.

[0133] By using the time-spectrum graph and pilot symbol sequence as joint features for forward inference, a multidimensional estimation vector for the multi-carrier 4096QAM signal is obtained, including:

[0134] Channel normalization is performed on the time-frequency spectrum, and phase unwrapping is performed on the pilot symbol sequence to obtain standardized time-frequency features and normalized pilot features;

[0135] By concatenating the standardized time-frequency features with the normalized pilot features, a joint feature vector of the multi-carrier 4096QAM signal is obtained.

[0136] A nonlinear transformation is performed on the multi-carrier 4096QAM signal based on the joint eigenvector to obtain the multidimensional estimation vector of the multi-carrier 4096QAM signal.

[0137] A nonlinear transformation is performed on the multi-carrier 4096QAM signal based on the joint eigenvector to obtain the multidimensional estimation vector of the multi-carrier 4096QAM signal. The specific method is as follows:

[0138] ;

[0139] In the formula, For multidimensional estimation vectors, It is the Sigmoid activation function. The hyperbolic tangent activation function is used. This is the dominant weight matrix in the nonlinear transformation. For joint feature vectors, This is the dominant bias vector in the nonlinear transformation. This is the gate weight matrix in the nonlinear transformation. This is the gated bias vector in the nonlinear transformation. For Hadama accumulation.

[0140] Specifically, when dividing a digital baseband signal into continuous signal frames, the continuously transmitted digital baseband signal is segmented according to a fixed time interval or data length standard. The segmentation process strictly adheres to the signal transmission timing, ensuring that the start and end positions of each signal frame are consistent with the natural transmission rhythm of the signal, without disrupting its integrity and continuity. Each segmented signal frame contains complete time-domain information, and the signal frames are seamlessly connected without overlap or omission, ultimately resulting in a series of continuous and structurally uniform signal frames.

[0141] Furthermore, when performing a short-time Fourier transform on each signal frame to obtain the corresponding time-frequency unit, a fixed time window function is applied to each signal frame segment. The time window function limits the analysis range of the signal frame, highlighting the signal's characteristics within a local time frame while reducing distortion at the frame edges. Based on the signal frames with the applied time window, the time-domain signal frames are transformed into time-frequency units containing both time and frequency information by analyzing the correspondence between the signal's frequency components and time. Each time-frequency unit accurately characterizes the frequency distribution of the corresponding signal frame within a specific time segment, fully preserving the signal's time-frequency coupling characteristics.

[0142] Furthermore, when combining time-frequency units in chronological order to construct the time-spectrum diagram of a multi-carrier signal, the time sequence of all time-frequency units is carefully considered to ensure that the arrangement of each time-frequency unit is synchronized with the transmission time of the original digital baseband signal. Following the chronological order, the time-frequency units are sequentially spliced ​​and integrated, ensuring a natural connection between the end position of the time-frequency unit corresponding to the previous signal frame and the beginning position of the time-frequency unit corresponding to the next signal frame. During the integration process, the frequency information and time markers of each time-frequency unit remain unchanged, ultimately forming a time-spectrum diagram that visually represents the frequency distribution of the multi-carrier signal at different time points.

[0143] Furthermore, based on the pilot pattern predefined in the communication protocol, pilot symbols are located and extracted from the frequency domain resources of the digital baseband signal to obtain the pilot symbol sequence of the multi-carrier signal. First, the pilot pattern predefined in the communication protocol is clarified. This pattern specifies in detail the distribution location, spacing rules, and identification characteristics of the pilot symbols in the frequency domain resources. Based on the provisions of the pilot pattern, a search is performed one by one in the frequency domain resources corresponding to the digital baseband signal to identify frequency domain positions that conform to the pilot identification characteristics; these positions are the distribution locations of the pilot symbols. The corresponding signal components are extracted from the identified frequency domain positions; these signal components are the pilot symbols. The extracted pilot symbols are arranged sequentially according to the order specified in the pilot pattern, ultimately forming an ordered pilot symbol sequence.

[0144] Specifically, when performing channel normalization on the time-frequency spectrum and phase unwrapping on the pilot symbol sequence to obtain standardized time-frequency features and regularized pilot features, for each channel of the time-frequency spectrum, the numerical distribution range of all time-frequency units within that channel is statistically analyzed. By adjusting the value of each time-frequency unit, the numerical distribution of that channel conforms to a unified standard, eliminating feature interference caused by amplitude differences between different channels, and obtaining standardized time-frequency features with consistent numerical scales. For the pilot symbol sequence, the phase information of each pilot symbol is analyzed, identifying and eliminating phase jump phenomena caused by signal transmission or processing, so that the phase of the pilot symbols exhibits a continuous and smooth change trend, avoiding the impact of phase discontinuity on subsequent feature processing, and obtaining regularized pilot features with regular phase patterns.

[0145] Furthermore, when concatenating standardized time-frequency features with normalized pilot features to obtain the joint feature vector of the multi-carrier signal, the time-frequency distribution information contained in the standardized time-frequency features and the pilot phase and amplitude information contained in the normalized pilot features are clearly defined, ensuring that the dimensions of the two types of features match and that the information does not overlap. Following a preset order, all feature elements of the standardized time-frequency features and all feature elements of the normalized pilot features are concatenated end-to-end to form a complete feature set. This set simultaneously covers the core information of both time-frequency domain features and pilot features, without losing any key details of either type of feature, ultimately yielding a joint feature vector that comprehensively reflects the characteristics of the multi-carrier signal.

[0146] Furthermore, when performing a nonlinear transformation on the multicarrier signal based on the joint eigenvector to obtain a multidimensional estimation vector, a fixed nonlinear transformation method is used to map each feature element in the joint eigenvector, breaking the linear correlation between feature elements and uncovering the complex nonlinear relationships hidden behind the features. The transformation converts the joint eigenvector from the original feature space to a new feature space, highlighting key characteristics of the multicarrier signal in the new space, such as signal distortion and channel attenuation. The transformed feature elements are then divided and reorganized according to preset dimensions, with each dimension corresponding to an estimation result of a signal characteristic, ultimately forming a multidimensional estimation vector containing estimates of multiple key signal characteristics.

[0147] Specifically, the joint feature vector is derived from the feature concatenation of standardized time-frequency features and normalized pilot features. First, the time-frequency spectrum is normalized to obtain standardized time-frequency features, and the pilot symbol sequence is unwrapped in phase to obtain normalized pilot features. Then, all elements of the two types of features are connected end to end in a preset order to form a joint feature vector that covers the core information of time-frequency domain features and pilot features.

