Substation network slicing and interference suppression method and system

Abstract

The application provides a substation network slice division and interference suppression method and system, and relates to the technical field of network slices. The method comprises the following steps: analyzing terminal request data and network channel data to establish a planning function, solving the planning function to generate an initial network slice, establishing a dynamic evolution equation according to the initial network slice, deducing an interference transfer function between the initial network slices, alternately solving to obtain an optimal power spectral density function and a time-frequency resource block allocation rule, performing interference component separation and phase compensation on the interfered signals, quantifying the network resource state deviation degree in combination with quality evaluation processing, reconstructing the initial network slice to obtain a network slice; the execution efficiency, reliability and anti-interference performance of the substation network slice division are improved, and the communication service quality is optimized.

Description

Technical Field

[0001] This invention relates to a method and system for substation network slicing and interference suppression, belonging to the field of network slicing technology. Background Technology

[0002] As the power grid continues to develop towards intelligence and informatization, the business channels within substation systems have increasingly diverse requirements for business interfaces and access methods. Wired communication methods can no longer meet the ever-growing business demands. Therefore, wireless LAN access within the substation domain has become an inevitable requirement. Network slicing technology, as a key means of ensuring network service quality, can provide dedicated communication channels for different services through logically isolated virtual networks.

[0003] In existing technologies, wireless communication solutions based on Wi-Fi or WAPI are typically used to provide wireless communication for substations, enabling remote monitoring of the health status of equipment within the substation and surrounding micro-environment parameters. However, existing technologies have the following shortcomings: in substation network data transmission scenarios, network data slicing cannot dynamically adapt to changes in substation services, and interference suppression is insufficient, resulting in a failure to guarantee the quality of communication services. Summary of the Invention

[0004] The purpose of this invention is to provide a method and system for substation network slicing and interference suppression to solve the problems of poor communication service quality and insufficient efficiency in the prior art.

[0005] To solve the above-mentioned technical problems, the present invention is implemented using the following technical solution.

[0006] Substation network slicing and interference suppression methods include:

[0007] A planning function is established by collecting terminal request data and network channel data. The planning function is solved by constructing a branch tree structure to obtain the initial network slice.

[0008] Based on the connection relationship and channel state of the initial network slice, the interference transfer function of the initial network slice is derived, and the optimal power spectral density function and time-frequency resource block allocation rule are obtained by iteratively solving the interference transfer function.

[0009] The filtering parameters are set according to the optimal power spectral density function, and the interference components are separated on the initial network slice affected by the interference. The signal compensation is performed in combination with the time-frequency resource block allocation rules to obtain the initial network slice after interference suppression and its corresponding quality evaluation vector.

[0010] A dynamic equilibrium function is constructed using the quality evaluation vector as the independent variable. The deviation of the network resource state is quantified by the dynamic equilibrium function. The time-frequency resource block allocation rule is updated and the initial network slice is optimized to obtain the network slice.

[0011] Specifically, a planning function is established by collecting terminal request data and network channel data. This planning function is then solved using a branch tree structure to obtain the initial network slices, including:

[0012] By analyzing the service characteristics of terminal request data, it is mapped to bandwidth demand level and latency sensitivity coefficient, and channel gain matrix and interference power spectral density are constructed by combining channel state information in network channel data.

[0013] A planning function is constructed based on the channel gain matrix and the interference power spectral density, combined with a penalty term. The penalty term is determined by measuring the product of the transmit power of the interference link and the power coupling coefficient.

[0014] By initializing the branch tree with the root node resource allocation state of zero, the solution space of the planning function is recursively searched, child nodes are generated and their function upper bounds are calculated. When the cumulative interference of the link corresponding to the child node exceeds the preset tolerance threshold, the path expansion is terminated, and the solution that makes the planning function obtain the global minimum value is selected.

[0015] The demapping is converted into a logical slice structure, and the priority scheduling rules of the logical slice structure are set based on the latency sensitivity coefficient. The initial network slice is then instantiated and generated.

