Data transmission methods for weak network environments
By constructing a channel characteristic surface and implementing channel state self-healing protection, the problem of channel state aging in high-speed mobile environments is solved, thereby improving the reliability and stability of data transmission and reducing the bit error rate without increasing pilot overhead.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- XINGYE (FUJIAN) INFORMATION TECHNOLOGY CO LTD
- Filing Date
- 2026-04-20
- Publication Date
- 2026-06-02
AI Technical Summary
In high-speed mobile environments, channel state aging leads to signal distortion and increased bit error rate. Existing technologies struggle to effectively eliminate the effects of channel state aging without increasing pilot overhead. Especially under high-reliability transmission requirements, traditional methods lack geometric constraints on the physical evolution of the channel response, resulting in predictive gradient divergence and increased phase noise.
By constructing channel feature surfaces and mapping multipath delay parameters, signal amplitude parameters, and discrete time axis coordinates to a spatiotemporal parametric coordinate system, the evolution gradient trend between channel feature surfaces is calculated. Under the channel state self-healing protection mechanism, external derivation and equalizer tap coefficient determination are performed to achieve closed-loop prediction and data compensation of channel state.
Without increasing pilot overhead, geometric topology constraints and self-healing protection mechanisms are used to lock the alignment of equalizer tap coefficients with the instantaneous channel state, effectively avoiding gradient divergence, improving the reliability and stability of data transmission, and reducing the bit error rate.
Smart Images
Figure CN122137708A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to a data transmission method for weak network environments, belonging to the field of wireless communication transmission technology. Background Technology
[0002] In current wireless communication transmission architectures, channel estimation and equalization techniques are core components for ensuring signal recovery quality at the receiver. The receiver typically obtains channel impulse response parameters based on pilot sequences and uses these parameters to compensate for distortions in data symbols affected by fading. In conventional communication scenarios, by periodically updating channel state information, the system can maintain a low bit error rate and ensure the smoothness of the transmission link. However, in high-speed mobile environments, especially when the receiver's speed exceeds 120 km / h, the propagation characteristics of electromagnetic waves in the three-dimensional physical scattering space undergo drastic changes. When the channel response parameters obtained by the receiver are applied to subsequent data symbols, they often experience substantial state aging due to processing delays, leading to spatial topological misalignment between the equalization parameters and the instantaneous channel characteristics. This prediction lag caused by the rapidly changing characteristics of the channel creates significant performance problems under the requirements of high-reliability transmission.
[0003] To address these challenges, traditional improvement methods primarily involve increasing pilot density or using algebraic linear difference for gradient fitting. However, increasing pilot overhead directly reduces the system's effective throughput. Furthermore, simple algebraic extrapolation logic lacks geometric constraints on the physical evolution of the channel response, making it prone to predictive gradient divergence when faced with nonlinear abrupt changes, thus introducing additional phase noise. This conflict between prediction accuracy and system overhead restricts the reliability of data transmission in weak network environments. While optimizing control logic alleviates signal distortion, existing techniques rely on discrete scalar statistical corrections, lacking the rigidity required for understanding the physical trajectory of channel evolution. For example, Chinese invention patent application CN121462011A discloses a method for suppressing signal interference in network communication chips. It suppresses interference by dynamically adjusting the bandwidth of the phase-locked loop by monitoring phase noise. Essentially, it is based on time-frequency domain closed-loop feedback adjustment. When rapid movement causes nonlinear topological deformation, the feedback parameter reconstruction mechanism lacks high-dimensional geometric modeling of the physical continuity of the electromagnetic environment. The deep fading point of the channel falls into gradient divergence due to the lack of boundary constraints, making it difficult to lock in the prediction bias at the source. The simple algebraic extrapolation logic lacks geometric constraints on the physical evolution process of the channel response, and nonlinear mutations introduce additional phase noise.