[0148] Furthermore, the dominant weight matrix is ​​pre-set based on a large amount of multi-carrier signal sample data. By analyzing the influence of each element in the joint feature vector on the multidimensional estimation vector, a corresponding weight value is assigned to each element to ensure that elements that contribute significantly to the estimation of key signal characteristics play a more significant role.

[0149] Furthermore, the dominant bias vector is pre-set based on the characteristic distribution law of multi-carrier signal samples. It is used to compensate for the deviation caused by the difference in characteristic distribution during the calculation of the joint feature vector and the dominant weight matrix, and to correct the calculation benchmark to improve the estimation accuracy.

[0150] Furthermore, the gating weight matrix is ​​pre-set based on the correlation between different feature elements in the joint eigenvector. It is used to regulate the degree of participation of each feature element in the final estimation result, strengthen the role of effective features, and suppress the interference of redundant features. The gating bias vector is pre-set based on the regulation requirements of the gating mechanism. It is used to fine-tune the calculation results of the gating weight matrix and the joint eigenvector, so that the gating output better fits the generation requirements of the multidimensional estimation vector.

[0151] Furthermore, the formula signifies that, through the synergistic effect of dual activation and gating control, the joint feature vector is transformed into a multidimensional estimation vector that accurately reflects the key characteristics of multi-carrier signals. During calculation, the joint feature vector is first fused with the dominant weight matrix, and then the dominant bias vector is superimposed to complete the benchmark correction, yielding the first intermediate result. This intermediate result is then input into the Sigmoid activation function, which restricts the result to a fixed interval, allowing the numerical value to intuitively reflect the effectiveness of the features, thus obtaining the dominant activation result.

[0152] In detail, the joint feature vector and the gating weight matrix are fused and calculated simultaneously, and the gating bias vector is superimposed for fine-tuning to obtain the second intermediate result. This intermediate result is then input into the hyperbolic tangent activation function, and the result is mapped to a specific interval through function processing to achieve nonlinear control of the feature response, thus obtaining the gating activation result.

[0153] Furthermore, a Hadamard product is performed on the dominant activation result and the gated activation result, and the two types of results are multiplied element-wise to achieve information fusion. This retains the core estimation information in the dominant activation result while filtering out invalid components through the gated activation result. Finally, a multi-dimensional estimation vector is output through the fusion operation. Each dimension of this vector corresponds to the estimated value of a key characteristic of the multi-carrier signal, providing accurate quantitative basis for subsequent signal analysis and processing.

[0154] In this embodiment, the digital baseband signal is divided into continuous signal frames according to fixed time intervals or data lengths, strictly following the signal transmission timing sequence to ensure that each frame contains complete time-domain information and that there is seamless connection between frames without overlap or omission. This division method avoids information redundancy caused by analyzing long signals in a single operation, while also fully covering the entire digital baseband signal. This provides a well-organized processing unit for subsequent frame-by-frame time-frequency analysis, ensuring the comprehensiveness of signal feature extraction.

[0155] A time window function is applied to each signal frame to limit the analysis range and reduce frame edge distortion. Then, a short-time Fourier transform is used to convert the time-domain signal frame into a time-frequency unit containing time and frequency information. This process can accurately capture the frequency distribution pattern of each signal frame within a specific time segment, fully preserve the time-frequency coupling characteristics of the signal, and avoid the problem of insufficient time-frequency resolution caused by directly analyzing long time-domain signals, thus providing high-precision time-frequency data support for the subsequent construction of time-spectrum diagrams.

[0156] By sequentially splicing the time-frequency units corresponding to each frame in chronological order, while keeping the time identifier and frequency information of each time-frequency unit unchanged, a time-frequency spectrum of the multi-carrier signal is finally formed. This time-frequency spectrum can intuitively display the frequency composition and variation trend of the signal at different time points, clearly reflecting the time-frequency domain characteristics of the multi-carrier signal. It provides an intuitive and comprehensive time-frequency feature basis for subsequent joint feature extraction and parameter estimation, solving the problem that single time-domain or frequency-domain analysis cannot take into account time-frequency correlation information.

[0157] Based on the pilot pattern predefined in the communication protocol, the distribution, spacing, and identification characteristics of pilot symbols in the frequency domain resources are clearly defined. Then, frequency domain positions matching these characteristics are retrieved one by one from the frequency domain resources of the digital baseband signal, and the corresponding signal components are extracted as pilot symbols and arranged sequentially to form a pilot symbol sequence. This extraction method strictly adheres to the protocol standard, avoiding extraction deviations caused by misjudgment of pilot positions, ensuring that the pilot symbol sequence accurately reflects channel characteristics, and providing reliable pilot feature support for subsequent joint feature inference and synchronization parameter estimation.

[0158] When performing channel normalization on the time-frequency spectrum, the values ​​of the time-frequency units within each channel are adjusted to ensure that the numerical distribution of each channel conforms to a unified standard. This eliminates characteristic interference caused by amplitude differences between different channels, resulting in standardized time-frequency features with consistent scale. Phase unwrapping is performed on the pilot symbol sequence to identify and eliminate phase jump phenomena, ensuring that the phase exhibits a continuous and smooth change trend, thus obtaining regularized pilot features. These two processes respectively address the problems of inconsistent time-frequency feature scale and discontinuous pilot phase, ensuring that the extracted features accurately reflect the essential characteristics of the signal and laying a high-quality foundation for subsequent feature fusion and inference.

[0159] Standardized time-frequency features and regularized pilot features are concatenated in a predetermined order to form a joint feature vector. This vector simultaneously encompasses the signal time-frequency distribution information reflected in the time-spectrum diagram, as well as the channel phase and amplitude information carried by the pilot symbol sequence. By integrating two complementary types of features, the limitations of single feature information are avoided, enabling the joint feature vector to comprehensively characterize the complex characteristics of multi-carrier signals and providing richer and more comprehensive feature support for subsequent accurate inference of multi-dimensional estimation vectors.

[0160] When performing nonlinear transformations based on joint eigenvectors, the linear correlation between eigenelements is broken, and hidden complex nonlinear relationships are uncovered. This transforms the joint eigenvectors from the original feature space to a new space that better highlights the key characteristics of the signal. This transformation effectively enhances the characterization of key synchronization parameters such as symbol timing error and carrier frequency deviation. The resulting multidimensional estimation vector can accurately quantify these key parameters, solving the problems of linear transformations being unable to adapt to complex signal characteristics and having insufficient estimation accuracy. This provides accurate parameter basis for subsequent synchronization adjustments.