[0016] Specifically, based on the connectivity of the initial network slices and the channel state, the interference transfer function of the initial network slices is derived. The optimal power spectral density function and time-frequency resource block allocation rules are obtained by iteratively solving the interference transfer function, including:

[0017] Based on the topology connection parameters, state transition matrix, and channel response parameters of the initial network slice, a dynamic evolution equation is constructed, where the state transition matrix is ​​calculated from the channel gain matrix and topology connection parameters.

[0018] The rate of change of the current channel unit output signal is calculated based on the dynamic evolution equation and the preset perturbation variables. Based on the analytical expression of the ratio function of the rate of change to the perturbation variables in the frequency domain, the interference transfer function is constructed.

[0019] The aggregated interference power of different channel units is calculated based on the interference transfer function, and an objective function is established with the minimum communication rate and the maximum transmit power as the boundary.

[0020] By initializing the power spectral density function and the time-frequency resource block allocation rule, the objective function is solved iteratively with the power spectral density function and the time-frequency resource block allocation rule fixed respectively, until its difference norm is less than the preset difference value, and the optimal power spectral density function and time-frequency resource block allocation rule are output.

[0021] Specifically, filtering parameters are set according to the optimal power spectral density function, interference component separation is performed on the initial network slice affected by interference, and signal compensation is performed in conjunction with time-frequency resource block allocation rules to obtain the initial network slice after interference suppression and its corresponding quality evaluation vector, including:

[0022] Based on the amplitude-frequency characteristics of the optimal power spectral density function, identify the frequency points and bandwidths where interference power is concentrated, and set the center frequency and stopband attenuation coefficient based on the frequency points and bandwidth respectively.

[0023] A filter function is constructed based on the center frequency and the stopband attenuation coefficient. The interference signal in the initial network slice is input into the filter function for convolution operation to obtain the effective signal component.

[0024] The effective signal components are compared with the ideal reference signal in the time and frequency domains. The group delay and amplitude distortion generated by the effective signal components are analyzed and calculated, and a compensation factor is generated.

[0025] The time base deviation and phase rotation of the effective signal components are corrected according to the time-frequency resource block allocation rules, and the initial network slice after interference suppression is generated by the compensation factor and the effective signal components.

[0026] The initial network slices after interference suppression are segmented, and the signal-to-noise ratio, bit error rate, and time delay jitter parameters of each segment are calculated and normalized. The segments are then combined to generate a quality assessment vector.

[0027] Specifically, the effective signal components are compared with the ideal reference signal in the time-frequency domain, and the group delay and amplitude distortion generated by the effective signal components are analyzed and calculated to generate compensation factors, including:

[0028] The ideal reference signal is extracted based on the pilot position defined by the time-frequency resource block allocation rules, and time-frequency synchronization alignment is performed by calculating the peak value of the cross-correlation function between the effective signal component and the ideal reference signal.

[0029] Based on the aligned effective signal components and the ideal reference signal, the complex value of the channel transfer function on each subcarrier is calculated in the frequency domain. The complex value of the channel transfer function is used as matrix elements to establish the channel response matrix.

[0030] The group delay analytical expression is established by performing difference operations and polynomial fitting on the channel response matrix along the frequency dimension, and the standard deviation of the amplitude value is calculated along the time dimension to obtain the amplitude distortion.

[0031] Based on the analytical expression of group delay and amplitude distortion, the phase compensation and amplitude compensation of the time-frequency resource block are calculated, and the phase compensation and amplitude compensation are converted into compensation factors in complex form.

[0032] Specifically, a dynamic equilibrium function is constructed using the quality evaluation vector as the independent variable. This function quantifies the deviation of network resource states, updates the time-frequency resource block allocation rules, and optimizes the initial network slices, resulting in network slices, including:

[0033] The quality assessment vector is mapped to coordinate points in three-dimensional space. The extreme path between the coordinate points and the ideal coordinate points is calculated. The dynamic equilibrium function is constructed by combining the vector field divergence of the coordinate points in each gradient direction. The ideal coordinate points are set according to the network service quality requirements.