[0004] Therefore, how to construct a channel evolution inference architecture with physical continuity constraints, and eliminate the effects of channel state aging in high dynamic scenarios without increasing pilot overhead, has become the technical problem to be solved by this invention. Summary of the Invention
[0005] To address the problems mentioned in the background art, the technical solution of the present invention is as follows: A data transmission method for weak network environments, comprising the following steps: Step 101: Obtain the first channel impulse response of the wireless channel in the first symbol period and the second channel impulse response in the second symbol period; Step 102: Extract the multipath delay parameters, signal amplitude parameters, and discrete time axis coordinates from the first channel impulse response and the second channel impulse response. Map the multipath delay parameters, signal amplitude parameters, and discrete time axis coordinates to the spatiotemporal parametric coordinate system to construct the first channel feature surface and the second channel feature surface that characterize the electromagnetic propagation environment. Step 103: Determine the evolution gradient trend between the first channel feature surface and the second channel feature surface, and calculate the surface fluctuation variance of the first channel feature surface; Step 104: When the surface fluctuation variance is lower than the preset correlation distortion threshold, the second channel feature surface is extrapolated and derived along the evolution gradient trend to obtain the predicted channel feature surface of the predicted symbol period. Step 105: Sample the predicted channel feature surface to obtain the predicted channel state information, and determine the equalizer tap coefficients based on the predicted channel state information. Step 106: Use equalizer tap coefficients to compensate for distortion in the received data symbols.
[0006] Preferably, in step 103, the surface fluctuation variance is calculated. Follow the formula below: ,in, Let be the surface fluctuation variance of the first channel feature surface, and N be the total number of channel feature points collected on the surface. Let be the local gradient value of the i-th feature point. The average gradient value of all feature points; the correlation distortion threshold is dynamically matched based on the moving speed of the receiver and the Doppler frequency shift amplitude in the range of 0.5Hz to 500Hz.
[0007] Preferably, it also includes a channel state self-healing protection mechanism, including the following steps: Step 301, when the surface fluctuation variance reaches or exceeds the correlation distortion threshold, it is determined that the channel state in the spatiotemporal parametric coordinate system has undergone a non-stationary abrupt change; Step 302, stop the extrapolation derivation of the predicted channel feature surface, and project the second channel feature surface onto the reference tangent plane to reset the state, so as to block the gradient divergence of the prediction model under non-stationary conditions.
[0008] Preferably, it also includes combining logic for multi-path reception, including the following steps: step 401, extracting the predicted gain scalar of each receiving branch according to the predicted channel feature surface; step 402, identifying degraded branches whose predicted gain scalar is lower than a preset 3dB gain threshold; step 403, before the combiner combines the multipath signals, turning off the degraded branches and reallocating the maximum ratio combining weight according to the signal-to-noise ratio of the remaining normal branches.
[0009] Preferably, in step 102, constructing the first channel feature surface and the second channel feature surface involves using the multipath delay parameter as the depth dimension, the signal amplitude parameter as the energy dimension, and the discrete time axis coordinate as the time dimension, and converting the discrete channel tap coefficients into a continuous surface model through linear interpolation.
[0010] Preferably, in step 103, determining the evolution gradient trend involves calculating the centroid offset vector and the local tensor change rate of the first channel feature surface and the second channel feature surface in the spatiotemporal parametric coordinate system.
[0011] Preferably, in step 105, determining the equalizer tap coefficients involves using predicted channel state information to perform phase pre-compensation on the time-domain filter coefficients of the equalizer, so that the equalization response period is aligned with the channel fading period in time.
[0012] Preferably, in step 104, the weak network environment includes a fast fading environment caused by the receiver moving at a speed of 30km / h to 350km / h. By maintaining the geometric continuity constraint of the predicted channel characteristic surface, the convergence step size of the equalizer at the channel depth fading point is locked.
[0013] Preferably, in step 105, sampling the predicted channel feature surface involves extracting the surface height value at the coordinate point corresponding to the predicted symbol period and mapping the height value to a channel gain weight in the complex domain.
[0014] Preferably, the data transmission method for weak network environments is applied to high-reliability communication terminals with latency requirements of less than 1ms. By reusing existing pilot sequences in the physical layer for surface reconstruction, closed-loop prediction of channel state and data compensation can be completed within 10ms.
[0015] Compared with the prior art, the beneficial effects of the present invention are: 1. In data transmission in weak network environments, by mapping discrete channel impulse response parameters to control vertices in three-dimensional Euclidean space and constructing continuous geometric surfaces using a non-uniform rational B-spline algorithm, this invention transforms traditional algebraic scalar extrapolation into evolutionary derivation based on geometric topological constraints. Under this mechanism, channel evolution possesses rigid continuity based on physical space. By utilizing the principal curvature and Gaussian curvature tensors inherent in the surface model, second-order geometric continuity constraints are applied to signal fluctuations. This locks the gradient divergence path that traditional linear difference algorithms inevitably generate when encountering severe Doppler frequency shifts caused by high-speed movement at the physical principle level, ensuring the alignment of equalizer tap coefficients with the instantaneous channel state in the time domain.