[0161] The Sigmoid activation function is used to process the calculation results of the joint feature vector with the dominant weight matrix and dominant bias vector, limiting the results to the range of 0 to 1, thus intuitively reflecting the effectiveness of the features in parameter estimation. Simultaneously, the hyperbolic tangent activation function is used to map the calculation results of the joint feature vector with the gated weight matrix and gated bias vector, adjusting the results to the range of -1 to 1, achieving non-linear control of the feature response. These two activation functions optimize the feature processing results from the perspectives of effectiveness representation and response control, respectively, avoiding the limitation of the feature response range caused by a single activation function, and providing more reasonable feature data for subsequent information fusion.

[0162] The dominant weight matrix assigns weights based on the influence of each element in the joint eigenvector on the multidimensional estimation vector, ensuring that eigenelements that contribute significantly to parameter estimation play a more prominent role. The dominant bias vector compensates for computational biases caused by differences in feature distributions, correcting the parameter estimation baseline. The gating weight matrix regulates the participation of each eigenele, strengthening effective features and suppressing redundant features; the gating bias vector fine-tunes the gating calculation results, making the gating output more closely match the parameter estimation requirements. This four-fold parameter synergy adapts to the characteristics of the joint eigenvector, avoiding estimation errors caused by feature weight imbalances or baseline biases, and improving the accuracy of parameter estimation.

[0163] By performing element-wise multiplication of the dominant feature result after Sigmoid activation and the gated feature result after hyperbolic tangent activation using the Hadamard product, the core information directly related to parameter estimation in the dominant feature is preserved, while invalid or interfering information is filtered out by the gated feature, achieving accurate fusion of the two types of feature information. This fusion method avoids information conflicts or redundancy caused by simple superposition, enabling the final multidimensional estimation vector to more accurately represent key synchronization parameters such as symbol timing error and carrier frequency deviation. This provides a reliable quantitative basis for subsequent multi-carrier signal synchronization adjustment and solves the problem that single feature processing cannot simultaneously achieve both accuracy and anti-interference capability.

[0164] S3. Based on the preliminary estimate of the symbol timing error in the multidimensional estimation vector, the digital baseband signal is time-domain interpolated and adjusted to obtain the preliminary aligned time-domain signal of the multi-carrier 4096QAM signal.

[0165] In this embodiment, based on the preliminary estimate of the symbol timing error in the multidimensional estimation vector, the digital baseband signal is time-domain interpolated and adjusted to obtain the preliminary aligned time-domain signal of the multi-carrier 4096QAM signal, including:

[0166] A preliminary estimate of the symbol timing error is extracted from the multidimensional estimation vector;

[0167] Based on the preliminary estimate of the symbol timing error and the sign, determine the direction and reference sampling point of the time-domain interpolation operation;

[0168] Based on the reference sampling points, the sampling sequence of the digital baseband signal is resampled to obtain interpolated sampling points and to discard the original sampling points.

[0169] The interpolated sampling points are recombined with the original sampling points after selection in chronological order to obtain the preliminary aligned time-domain signal of the multi-carrier 4096QAM signal.

[0170] Specifically, when extracting the preliminary estimate of the symbol timing error from the multidimensional estimation vector, the specific dimension corresponding to the symbol timing error estimation in the multidimensional estimation vector is identified. The value of this dimension directly reflects the degree to which the symbol timing deviates from the standard position during the transmission of the digital baseband signal. By reading the specific value of this dimension, the preliminary estimate of the symbol timing error is accurately obtained, ensuring that the value can truly represent the magnitude and direction of the timing deviation, providing a clear error basis for subsequent time-domain interpolation adjustments.

[0171] Furthermore, based on the preliminary estimate and sign of the symbol timing error, when determining the direction of the time-domain interpolation operation and the reference sampling point, the sign corresponding to the preliminary estimate of the symbol timing error is analyzed. If the sign is positive, it indicates that the actual sampling point lags behind the standard sampling position, and the interpolation operation direction needs to add sampling points forward; if the sign is negative, it indicates that the actual sampling point is ahead of the standard sampling position, and the interpolation operation direction needs to discard some sampling points backward. Using the sampling points in the digital baseband signal that meet the standard timing requirements as a reference, and combined with the magnitude of the timing error, the reference sampling point used for subsequent interpolation calculations is determined. This reference sampling point can provide a stable reference for the interpolation operation, ensuring that the interpolation direction and range accurately match the timing error correction requirements.

[0172] Furthermore, based on the reference sampling point, the sampling sequence of the digital baseband signal is resampled to obtain interpolated sampling points. When discarding original sampling points, new sampling points are added to the gaps in the original sampling sequence, centered on the reference sampling point, according to the magnitude of the timing error and the interpolation direction. During the addition process, based on the signal amplitude of the reference sampling point and its adjacent original sampling points, the signal amplitude at the gap position is calculated to generate interpolated sampling points, ensuring that the amplitude of the interpolated sampling points is consistent with the trend of the original signal. At the same time, based on the sign and magnitude of the timing error, sampling points in the original sampling sequence that exceed the standard timing range or are meaningless for signal alignment are discarded, while original sampling points that meet the adjustment requirements are retained, so that the remaining sampling points can work together with the interpolated sampling points to construct a regular sampling sequence.

[0173] Furthermore, when recombining the interpolated sampling points with the selected original sampling points in chronological order to obtain the preliminary aligned time-domain signal of the multi-carrier signal, the chronological order of the interpolated sampling points and the selected original sampling points is carefully considered to ensure that the arrangement of each sampling point strictly follows the time logic of signal transmission. According to the chronological order, the interpolated sampling points are inserted into the gaps between the corresponding original sampling points, so that all sampling points form a continuous, non-overlapping, and complete sampling sequence. During the recombination process, the signal amplitude of each sampling point remains unchanged to ensure that the original characteristics of the signal are not affected. Finally, a preliminary aligned time-domain signal with sampling point timing consistent with the standard position and regular time-domain characteristics is obtained.

[0174] In this embodiment, a preliminary estimate of the symbol timing error is parsed from the multidimensional estimation vector. This estimate can directly reflect the degree and direction of the symbol timing deviation from the standard position during the transmission of the digital baseband signal. It provides a clear and accurate basis for error quantification for subsequent time-domain interpolation adjustment, avoids blind adjustment due to fuzzy error information, and ensures that subsequent operations always revolve around correcting timing deviation.