[0034] The partial derivatives of each component of the quality assessment vector are calculated using the dynamic equilibrium function to generate the gradient vector. The product of the gradient vector and the rate of change of each component of the quality assessment vector is calculated and integrated. The square root of the integral result is taken as the deviation of the network resource state.

[0035] Based on the magnitude and trend of network resource status deviation, a correction parameter generation function is constructed. The adjustment coefficient that obtains the global minimum value of the correction parameter generation function is found through preset conditions, and the time-frequency resource block allocation rules are adjusted.

[0036] The mapping relationship between time-frequency resource blocks and terminals is reconstructed based on the adjustment coefficients, and the interference coupling matrix of the initial network slice is calculated in combination with the transmit power.

[0037] When the maximum eigenvalue of the interference coupling matrix is ​​greater than the stability threshold, the tolerance threshold is updated by the network resource state deviation, and the network slice is obtained by solving the planning function. The stability threshold is determined by the network resource state deviation.

[0038] Specifically, the quality assessment vector is mapped to coordinate points in three-dimensional space, the extreme path between the coordinate points and the ideal coordinate points is calculated, and a dynamic equilibrium function is constructed by combining the vector field divergence of the coordinate points in each gradient direction. The ideal coordinate points are set according to the network service quality requirements, including:

[0039] The tangent, negative logarithm, and reciprocal square root values ​​of the components of the quality assessment vector are calculated, and these are used as basis vectors for orthogonalization to generate an orthogonal feature set.

[0040] Construct a metric tensor for three-dimensional space based on the orthogonal feature set, determine the coordinate points under the metric tensor, calculate the extreme path between the coordinate points and the ideal coordinate points, and take the length of the extreme path as the minimum path length.

[0041] Construct a mass state vector field in three-dimensional space, connect it to a metric tensor, calculate the covariant derivative of the mass state vector field, and obtain the divergence of the mass state vector field by finding the trace of the covariant derivative. The components of the mass state vector field are obtained from the time derivative of the orthogonal feature set.

[0042] A dynamic equilibrium function is constructed based on the minimum path length, the divergence of the quality state vector field, and the stability constraint term, wherein the stability constraint term is obtained based on the fluctuation amplitude of each component of the orthogonal feature set.

[0043] Specifically, based on the magnitude and trend of network resource status deviation, a correction parameter generation function is constructed. Adjustment coefficients that achieve the global minimum value of the correction parameter generation function are found using preset conditions. The time-frequency resource block allocation rules are then adjusted, including:

[0044] The first and second derivatives of the network resource state deviation are calculated, and a third-order eigenvector is established using the network resource state deviation, its first and second derivatives.

[0045] Using the third-order eigenvector as the independent variable, a modified parameter generation function containing an exponential decay term and a periodic oscillation term is constructed. The exponential decay term is determined by the magnitude of the network resource state deviation, and the periodic oscillation term is determined by the frequency of the network resource state deviation.

[0046] The feasible range of the adjustment coefficient is set according to the number of time-frequency resource blocks. The initial candidate solution set is generated by uniform sampling. The square root of the sum of squares of the differences between the initial candidate solution and each component of the third-order feature vector is calculated to obtain the function value set. The probability distribution function is established based on the function value set.

[0047] A candidate solution set is generated based on the probability distribution function. The function value set is iteratively calculated until the average function value of the function value set is less than the preset convergence value. The candidate solution with the smallest function value is taken as the global optimal adjustment coefficient.

[0048] The current time-frequency resource block allocation rule is updated and calculated based on the globally optimal adjustment coefficient to obtain the updated time-frequency resource block allocation rule.

[0049] The substation network slicing and interference suppression system includes a slice initialization module, an interference analysis module, a signal processing module, and a dynamic optimization module.