[0016] 2. A topology tear monitoring mechanism based on local curvature variance of the surface is introduced. By identifying whether there are non-differentiable physical folds in the geometric surface, the essential distinction between sudden impulse noise and normal fast fading process is realized. When the curvature singularity, i.e. the topology tear threshold is touched, is detected, the system automatically suspends the surface extension process and performs tangent plane projection reset. This fuse protection mechanism is not a simple signal amplitude gating, but a real-time diagnosis of the integrity of the channel physical modeling structure. It effectively avoids the blind extrapolation of the prediction model under extreme non-stationary conditions and improves the self-healing capability and robustness of the baseband processing system in complex interference environments.
[0017] 3. By deeply coupling the temporal prediction mechanism with spatial diversity reception, an antenna branch combining strategy is formed. The system uses the predicted signal gain scalar obtained by resampling the prediction surface to identify and shield degraded branches that will enter the deep fading trap in the next symbol period before the multipath signal enters the combiner. This active combining weight adjustment based on the law of physical evolution effectively blocks the negative pollution of the maximum ratio combining result by pure noise branches. Without changing the physical layer detection overhead, the output signal-to-noise ratio of the receiver is purified, and the nonlinear superposition of the spatial and temporal dimension processing gains is realized. Attached Figure Description
[0018] Figure 1 This is a flowchart of the data transmission method for geometric surface continuity constraints according to the present invention; Figure 2 This is a system architecture diagram of channel prediction and self-healing compensation in a highly dynamic weak network environment according to the present invention.
[0019] The objectives, features, and advantages of this invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation
[0020] The technical solutions of the embodiments of this application will be clearly described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of this application are within the scope of protection of this application.
[0021] A data transmission method for weak network environments includes the following steps: Step 101: Obtain the first channel impulse response of the wireless channel in the first symbol period and the second channel impulse response in the second symbol period; Step 102: Extract the multipath delay parameters, signal amplitude parameters, and discrete time axis coordinates from the first channel impulse response and the second channel impulse response. Map the multipath delay parameters, signal amplitude parameters, and discrete time axis coordinates to the spatiotemporal parametric coordinate system to construct the first channel feature surface and the second channel feature surface that characterize the electromagnetic propagation environment. Step 103: Determine the evolution gradient trend between the first channel feature surface and the second channel feature surface, and calculate the surface fluctuation variance of the first channel feature surface; Step 104: When the surface fluctuation variance is lower than the preset correlation distortion threshold, the second channel feature surface is extrapolated and derived along the evolution gradient trend to obtain the predicted channel feature surface of the predicted symbol period. Step 105: Sample the predicted channel feature surface to obtain the predicted channel state information, and determine the equalizer tap coefficients based on the predicted channel state information. Step 106: Use equalizer tap coefficients to compensate for distortion in the received data symbols.
[0022] Preferably, in step 103, the surface fluctuation variance is calculated. Follow the formula below: ,in, Let be the surface fluctuation variance of the first channel feature surface, and N be the total number of channel feature points collected on the surface. Let be the local gradient value of the i-th feature point. The average gradient value of all feature points; the correlation distortion threshold is dynamically matched based on the moving speed of the receiver and the Doppler frequency shift amplitude in the range of 0.5Hz to 500Hz.
[0023] Preferably, it also includes a channel state self-healing protection mechanism, including the following steps: Step 301, when the surface fluctuation variance reaches or exceeds the correlation distortion threshold, it is determined that the channel state in the spatiotemporal parametric coordinate system has undergone a non-stationary abrupt change; Step 302, stop the extrapolation derivation of the predicted channel feature surface, and project the second channel feature surface onto the reference tangent plane to reset the state, so as to block the gradient divergence of the prediction model under non-stationary conditions.
[0024] Preferably, it also includes combining logic for multi-path reception, including the following steps: step 401, extracting the predicted gain scalar of each receiving branch according to the predicted channel feature surface; step 402, identifying degraded branches whose predicted gain scalar is lower than a preset 3dB gain threshold; step 403, before the combiner combines the multipath signals, turning off the degraded branches and reallocating the maximum ratio combining weight according to the signal-to-noise ratio of the remaining normal branches.
[0025] Preferably, in step 102, constructing the first channel feature surface and the second channel feature surface involves using the multipath delay parameter as the depth dimension, the signal amplitude parameter as the energy dimension, and the discrete time axis coordinate as the time dimension, and converting the discrete channel tap coefficients into a continuous surface model through linear interpolation.