[0175] Based on the preliminary estimate of the symbol timing error and the sign, it is determined whether the actual sampling point lags or leads the standard sampling position, thereby determining the direction of the time-domain interpolation operation. Simultaneously, a reference sampling point is selected based on the error magnitude. This dynamic parameter determination method enables the interpolation operation to accurately match the actual timing deviation, avoiding adjustment deviations caused by fixed interpolation strategies and ensuring that the interpolation direction and reference point meet signal alignment requirements.

[0176] The sampling sequence of the digital baseband signal is resampled based on a reference sampling point. During the gaps in the original sampling sequence, interpolated sampling points are generated based on the signal amplitude of the reference point and adjacent sampling points, ensuring that the interpolated points closely match the changing trend of the original signal. Simultaneously, original sampling points that exceed the standard timing range or are meaningless for signal alignment are discarded. This resampling process both supplements missing effective sampling information and eliminates redundant sampling points, optimizing the integrity and regularity of the sampling sequence.

[0177] The interpolated sampling points are recombined with the original, selected sampling points in chronological order to form a continuous, non-overlapping, and complete sampling sequence, resulting in a preliminarily aligned time-domain signal. The timing of the sampling points in this signal is nearly identical to the standard position, effectively correcting the symbol timing error. This lays a solid foundation for subsequent synchronization operations such as frequency offset pre-compensation and reduces the risk of signal distortion or synchronization failure due to timing deviations in later processing.

[0178] S4. Based on the preliminary estimate of the carrier frequency deviation in the multidimensional estimation vector, perform frequency offset pre-compensation on the pre-aligned time domain signal to obtain the frequency offset compensation signal of the multi-carrier 4096QAM signal.

[0179] In this embodiment, based on the preliminary estimate of the carrier frequency deviation in the multidimensional estimation vector, frequency offset pre-compensation is performed on the initially aligned time-domain signal to obtain the frequency offset compensation signal for the multi-carrier 4096QAM signal, including:

[0180] A preliminary estimate of the carrier frequency deviation is extracted from the multidimensional estimation vector;

[0181] Based on the preliminary estimate of the carrier frequency deviation and the sampling period, the angular velocity of the multi-carrier 4096QAM signal is determined.

[0182] The phase rotation sequence is obtained by multiplying the phase value of each element in the multi-carrier 4096QAM signal by the angular velocity.

[0183] In the time domain, the phase of the initially aligned time-domain signal is reversed based on the phase rotation sequence to obtain the frequency offset compensation signal of the multi-carrier 4096QAM signal.

[0184] Specifically, when extracting the preliminary estimate of the carrier frequency deviation from the multidimensional estimation vector, the specific dimension corresponding to the carrier frequency deviation estimation in the multidimensional estimation vector is identified. The value of this dimension directly reflects the degree of deviation between the actual carrier frequency of the multi-carrier signal and the standard carrier frequency. By reading the specific value of this dimension, the preliminary estimate of the carrier frequency deviation is accurately obtained, ensuring that this value can fully characterize the magnitude and direction of the frequency deviation, providing an accurate basis for subsequent frequency deviation pre-compensation.

[0185] Furthermore, when determining the angular velocity of the multi-carrier signal based on the preliminary estimate of the carrier frequency deviation and the sampling period, the sampling period is defined as a fixed time interval between two adjacent sampling points during signal sampling. This interval is a fundamental parameter for signal processing and remains constant. The preliminary estimate of the carrier frequency deviation is correlated with the sampling period, and through their interaction, it is transformed into a physical quantity characterizing the rate of phase change of the signal. This physical quantity is the angular velocity of the multi-carrier signal, whose magnitude and direction are directly related to the frequency deviation and can accurately reflect the phase change pattern caused by the frequency deviation.

[0186] Furthermore, when using the product of the phase value of each element in the multi-carrier signal and the angular velocity as the phase rotation sequence, the original phase value of each element in the multi-carrier signal is extracted one by one to ensure that the phase information of each element is complete and without deviation. Based on the extracted original phase value and the determined angular velocity, the phase value of each element is changed in a relationship proportional to the angular velocity, generating a rotation phase value corresponding to each element. The rotation phase values ​​of all elements are arranged sequentially according to the original order of the signal elements to form a continuous phase rotation sequence, which can accurately correspond to the phase shift caused by frequency offset.

[0187] Furthermore, when performing phase inverse rotation on the initially aligned time-domain signal based on the phase rotation sequence to obtain the frequency offset compensation signal for the multi-carrier signal, the phase rotation sequence is precisely aligned with the initially aligned time-domain signal in the time domain to ensure that each signal element corresponds to a unique rotated phase value in the phase rotation sequence. For each element in the initially aligned time-domain signal, a reverse phase rotation operation is performed according to its corresponding rotated phase value, which cancels the additional phase offset caused by the frequency offset, restoring the phase of each element to the normal state at the standard carrier frequency. After the phase inverse rotation processing, the frequency offset effect in the initially aligned time-domain signal is completely eliminated, ultimately obtaining a frequency offset compensation signal with a regular phase and a frequency consistent with the standard carrier frequency.

[0188] In this embodiment, a preliminary estimate of the carrier frequency deviation is parsed from the multidimensional estimation vector. This estimate can fully characterize the magnitude and direction of the deviation between the actual carrier frequency of the multi-carrier signal and the standard carrier frequency, providing an accurate basis for deviation quantification for subsequent frequency offset pre-compensation. This avoids errors in compensation direction or insufficient compensation due to inaccurate frequency offset information, ensuring that the compensation operation always revolves around the actual frequency offset problem.

[0189] Based on a fixed sampling period, it is correlated with a preliminary estimate of the carrier frequency deviation to calculate and transform it into angular velocity, which characterizes the rate of signal phase change. Angular velocity can accurately reflect the phase change pattern caused by frequency deviation, avoiding the error caused by roughly calculating the phase offset based solely on the frequency deviation value. This provides key parameter support for the subsequent generation of accurate phase rotation sequences and improves the accuracy of frequency deviation compensation.

[0190] The original phase value of each element in the multi-carrier signal is extracted one by one. The phase value of each element is then changed in a manner proportional to the angular velocity, generating a rotational phase value for each element. These rotational phase values ​​are then arranged in their original order to form a phase rotation sequence. This sequence accurately corresponds to the phase shift of each signal element caused by frequency offset, avoiding the problem of inadequate compensation for some elements due to uniform phase adjustment, and ensuring that the compensation operation covers all signal elements.