[0050] The slice initialization module is used to collect terminal request data and network channel data to establish a planning function, and solve the planning function by constructing a branch tree structure to obtain the initial network slice;

[0051] The interference analysis module is used to derive the interference transfer function of the initial network slice based on the connection relationship and channel state of the initial network slice, and iteratively solve the interference transfer function to obtain the optimal power spectral density function and time-frequency resource block allocation rules.

[0052] The signal processing module is used to set filtering parameters according to the optimal power spectral density function, perform interference component separation processing on the interfered initial network slice, and perform signal compensation in combination with the time-frequency resource block allocation rules to obtain the interference-suppressed initial network slice and its corresponding quality evaluation vector.

[0053] The dynamic optimization module is used to construct a dynamic equilibrium function with the quality evaluation vector as the independent variable, quantify the deviation of the network resource state through the dynamic equilibrium function, update the time-frequency resource block allocation rules and optimize the initial network slice to obtain the network slice.

[0054] Specifically, the signal processing module includes a filter control unit and a quality assessment unit:

[0055] The filter control unit is used to identify the interference frequency and bandwidth according to the optimal power spectral density function, and to calculate and set the filter parameters.

[0056] The quality assessment unit is used for segmented quality monitoring, calculating signal-to-noise ratio, bit error rate, and delay jitter parameters, performing numerical normalization processing, and generating a quality assessment vector.

[0057] Compared with existing technologies, the beneficial effects achieved by this invention are as follows: By solving the planning function and dynamically analyzing the interference transfer function through a branch tree structure, combined with the feedback mechanism of the quality assessment vector, dynamic interference suppression and adaptive resource allocation can be achieved. By employing alternating iterative optimization and adaptive filtering techniques, the power spectral density function and time-frequency resource block allocation rules can be accurately reconstructed. Through the quantification of network resource state deviation and dynamic balance function control, network slices are reconstructed. This invention solves the problems of poor communication service quality and insufficient efficiency in existing technologies, improves the execution efficiency, reliability, and anti-interference performance of substation network slice division, and optimizes communication service quality.

[0058] It should be understood that the above general description and the following detailed description are merely exemplary and explanatory, and are not intended to limit the technical solutions of this disclosure. Attached Figure Description

[0059] Figure 1 The flowchart of the substation network slicing and interference suppression method provided by the present invention;

[0060] Figure 2 This is an example diagram of a substation network provided by the present invention;

[0061] Figure 3 This is a schematic diagram of the computational quality assessment vector provided by the present invention;

[0062] Figure 4 The diagram shows the substation network slicing and interference suppression system structure provided by this invention. Detailed Implementation

[0063] The technical solution of the present invention will be described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the embodiments of the present invention and the specific features in the embodiments are detailed descriptions of the technical solution of the present invention, rather than limitations thereof. In the absence of conflict, the embodiments of the present invention and the technical features in the embodiments can be combined with each other.

[0064] The term "and / or" simply describes the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A alone, A and B simultaneously, or B alone. Additionally, the character " / " generally indicates that the preceding and following related objects have an "or" relationship.

[0065] Example 1:

[0066] Please see Figures 1-3 This invention provides an embodiment of a method for substation network slicing and interference suppression, which includes the following specific steps:

[0067] Step S1: Collect terminal request data and network channel data to establish a planning function. Solve the planning function by constructing a branch tree structure to obtain the initial network slice.

[0068] The specific steps of step S1 are as follows:

[0069] Step S101: By parsing the service characteristics of the terminal request data, map them to bandwidth demand level and latency sensitivity coefficient, and construct the channel gain matrix and interference power spectral density by combining the channel state information in the network channel data.