[0026] Preferably, in step 103, determining the evolution gradient trend involves calculating the centroid offset vector and the local tensor change rate of the first channel feature surface and the second channel feature surface in the spatiotemporal parametric coordinate system.
[0027] Preferably, in step 105, determining the equalizer tap coefficients involves using predicted channel state information to perform phase pre-compensation on the time-domain filter coefficients of the equalizer, so that the equalization response period is aligned with the channel fading period in time.
[0028] Preferably, in step 104, the weak network environment includes a fast fading environment caused by the receiver moving at a speed of 30km / h to 350km / h. By maintaining the geometric continuity constraint of the predicted channel characteristic surface, the convergence step size of the equalizer at the channel depth fading point is locked.
[0029] Preferably, in step 105, sampling the predicted channel feature surface involves extracting the surface height value at the coordinate point corresponding to the predicted symbol period and mapping the height value to a channel gain weight in the complex domain.
[0030] Preferably, the data transmission method for weak network environments is applied to high-reliability communication terminals with latency requirements of less than 1ms. By reusing existing pilot sequences in the physical layer for surface reconstruction, closed-loop prediction of channel state and data compensation can be completed within 10ms.
[0031] Example 1: When the receiver travels through a complex multipath reflection environment at a speed of 350 km / h, the Doppler frequency shift causes channel state information aging. The three-dimensional geometric topological deformation of the electromagnetic wave in the physical scattering space leads to topological misalignment and equalization mismatch when the discrete channel state information extracted based on the periodic pilot is applied to adjacent data symbols. The receiver's baseband processor continuously acquires the first channel impulse response of the wireless channel in the first symbol period and the second channel impulse response in the second symbol period. The baseband processor extracts the multipath delay parameters, signal amplitude parameters, and discrete time axis coordinates from the first and second channel impulse responses. The baseband processor maps the multipath delay parameters, signal amplitude parameters, and discrete time axis coordinates to a spatiotemporal parametric coordinate system to construct a first channel feature surface and a second channel feature surface that characterize the electromagnetic propagation environment.
[0032] To address the nonlinear gradient change of channel distortion over time, the baseband processor determines the evolution gradient trend between the first and second channel characteristic surfaces and calculates the surface fluctuation variance of the first channel characteristic surface. ; Calculate the surface fluctuation variance Follow the formula below: ,in, Let be the surface fluctuation variance of the first channel feature surface, and N be the total number of channel feature points collected on the surface. Let be the local gradient value of the i-th feature point. The average gradient value for all feature points; simultaneously, the baseband processing system dynamically matches and determines the correlation distortion threshold based on the receiver's moving speed and the Doppler frequency shift amplitude within the range of 0.5Hz to 500Hz; when the surface fluctuation variance... When the correlation distortion threshold is lower than the preset threshold, the solver, constrained by maintaining the geometric continuity of the surface, extrapolates the second channel feature surface along the evolution gradient trend to obtain the predicted channel feature surface of the predicted symbol period. In this operation, the surface model generated by discrete parameter mapping provides a high-dimensional geometric boundary for topology evolution solution, and the non-divergent extrapolation of the continuous surface is dominated in the reverse direction according to the evolution gradient trend. The baseband processor samples the predicted channel feature surface to obtain the predicted channel state information. The baseband processor determines the equalizer tap coefficients based on the predicted channel state information and uses the equalizer tap coefficients to compensate for the distortion of the received data symbols, maintaining the convergence step size lock at the channel deep fading point.