[0191] In the time domain, the phase rotation sequence is precisely aligned with the initially aligned time domain signal. For each signal element, a reverse phase rotation operation is performed based on the corresponding rotation phase value to cancel out the additional phase offset caused by frequency deviation, restoring the phase of each element to its normal state at the standard carrier frequency. After this processing, the frequency deviation effect in the initially aligned time domain signal is completely eliminated. The resulting frequency deviation compensation signal has a regular phase and a frequency consistent with the standard carrier frequency, providing a high-quality signal source for subsequent phase trajectory state prediction and signal synchronization, and reducing synchronization errors caused by residual frequency deviation.

[0192] S5. Perform state prediction on the phase trajectory of the frequency offset compensation signal. Based on the output value of the state prediction, apply a reverse phase rotation to the frequency offset compensation signal to obtain the synchronized signal of the multi-carrier 4096QAM signal.

[0193] In this embodiment, state prediction of the phase trajectory of the frequency offset compensation signal includes:

[0194] Extract a sequence of historical phase trajectories containing at least one symbol period length from the memory cells of the nonlinear phase tracker;

[0195] By capturing local fluctuation patterns and global change trends in historical phase trajectory sequences, a phase dynamic feature vector is obtained;

[0196] The optimal estimated state vector from the previous moment is concatenated with the phase dynamic feature vector to obtain the enhanced state vector of the multi-carrier 4096QAM signal.

[0197] The enhanced state vector is nonlinearly evolved under the influence of phase dynamic characteristics to obtain the predicted state vector of the multi-carrier 4096QAM signal.

[0198] The formula for calculating the predicted state vector is as follows:

[0199] ;

[0200] In the formula, for The predicted state vector at time t. This is a linear state transition function based on the classical state-space model. for The optimal estimated state vector at time t. As a time factor, It is a nonlinear mapping function. For phase dynamic eigenvectors, This is a historical phase trajectory sequence.

[0201] Specifically, when extracting a historical phase trajectory sequence containing at least one symbol period from the memory unit of the nonlinear phase tracker, it is clarified that the memory unit of the nonlinear phase tracker is specifically used to store phase trajectory data generated during the transmission of frequency offset compensation signals, and the data is recorded continuously in chronological order. According to preset extraction rules, phase trajectory data with a time span covering at least one symbol period is selected from the memory unit to ensure that the extracted historical phase trajectory sequence completely contains all information about phase changes within that time period, without omitting key phase fluctuation details, thus providing comprehensive historical data support for subsequent analysis of phase change patterns.

[0202] Furthermore, by capturing local fluctuation patterns and global trends in historical phase trajectory sequences to obtain a phase dynamic feature vector, the historical phase trajectory sequences are analyzed segment by segment to identify phase fluctuations occurring within short periods. The frequency, amplitude, and duration of these local fluctuations are summarized to form fluctuation patterns characterizing local features. Simultaneously, from the time dimension of the entire sequence, the overall trend of phase change is observed to determine whether it exhibits an upward, downward, or stable trend, extracting the global trend reflecting the overall characteristics. The relevant information of local fluctuation patterns and global trends is organized and integrated in a fixed format, transforming it into a phase dynamic feature vector that quantifies the dynamic characteristics of phase changes.

[0203] Furthermore, when concatenating the optimal estimated state vector from the previous moment with the phase dynamic feature vector to obtain the enhanced state vector of the multi-carrier signal, it is clear that the optimal estimated state vector from the previous moment is calculated based on all previous phase data and can accurately reflect the optimal estimated state of the phase at the previous moment. Ensuring that the dimensions of this vector match the phase dynamic feature vector and that there are no information conflicts, all feature elements of both are concatenated in a preset order. This ensures that the enhanced state vector contains both the optimal estimated phase information from the previous moment and incorporates the dynamic change characteristics of the current phase, comprehensively covering the core data of historical states and real-time dynamics without losing any type of key information.

[0204] Furthermore, when nonlinearly evolving the enhanced state vector under the influence of phase dynamic characteristics to obtain the predicted state vector of the multi-carrier signal, a fixed nonlinear evolution method is adopted. Based on the change law represented by the phase dynamic feature vector, each feature element in the enhanced state vector is dynamically adjusted. During the adjustment process, local fluctuation patterns and global change trends are fully integrated, allowing the feature elements to undergo nonlinear transformations according to the phase change law, uncovering the complex hidden correlations between features. Through the evolution process, the historical and real-time information contained in the enhanced state vector is transformed into predictive information for future phase states, ultimately forming a predicted state vector that accurately represents subsequent phase change states.

[0205] Specifically, The optimal estimated state vector at time t comes from the... The optimal estimation result of the phase trajectory at time moment is obtained by... All phase trajectory data prior to a given time are analyzed and calculated. Combined with the variation patterns of the error-corrected trajectory and error correction, a result is obtained that accurately reflects... The optimal estimate of the phase state at time step, presented in vector form, contains... Key features of the time phase.

[0206] Furthermore, the historical phase trajectory sequence is derived from the memory cell of the nonlinear phase tracker. Phase trajectory data containing at least one symbol period length is extracted from the memory cell. These data continuously record the phase changes during the transmission of the frequency offset compensation signal in chronological order, fully preserving the local fluctuations and global changes in the phase.

[0207] Furthermore, the linear state transition function is pre-defined based on the classical state-space model, which follows the fundamental law that the phase state changes linearly with time. The function works by... The optimal estimated state vector at time step 1 is linearly transformed to predict its natural transition over time. The state at any given moment reflects the linear evolution trend of the phase state.

[0208] Furthermore, the nonlinear mapping function is pre-defined based on the nonlinear change characteristics of the phase trajectory. It is used to nonlinearly transform the phase dynamic feature vector processed by the convolutional neural network, so that it can be consistent with the output of the linear state transition function in the feature space, which facilitates subsequent information fusion.

[0209] Furthermore, the convolutional neural network extracts features by performing multi-layer convolution operations on the historical phase trajectory sequence. First, the historical phase trajectory sequence is processed by sliding window, with each window corresponding to a local segment in the sequence. Feature values ​​are calculated through the interaction between the convolution kernel and the local segment to capture local phase fluctuation patterns. Then, pooling operations are used to simplify the features while retaining key information. After multi-layer processing, the extracted local features are integrated with global features, and finally, a phase dynamic feature vector that can characterize the dynamic characteristics of the historical phase trajectory sequence is output.

[0210] Furthermore, the formula signifies that by fusing the prediction results of linear state transitions with a nonlinear correction term based on historical phase trajectory characteristics, it generates a formula that accurately reflects... The predicted state vector of the phase state at time step. The calculation first involves... The optimal estimated state vector at time step (t) is input into the linear state transition function. The function transforms the vector according to the linear change law of the classical state-space model, resulting in a linear evolution-based state transition function. The predicted state value at time moment reflects the natural trend of phase state changes when there is no significant nonlinear disturbance.

[0211] In detail, the historical phase trajectory sequence is simultaneously input into a convolutional neural network. Through multi-layer convolution and pooling operations, local fluctuation patterns and global change trends in the sequence are extracted to generate a phase dynamic feature vector, which contains the nonlinear change characteristics of the phase trajectory.

[0212] Furthermore, the phase dynamic feature vector is input into a nonlinear mapping function, which performs a nonlinear transformation on it to match the output of the linear state transition function in terms of feature dimension and numerical range, thus obtaining a nonlinear correction term. This correction term is used to compensate for nonlinear factors not considered in the linear prediction.

[0213] Furthermore, the linear prediction value output by the linear state transition function is added to the nonlinear correction term output by the nonlinear mapping function to achieve information fusion of linear trends and nonlinear characteristics, ultimately yielding a result that comprehensively reflects... The predicted state vector of the phase state at any given time contains both the natural evolution trend of the phase and the dynamic features in the historical trajectory, thus improving the accuracy of the prediction.

[0214] In this embodiment, a historical phase trajectory sequence containing at least one symbol period length is extracted from the memory cell of the nonlinear phase tracker. This sequence continuously records the phase changes of the frequency offset compensation signal in chronological order, which can completely cover all the fluctuation details of the phase within one symbol period. This avoids incomplete analysis of phase change patterns caused by fragmented historical data, and provides complete and continuous data source support for subsequent accurate capture of phase dynamic features.

[0215] By analyzing historical phase trajectory sequences segment by segment, we can identify phase fluctuations within a short period to obtain local fluctuation patterns. Simultaneously, we observe the overall phase change trend over time to extract global trends. These two types of features are integrated to form a dynamic phase feature vector. This vector contains both local detailed phase features and overall change patterns, avoiding the limitations of single-dimensional feature representation. It more comprehensively and accurately reflects the dynamic characteristics of phase changes, providing high-quality feature data for subsequent state prediction.

[0216] The vector that accurately reflects the optimal phase estimate state at the previous moment is concatenated with a vector containing phase dynamic features to form an enhanced state vector. This vector simultaneously encompasses the historical optimal estimate state and real-time dynamic features, preserving the accurate results of previous phase analysis while incorporating the latest phase change trend information. This avoids insufficient prediction information caused by relying solely on a single state vector, providing richer and more comprehensive state data for subsequent nonlinear evolution and improving the reliability of the prediction basis.

[0217] Based on the changing patterns represented by the phase dynamic feature vector, each feature element in the enhanced state vector is dynamically adjusted. Through nonlinear evolution, the complex relationships hidden between features are uncovered, transforming historical states and real-time dynamic information into predictive information for future phase states. This evolutionary approach can adapt to potential nonlinear changes in the phase trajectory, avoiding the limitations of linear prediction in handling complex phase changes. The resulting predicted state vector accurately represents subsequent phase changes, providing a precise predictive basis for subsequent reverse phase rotation to achieve signal synchronization and reducing phase synchronization errors.

[0218] By using a linear state transition function based on the classical state-space model, for The optimal estimated state vector at time step is processed, and based on the fundamental law that the phase state changes linearly with time, its natural transition to... The state at any given moment. This linear prediction can capture the regular trend of phase changes, avoid prediction deviations caused by ignoring the fundamental laws of change, provide a stable basic framework for subsequent predictions, and ensure that the prediction results do not deviate from the basic logic of phase changes.

[0219] By processing historical phase trajectory sequences using a convolutional neural network, local fluctuation patterns are captured through a sliding window, and key information is preserved through pooling operations to extract a phase dynamic feature vector that characterizes the dynamic properties of the phase. This vector is then transformed using a nonlinear mapping function to ensure consistency between the vector and the linear prediction result in the feature space, forming a nonlinear correction term. This correction term compensates for phase nonlinear changes that linear prediction cannot cover, adapts to the complex fluctuations that may exist in the phase of multi-carrier signals, and avoids the limitations of single linear prediction in handling complex scenarios.

[0220] By adding the basic prediction result output by the linear state transition function to the correction term output by the nonlinear mapping function, a deep fusion of linear trends and nonlinear characteristics is achieved. This fusion method preserves the conventional laws of phase changes while incorporating dynamic details from historical trajectories, avoiding the one-sidedness of a single prediction model and resulting in a more comprehensive final prediction. The time-predicted state vector can comprehensively reflect the overall trend and local fluctuations of the phase, accurately characterize the subsequent phase change state, and provide a reliable basis for the phase calibration of the frequency offset compensation signal.

[0221] The entire calculation process uses a historical phase trajectory sequence as a crucial input. This sequence fully records the phase changes over at least one symbol period, containing rich dynamic phase information. By leveraging a convolutional neural network to fully extract effective features from the historical data, the nonlinear correction term is generated based on the true phase change patterns, avoiding unfounded subjective corrections. Furthermore, by combining the optimal estimated state vector from the previous moment, the prediction results are ensured to both maintain the accuracy of previous analyses and incorporate the dynamic characteristics of historical data, further enhancing the reliability and practicality of the predicted state vector.

[0222] S6. Equalize and demodulate the synchronized signal with 4096QAM to obtain the data bit stream of the multi-carrier 4096QAM signal.

[0223] In this embodiment, the synchronized signal is equalized and demodulated using 4096QAM to obtain a data bitstream of a multi-carrier 4096QAM signal.

[0224] Specifically, when equalizing the synchronized signal, the channel interference experienced by the signal during transmission is first analyzed to identify the signal distortion characteristics caused by interference such as channel attenuation and multipath propagation. A fixed equalizer structure is used, and the synchronized signal is input into the equalizer. The equalizer, based on preset interference compensation rules, performs reverse correction on the distorted components of the signal. For example, it enhances the amplitude reduction caused by channel attenuation and cancels inter-symbol interference caused by multipath propagation. By continuously adjusting the internal parameters of the equalizer, the amplitude and phase of the output signal are restored to the ideal transmission state, eliminating the influence of channel interference on the signal, thus obtaining the equalized signal.