[0033] When faced with sudden impulse noise causing the surface fluctuation variance to reach or exceed the correlation distortion threshold, the baseband processor triggers a channel state self-healing protection mechanism. The baseband processor suspends the extrapolation process of the predicted channel feature surface and projects the second channel feature surface onto a reference tangent plane. This tangent plane projection resets the state, preventing gradient divergence of the prediction model under non-stationary conditions. For communication terminals equipped with a multi-branch receiver architecture, the baseband processor extracts the predicted gain scalar for each receiver branch based on the predicted channel feature surface. The multipath signal combiner identifies the predicted gain scalar before signal combining. For degraded branches with a gain below 3dB, the degraded branch is shut down, and the maximum ratio combining weights are reallocated based on the signal-to-noise ratio of the remaining normal branches to prevent pure noise branches from contaminating the maximum ratio combining result. The baseband processor sends pulse control signals to the general-purpose input / output interface to drive the RF switch matrix to disconnect the feeder connection of the degraded branch on the physical link, or to force the corresponding branch multiplier weight register in the baseband combiner to be overwritten to zero. This shielding logic is executed by a hardware comparator array operating at a frequency of 200kHz, which checks the predicted gain scalar of each branch every 0.5ms. If, during one polling cycle, the gain scalar of a branch is below the 3dB threshold for four consecutive sampling periods, the determiner will send a 0 signal pulse to the corresponding 8-bit branch gating register to forcibly clear the multiplication weights of that branch participating in maximum ratio combining, ensuring that pure noise interference does not enter the back-end processing link. To address background noise drift caused by communication terminals traversing different terrain environments, the baseband processor is configured with a sliding state observation window of length M symbol periods. The system continuously calculates the average signal-to-noise ratio (SNR) after multipath signal combining within the sliding state observation window, detecting when the average SNR continuously exceeds a preset operating confidence level. When the upper limit of the interval reaches a specific time constant, the baseband processor extracts the mean variance of the surface fluctuation corresponding to the channel impulse response within the current sliding state observation window, and uses the mean to overwrite the original initial static reference constant C in the hardware register online, establishing an adaptive tracking baseline that conforms to the current electromagnetic topology environment. The above processing flow reuses the existing pilot sequence in the physical layer, completes the closed-loop prediction and distortion compensation of the channel state within 10ms, and locks the gradient divergence path of the linear algorithm based on the physical continuity constraint of the geometric surface, outputting demodulated data with a low bit error rate under the boundary condition of maintaining the predetermined pilot overhead.
[0034] Example 2: For data transmission reliability testing of communication terminals under strong multipath and high-frequency bias environments, a wireless channel hardware simulator and a baseband processing board are connected to generate a continuous test data stream. The simulator is configured with a tapped delay line multipath fading channel model following a standardized protocol, and the maximum Doppler frequency shift amplitude corresponding to a moving speed of 120km / h to 500km / h is set. 15dB of Gaussian white noise and periodic burst impulse noise are superimposed on the channel source to restore the broadband noise floor and transient change interference. The technical consideration for setting the correlation distortion threshold is to balance the timeliness of channel state tracking and the frequency of system state reset. When the Doppler frequency shift amplitude increases and the coherence time shortens, the baseband processor reduces the correlation distortion threshold to match the channel fading rate. Under the test condition of 350km / h, based on the local gradient mean distribution of the first channel characteristic surface within the coherence time block, the baseband processor determines the correlation distortion threshold to be 0.15.
[0035] Upon test initiation, the baseband processor acquires the distorted signal containing noise interference, extracts multipath delay parameters, signal amplitude parameters, and discrete time axis coordinates to construct a first channel feature surface and a second channel feature surface, and calculates the surface fluctuation variance of the first channel feature surface. Test data comparison shows that under continuous Doppler frequency shift interference, the control group, using a one-dimensional linear interpolation algorithm, experiences an exponential increase in equalizer tap coefficient error after 2ms, leading to system demodulation failure. The experimental group, employing the technical solution of this invention, extrapolates and derives the second channel feature surface along the evolution gradient trend, controlling the equalization error within 4.5%, exhibiting a physical constraint of high-dimensional geometric boundaries on the divergence path. The continuous data monitoring revealed the inflection point of nonlinear performance caused by impulse noise injection. When the surface fluctuation variance jumped to 0.16 due to sudden impulse noise, the self-healing protection mechanism of the missing channel state in the partial missing control group was missing. It was forced to derive along the original evolution gradient trend, which caused topological distortion of the predicted channel feature surface. The data demodulation bit error rate suddenly increased to 15.3%. When the variance value of the experimental group reached the limit of 0.15, the baseband processor suspended the extrapolation process of the predicted channel feature surface, projected the second channel feature surface onto the reference tangent plane, and re-determined the equalizer tap coefficients according to the reset state, blocking the error propagation. The bit error rate was maintained in the stable range of 2.1%.
[0036] Boundary parameter stress tests were conducted to construct an out-of-range control group that deviated from the defined values. Test data showed that when the correlation distortion threshold was set to an absolute lower limit of 0.05, conventional Gaussian white noise disturbances caused the system to frequently suspend the derivation process, and the baseband processor continuously degraded to the tangent plane projection state, resulting in a decrease in the prediction gain of the predicted channel feature surface and an increase in the overall bit error rate to 8.2%. When the threshold was set to an absolute upper limit of 0.35, the system delayed in recognizing the geometric manifold distortion caused by sudden impulse noise, resulting in a deterioration in the demodulation bit error rate to 12.7%. The parameter configuration of 0.15 constituted an effective working range for maintaining surface continuity and blocking abrupt divergence. Quantitative comparison of spatiotemporal coordinate mapping and geometric feature monitoring confirmed that extracting multipath delay parameters, signal amplitude parameters, and discrete time axis coordinates to construct the channel feature surface, and triggering a self-healing protection mechanism based on surface fluctuation variance, suppressed the gradient divergence caused by the superposition of high dynamic multipath fading environment and transient interference, and maintained the stability of the data transmission link.
[0037] Example 3: When a communication terminal accesses a wireless network with frequency offset characteristics, the baseband processor acquires the first channel impulse response of the wireless channel in the first symbol period and the second channel impulse response in the second symbol period. The baseband processor extracts the multipath delay parameters, signal amplitude parameters, and discrete time axis coordinates from the first and second channel impulse responses. The baseband processor establishes a three-dimensional spatiotemporal parametric coordinate system, setting the discrete time axis coordinates as the X-axis, the multipath delay parameters as the Y-axis, and the signal amplitude parameters as the Z-axis. The baseband processor converts the extracted multipath delay parameters, signal amplitude parameters, and discrete time axis coordinates into a three-dimensional spatial scatter set in the three-dimensional spatiotemporal parametric coordinate system, and uses a non-uniform rational B-spline algorithm to calculate the control vertex and node vectors of the three-dimensional spatial scatter set, constructing a first channel characteristic representing the electromagnetic propagation environment. Based on the multipath scattering theory, the overall energy envelope change rate and surface phase rotation rate of the signal have inherent differences in physical time scale. To ensure accurate mapping and reconstruction of the channel state in the complete complex domain, including amplitude and phase, the baseband processor converts the discrete channel impulse response into an independent amplitude scalar matrix and an unwound phase matrix when constructing the feature surface. The Z-axis coordinate of the three-dimensional spatiotemporal parametric coordinate system is specifically assigned by the amplitude scalar matrix, generating a continuous feature surface characterizing energy fading. The baseband processor extracts the partial derivatives of the unwound phase matrix on the discrete time axis and establishes a dynamic phase gradient table independent of the feature surface. After constructing the first channel feature surface, the baseband processor determines the current Doppler frequency shift amplitude by analyzing the phase deflection rate of the continuous pilot symbols and calculates the correlation distortion threshold by combining the receiver's moving speed output by the velocity measurement module. Correlation distortion threshold The calculation formula is: ,in, Here, is the correlation distortion threshold and is a dimensionless parameter; C is the initial static reference constant; and μ is the attenuation coefficient that is positively correlated with the receiver's moving speed. The Doppler frequency shift amplitude is defined as the range from 0.5 Hz to 500 Hz. When the Doppler frequency shift amplitude approaches 0.5 Hz, the correlation distortion threshold is defined as follows. Approaching the initial static reference constant; when the Doppler frequency shift amplitude approaches 500Hz, the attenuation coefficient reduces the correlation distortion threshold. It tends towards the minimum value.
[0038] Calculate the correlation distortion threshold Subsequently, the baseband processor calculates the surface fluctuation variance of the first channel characteristic surface; based on the differential geometric curvature energy evolution mechanism, the local gradient magnitude of the surface corresponds to the continuous distribution of the channel physical scattering path with a smooth change; the baseband processor calculates the orthogonal partial derivative vectors along the time delay axis and time axis for each discrete sampling point on the first channel characteristic surface, and extracts the L2 norm of the orthogonal partial derivative vector as the local gradient magnitude. Calculate the local gradient magnitude of all discrete sampling points. The sum of squares and mean of the differences between the surface variance and the arithmetic mean G is used to output a quantized scalar, namely the surface variance. After determining that the geometric continuity is achieved, the baseband processor extracts the control vertex matrix corresponding to the second channel feature surface. It then uses the centroid offset vector as a translation step size operator, superimposing it onto the coordinates of each element of the control vertex matrix to generate a new predictive control vertex matrix. Subsequently, it reconstructs the predictive channel feature surface based on the non-uniform rational B-spline basis function. The height values of the corresponding coordinate points on the predictive channel feature surface are then recombined with the extrapolated phase values in the aforementioned dynamic phase gradient table. The baseband processor then compares the surface fluctuation variance. Correlation distortion threshold When surface fluctuation variance Below the correlation distortion threshold At that time, the baseband processor determines the evolution gradient trend between the first channel feature surface and the second channel feature surface, and extrapolates the second channel feature surface along the evolution gradient trend to obtain the predicted channel feature surface of the predicted symbol period; the baseband processor samples the predicted channel feature surface to obtain the predicted channel state information; the baseband processor determines the equalizer tap coefficients based on the predicted channel state information, and uses the equalizer tap coefficients to compensate for distortion in the received data symbols; the above is achieved by using a correlation distortion threshold. The parameterized matching and geometric feature extrapolation derivation prevented gradient divergence caused by channel distortion, thus maintaining the stability of the data transmission link.