[0225] Furthermore, when performing quadrature amplitude demodulation on the equalized signal, the equalized signal is first decomposed into in-phase and quadrature components. These two components correspond to the amplitude information of the signal in the horizontal and vertical directions in the orthogonal coordinate system, respectively. According to the constellation mapping rules of quadrature amplitude modulation, which predefine the correspondence between different amplitude combinations and digital symbols, the amplitude combinations of the decomposed in-phase and quadrature components are compared one by one with the standard amplitude combinations in the constellation diagram. The standard amplitude combination that best matches the current amplitude combination is found, and the digital symbol corresponding to this standard amplitude combination is determined. All the digital symbols corresponding to the equalized signals are arranged sequentially in chronological order to form a continuous sequence of digital symbols, resulting in the data bitstream of the multi-carrier signal.

[0226] In this embodiment, when equalizing the synchronized signal, the distortion characteristics caused by channel attenuation and multipath propagation during transmission are first analyzed. Then, the equalizer corrects the distortion components in reverse according to preset rules, such as enhancing the amplitude reduction caused by channel attenuation and canceling inter-symbol interference caused by multipath propagation. At the same time, the equalizer parameters are continuously adjusted to ensure that the amplitude and phase of the output signal are restored to the ideal state. This process completely eliminates the influence of channel interference on the signal, avoids interference residues that cause subsequent demodulation errors, and provides a clean signal source for high-quality demodulation.

[0227] When demodulating the equalized signal, it is first decomposed into mutually perpendicular in-phase and quadrature components. Then, according to the preset mapping rules of the quadrature amplitude modulation constellation diagram, the amplitude combination of the two components is compared one by one with the standard amplitude combination of the constellation diagram to determine the corresponding digital symbols and arrange them in chronological order to form a data bitstream. This demodulation method accurately matches the characteristics of 4096QAM high-order modulation signals, accurately restores the digital information carried by the signal, avoids data loss or bit errors caused by mismatch between demodulation method and modulation type, and finally outputs a valid data bitstream that meets the requirements, completing the entire process from reception processing to data extraction of multi-carrier 4096QAM signals.

[0228] Equalization provides a clean, interference-free signal for demodulation, preventing channel interference from affecting demodulation accuracy. The demodulation process performs precise symbol mapping based on the ideal equalized signal, ensuring reliable data extraction. These two processes work together to solve signal transmission distortion and achieve accurate decoding of high-order modulated signals, effectively reducing the bit error rate of the data bitstream, ensuring high-quality output data, meeting the high data transmission accuracy requirements of multi-carrier 4096QAM signals, and providing reliable support for subsequent data applications.

[0229] A multi-carrier 4096QAM signal coordination and synchronization system 100 is disclosed, which can be installed in electronic devices. For example... Figure 2 As shown, depending on the functions implemented, the system may include a signal processing module 101, a multidimensional estimation vector extraction module 102, a time-domain interpolation module 103, a frequency offset pre-compensation module 104, a signal synchronization module 105, and a signal modulation module 106. These modules can also be called units, which refer to a series of computer program segments that can be executed by the processor of an electronic device and can perform a fixed function, and are stored in the memory of the electronic device.

[0230] In this embodiment, the functions of each module / unit are as follows:

[0231] Signal processing module 101 is used to perform down-conversion and analog-to-digital conversion on the received multi-carrier 4096QAM signal to obtain the digital baseband signal of the multi-carrier 4096QAM signal;

[0232] The multidimensional estimation vector extraction module 102 is used to extract the time spectrum and pilot symbol sequence of the multi-carrier 4096QAM signal, and use the time spectrum and pilot symbol sequence as joint features for forward inference to obtain the multidimensional estimation vector of the multi-carrier 4096QAM signal;

[0233] The time-domain interpolation module 103 is used to perform time-domain interpolation adjustment on the digital baseband signal based on the preliminary estimate of the symbol timing error in the multidimensional estimation vector, so as to obtain the preliminary aligned time-domain signal of the multi-carrier 4096QAM signal.

[0234] The frequency offset pre-compensation module 104 is used to perform frequency offset pre-compensation on the initially aligned time domain signal based on the preliminary estimate of the carrier frequency offset in the multidimensional estimation vector, so as to obtain the frequency offset compensation signal of the multi-carrier 4096QAM signal.

[0235] The signal synchronization module 105 is used to predict the phase trajectory of the frequency offset compensation signal and apply a reverse phase rotation to the frequency offset compensation signal based on the output value of the state prediction to obtain the synchronized signal of the multi-carrier 4096QAM signal.

[0236] The signal modulation module 106 is used to equalize and demodulate the synchronized signal with 4096QAM to obtain the data bit stream of the multi-carrier 4096QAM signal.

[0237] The above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A method for coordinating and synchronizing multi-carrier 4096QAM signals, characterized in that, Includes the following steps: Step 1: Perform down-conversion and analog-to-digital conversion on the received multi-carrier 4096QAM signal to obtain the digital baseband signal of the multi-carrier 4096QAM signal; Step 2: Extract the time spectrum and pilot symbol sequence of the multi-carrier 4096QAM signal, and use the time spectrum and pilot symbol sequence as joint features for forward inference to obtain the multidimensional estimation vector of the multi-carrier 4096QAM signal; Step 3: Based on the preliminary estimate of the symbol timing error in the multidimensional estimation vector, perform time-domain interpolation adjustment on the digital baseband signal to obtain the preliminary aligned time-domain signal of the multi-carrier 4096QAM signal; Step 4: Based on the preliminary estimate of the carrier frequency deviation in the multidimensional estimation vector, perform frequency offset pre-compensation on the pre-aligned time-domain signal to obtain the frequency offset compensation signal of the multi-carrier 4096QAM signal. Step 5: Perform state prediction on the phase trajectory of the frequency offset compensation signal. Based on the output value of the state prediction, apply a reverse phase rotation to the frequency offset compensation signal to obtain the synchronized signal of the multi-carrier 4096QAM signal. Step 6: Equalize and demodulate the synchronized signal using 4096QAM to obtain the data bit stream of the multi-carrier 4096QAM signal.

2. The multi-carrier 4096QAM signal coordination and synchronization method as described in claim 1, characterized in that, The multi-carrier 4096QAM signal received in step 1 is a radio frequency signal. The specific method of step 1 is as follows: Step 101: The received radio frequency signal is mixed with the local oscillator signal generated by the local oscillator through a mixer, thereby downconverting the radio frequency signal to an intermediate frequency signal. Step 102: Use an anti-aliasing filter to filter the intermediate frequency signal, suppress out-of-band noise and image frequency interference, and obtain the filtered intermediate frequency signal; Step 103: Sample and quantize the filtered intermediate frequency signal to obtain a digital intermediate frequency signal; Step 104: Convert the digital intermediate frequency signal to the baseband frequency and separate the in-phase component and quadrature component to obtain the digital baseband signal of the multi-carrier 4096QAM signal.