[0039] Example 4: When the communication terminal is deployed on a high-speed mobile physical architecture, the test vehicle travels at a constant speed at a basic reference speed. The baseband processor extracts the multipath delay parameters, signal amplitude parameters, and discrete time axis coordinates of the interference-free channel impulse response to construct a basic channel feature surface. The baseband processor extracts the mean upper limit of the surface fluctuation variance of the basic channel feature surface and assigns it to the initial static reference constant C. The test vehicle accelerates to its maximum operating speed, and the baseband processor records the limiting Doppler frequency shift amplitude and the minimum variance critical value for maintaining demodulation convergence. The baseband processor calculates the difference between the initial static reference constant C and the minimum variance critical value, divides the difference by the limiting Doppler frequency shift amplitude, generates the attenuation coefficient μ, and writes it into the hardware register.
[0040] After the attenuation coefficient μ is written to the hardware register, the communication terminal switches to the service network. During the symbol resolution cycle, the baseband processor reads the initial static reference constant C and the attenuation coefficient μ, and calculates the correlation distortion threshold by combining it with the real-time Doppler frequency shift amplitude. When the real-time Doppler frequency shift amplitude fluctuates, the baseband processor uses the initial static reference constant C and the attenuation coefficient μ to refresh the correlation distortion threshold. The correlation distortion threshold The surface fluctuation variance used to determine the first channel characteristic surface Constraints; Baseband processor comparison surface fluctuation variance Correlation distortion threshold The magnitude of the value triggers the extrapolation or tangent plane projection transformation of the second channel characteristic surface, and the receiver outputs demodulated data symbols in the multipath interference fading channel.
[0041] Example 5: When the communication terminal faces high-speed mobile conditions, the baseband processor receives a discretely distributed set of multipath points and converts the discrete signal into a continuous topology model. The baseband processor uses a chord length parameterization algorithm to calculate node vectors for the three-dimensional spatial set of scattered points, which includes multipath delay parameters, signal amplitude parameters, and discrete time axis coordinates. Based on the node vectors, the control vertex matrix is solved by least multiplication. The baseband processor generates non-uniform rational B-spline basis functions based on the node vectors and the control vertex matrix. It then uses the non-uniform rational B-spline basis functions to construct the first channel feature surface and the second channel feature surface, determining the numerical conversion method from physical sampling points to continuous geometric surfaces.
[0042] When sudden impulse noise causes surface fluctuation variance of the first channel characteristic surface Relevance distortion threshold reached At this time, the baseband processor starts the state reset step; the baseband processor extracts the geometric center point of the second channel feature surface in the stationary period, calculates the normal direction at the geometric center point as the center normal vector; the baseband processor determines the plane perpendicular to the center normal vector as the reference tangent plane, and orthogonally projects all control vertices of the second channel feature surface along the direction parallel to the center normal vector to the reference tangent plane to generate the reset topology matrix. In order to ensure the physical stability of the reference tangent plane, the reference position of the center normal vector is no longer determined by the surface at the moment of distortion, but is extracted from the geometric center coordinates of the first 10 normal symbol periods without alarms from the sliding window with a length of 10 built into the baseband processor and the arithmetic mean is calculated. The center point is used as the physical origin, and the Z-axis energy dimension coordinates of all abnormal control vertices are forcibly mapped to the horizontal tangent plane where the average center point is located. The residual accuracy of the projection calculation is controlled within the order of 0.0001 mm. This projection operation is used to block the surface topology distortion caused by sudden noise.
[0043] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.
[0044] Finally, it should be noted that 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 data transmission method for weak network environments, characterized in that, Includes the following steps: Step 101: Obtain the first channel impulse response of the wireless channel in the first symbol period and the second channel impulse response in the second symbol period; Step 102: Extract the multipath delay parameters, signal amplitude parameters, and discrete time axis coordinates from the first channel impulse response and the second channel impulse response. Map the multipath delay parameters, signal amplitude parameters, and discrete time axis coordinates to the spatiotemporal parametric coordinate system to construct the first channel feature surface and the second channel feature surface that characterize the electromagnetic propagation environment. Step 103: Determine the evolution gradient trend between the first channel feature surface and the second channel feature surface, and calculate the surface fluctuation variance of the first channel feature surface; Step 104: When the surface fluctuation variance is lower than the preset correlation distortion threshold, the second channel feature surface is extrapolated and derived along the evolution gradient trend to obtain the predicted channel feature surface of the predicted symbol period. Step 105: Sample the predicted channel feature surface to obtain the predicted channel state information, and determine the equalizer tap coefficients based on the predicted channel state information. Step 106: Use equalizer tap coefficients to compensate for distortion in the received data symbols.
2. The data transmission method for weak network environments according to claim 1, characterized in that, In step 103, the surface fluctuation variance is calculated. Follow the formula below: ,in, Let be the surface fluctuation variance of the first channel feature surface, and N be the total number of channel feature points collected on the surface. Let be the local gradient value of the i-th feature point. The average gradient value of all feature points; the correlation distortion threshold is dynamically matched based on the moving speed of the receiver and the Doppler frequency shift amplitude in the range of 0.5Hz to 500Hz.
3. The data transmission method for weak network environments according to claim 1, characterized in that, It also includes a channel state self-healing protection mechanism, including the following steps: Step 301, when the surface fluctuation variance reaches or exceeds the correlation distortion threshold, it is determined that the channel state in the spatiotemporal parametric coordinate system has undergone a non-stationary abrupt change; Step 302, stop the extrapolation derivation of the predicted channel feature surface, and project the second channel feature surface onto the reference tangent plane to reset the state, so as to block the gradient divergence of the prediction model under non-stationary conditions.
4. The data transmission method for weak network environments according to claim 1, characterized in that, It also includes combining logic for multi-path reception, including the following steps: Step 401, extract the predicted gain scalar of each receiving branch according to the predicted channel feature surface; Step 402, identify degraded branches whose predicted gain scalar is lower than the preset 3dB gain threshold; Step 403, before the combiner combines the multipath signals, turn off the degraded branches and reallocate the maximum ratio combining weight according to the signal-to-noise ratio of the remaining normal branches.
5. A data transmission method for weak network environments according to claim 1, characterized in that, In step 102, the first channel feature surface and the second channel feature surface are constructed. This involves using the multipath delay parameter as the depth dimension, the signal amplitude parameter as the energy dimension, and the discrete time axis coordinate as the time dimension. The discrete channel tap coefficients are converted into a continuous surface model through linear interpolation.
6. The data transmission method for weak network environments according to claim 1, characterized in that, In step 103, determining the evolution gradient trend involves calculating the centroid offset vector and the local tensor change rate of the first channel feature surface and the second channel feature surface in the spatiotemporal parametric coordinate system.
7. A data transmission method for weak network environments according to claim 1, characterized in that, In step 105, the equalizer tap coefficients are determined, which involves using predicted channel state information to perform phase pre-compensation on the time-domain filter coefficients of the equalizer so that the equalization response period is aligned with the channel fading period in time.
8. A data transmission method for weak network environments according to claim 1, characterized in that, In step 104, the weak network environment includes a fast fading environment caused by the receiver moving at a speed of 30km / h to 350km / h. By maintaining the geometric continuity constraint of the predicted channel characteristic surface, the convergence step size of the equalizer at the channel depth fading point is locked.
9. A data transmission method for weak network environments according to claim 1, characterized in that, In step 105, the predicted channel feature surface is sampled, which involves extracting the surface height value at the coordinate point corresponding to the predicted symbol period and mapping the height value to the channel gain weight in the complex domain.
10. A data transmission method for weak network environments according to claim 1, characterized in that, The data transmission method for weak network environments is applied to high-reliability communication terminals with latency requirements of less than 1ms. By reusing existing pilot sequences in the physical layer for surface reconstruction, closed-loop prediction of channel state and data compensation can be completed within 10ms.