3. The multi-carrier 4096QAM signal coordination and synchronization method as described in claim 1, characterized in that, In step 2, the time-spectrum diagram and pilot symbol sequence of the multi-carrier 4096QAM signal are extracted, specifically as follows: Step 201: Divide the digital baseband signal into continuous signal frames; Step 202: Perform a short-time Fourier transform on each signal frame to obtain the time-frequency unit corresponding to the signal frame; Step 203: Combine the time-frequency units in time order to form the time-spectrum diagram of the multi-carrier 4096QAM signal; Step 204: Based on the pilot pattern predefined in the communication protocol, locate and extract pilot symbols from the frequency domain resources of the digital baseband signal to obtain the pilot symbol sequence of the multi-carrier 4096QAM signal.

4. The multi-carrier 4096QAM signal coordination and synchronization method as described in claim 3, characterized in that, In step 2, the time-spectrum diagram and pilot symbol sequence are used as joint features for forward inference to obtain the multidimensional estimation vector of the multi-carrier 4096QAM signal. The specific method is as follows: Step 205: Perform channel normalization on the time-frequency spectrum and phase unwinding on the pilot symbol sequence to obtain standardized time-frequency features and normalized pilot features; Step 206: Concatenate the standardized time-frequency features with the normalized pilot features to obtain the joint feature vector of the multi-carrier 4096QAM signal; Step 207: Perform a nonlinear transformation on the multi-carrier 4096QAM signal based on the joint eigenvector to obtain the multidimensional estimation vector of the multi-carrier 4096QAM signal.

5. The multi-carrier 4096QAM signal coordination and synchronization method as described in claim 4, characterized in that, The specific method for step 207 is as follows: ； In the formula, For multidimensional estimation vectors, It is the Sigmoid activation function. The hyperbolic tangent activation function is used. This is the dominant weight matrix in the nonlinear transformation. For joint feature vectors, This is the dominant bias vector in the nonlinear transformation. This is the gate weight matrix in the nonlinear transformation. This is the gated bias vector in the nonlinear transformation. For Hadama accumulation.

6. The multi-carrier 4096QAM signal coordination and synchronization method as described in claim 1, characterized in that, The specific steps for step 3 are as follows: Step 301: Extract the preliminary estimate of the symbol timing error from the multidimensional estimation vector; Step 302: Based on the preliminary estimate of the symbol timing error and the symbol, determine the direction and reference sampling point of the time-domain interpolation operation; Step 303: Based on the reference sampling points, the sampling sequence of the digital baseband signal is resampled to obtain interpolated sampling points, and the original sampling points are discarded. Step 304: Recombine the interpolated sampling points with the original sampling points after selection in chronological order to obtain the preliminary aligned time domain signal of the multi-carrier 4096QAM signal.

7. The multi-carrier 4096QAM signal coordination and synchronization method as described in claim 1, characterized in that, The specific method for step 4 is as follows: Step 401: Extract a preliminary estimate of the carrier frequency deviation from the multidimensional estimation vector; Step 402: Determine the angular velocity of the multi-carrier 4096QAM signal based on the preliminary estimate of the carrier frequency deviation and the sampling period; Step 403: The phase value of each element in the multi-carrier 4096QAM signal is proportional to the product of the angular velocity and used as the phase rotation sequence; Step 404: In the time domain, the phase of the initially aligned time domain signal is reversed based on the phase rotation sequence to obtain the frequency offset compensation signal of the multi-carrier 4096QAM signal.

8. The multi-carrier 4096QAM signal coordination and synchronization method as described in claim 1, characterized in that, In step 5, the phase trajectory of the frequency offset compensation signal is predicted in the following way: Step 501: Extract a sequence of historical phase trajectories containing at least one symbol period length from the memory cell of the nonlinear phase tracker; Step 502: Capture the local fluctuation patterns and global change trends in the historical phase trajectory sequence to obtain a phase dynamic feature vector; Step 503: The optimal estimated state vector of the previous moment is concatenated with the phase dynamic feature vector to obtain the enhanced state vector of the multi-carrier 4096QAM signal. Step 504: Under the influence of phase dynamic characteristics, the enhanced state vector is nonlinearly evolved to obtain the predicted state vector of the multi-carrier 4096QAM signal.

9. The multi-carrier 4096QAM signal coordination and synchronization method as described in claim 8, characterized in that, The formula for calculating the predicted state vector is as follows: ； In the formula, for The predicted state vector at time t. This is a linear state transition function based on the classical state-space model. for The optimal estimated state vector at time t. As a time factor, It is a nonlinear mapping function. For phase dynamic eigenvectors, This is a historical phase trajectory sequence.

10. A multi-carrier 4096QAM signal coordination and synchronization system, characterized in that, include: The signal processing module is used to perform down-conversion and analog-to-digital conversion on the received multi-carrier 4096QAM signal to obtain the digital baseband signal of the multi-carrier 4096QAM signal; The multidimensional estimation vector extraction module is used to extract the time spectrum and pilot symbol sequence of the multi-carrier 4096QAM signal, and use the time spectrum and pilot symbol sequence as joint features for forward inference to obtain the multidimensional estimation vector of the multi-carrier 4096QAM signal; The time-domain interpolation module is used to perform time-domain interpolation adjustment on the digital baseband signal based on the preliminary estimate of the symbol timing error in the multidimensional estimation vector, so as to obtain the preliminary aligned time-domain signal of the multi-carrier 4096QAM signal. The frequency offset pre-compensation module is used to perform frequency offset pre-compensation on the initially aligned time domain signal based on the preliminary estimate of the carrier frequency offset in the multidimensional estimation vector, so as to obtain the frequency offset compensation signal of the multi-carrier 4096QAM signal. The signal synchronization module is used to predict the phase trajectory of the frequency offset compensation signal. Based on the output value of the state prediction, a reverse phase rotation is applied to the frequency offset compensation signal to obtain the synchronized signal of the multi-carrier 4096QAM signal. The signal conditioning module is used to equalize and demodulate the synchronized signal using 4096QAM to obtain a data bit stream of a multi-carrier 4096QAM signal.

Images