[0070] In this embodiment, by parsing the service type identifier and bandwidth requirement parameters in the terminal request data, the service criticality level and data throughput requirement are extracted. The service criticality level is mapped to a latency sensitivity coefficient. Based on the data throughput requirement and combined with the peak bandwidth data in historical transmission records, the service bandwidth requirement level is divided into high, medium, and low. Signal received power and phase noise samples in network channel data are collected. The cross-correlation function values ​​of the received signal and the reference signal at different sampling points are calculated to construct a channel gain matrix. Frequency domain analysis is performed on the phase noise samples. The interference power spectral density is generated by calculating the statistical expectation value of the noise power. The interference power spectral density is a description of existing interference sources in the wireless environment. The terminal request data includes, but is not limited to, bandwidth requirement parameters and service type identifiers. The network channel data includes, but is not limited to, channel state information, signal received power, and phase noise samples.

[0071] exist Figure 2In the diagram, the left rectangle represents the substation control room, which includes switches, controllers, and data networks. The right rectangle represents the substation switchyard, which includes equipment groups connected to the switches via wired connections. Data transmission between devices is wireless.

[0072] Step S102: Based on the channel gain matrix and the interference power spectral density, a planning function is constructed by combining a penalty term, which is determined by measuring the product of the transmit power of the interference link and the power coupling coefficient.

[0073] In this embodiment, the normalized power coupling coefficient is obtained by multiplying the squared magnitude of the off-diagonal elements in the channel gain matrix with the noise power and interference power spectral density of the corresponding receiving node. The actual transmit power of the interference link is measured, and the interference cost is obtained by multiplying it point by point with the power coupling coefficient. The product of the interference cost and the dynamic weighting factor is used as a penalty term. A planning function is constructed in combination with the spectral efficiency. The dynamic weighting factor is adaptively adjusted according to the ratio of the current network load rate to the historical interference level. The spectral efficiency is obtained by calculating the theoretical transmission rate of the interference link using the Shannon formula and summing the results.

[0074] Step S103: By initializing the branch tree with the root node resource allocation state of zero, recursively search the solution space of the planning function, generate child nodes and calculate their function upper bounds. When the cumulative interference of the link corresponding to the child node is greater than the preset tolerance threshold, terminate the path expansion and filter the solution that makes the planning function obtain the global minimum value.

[0075] In this embodiment, the root node is initialized with a branch tree structure with all zeros, the current optimal solution is set to infinity, and an empty feasible solution set is created. All available channel units are allocated to terminals without allocated channel units at each node, and corresponding child nodes are generated. The function value of the child node is calculated and added to the feasible solution set. The maximum throughput when the terminal is in the optimal channel and experiences minimal interference under ideal conditions is added to the current penalty term to obtain the upper bound of the node's function. The interference power generated by the child node on the current channel unit is added to the cumulative interference value of the parent node. The interference power is calculated by multiplying the terminal's transmit power by the corresponding element in the channel gain matrix. When the cumulative interference value of the current node is detected to be greater than a preset tolerance threshold, branch expansion is terminated. The solution that makes the planning function achieve a global minimum is selected from the feasible solution set. The tolerance threshold is set by those skilled in the art based on the actual situation.

[0076] Step S104: Demap the network into a logical slice structure, set the priority scheduling rules of the logical slice structure based on the latency sensitivity coefficient, and instantiate the initial network slice.

[0077] In this embodiment, a mapping table is generated by extracting the terminal and channel unit number corresponding to the solution. Terminals with the same channel unit number are divided into the same logical group according to the mapping table. A unique slice identifier is assigned to each logical group to construct a logical slice structure. The scheduling priority weight of each logical slice structure is calculated based on the latency sensitivity coefficient, where the scheduling priority weight is exponentially positively correlated with the latency sensitivity coefficient. A time-slice round-robin scheduling table is constructed based on the scheduling priority weight, where high-weight slices obtain more time-slice quotas in a unit period, and the interval between adjacent time slices is inversely proportional to the weight value. A time-frequency resource pool is configured for each logical slice structure. The logical slice structure, scheduling priority weight, and time-frequency resource pool configuration parameters are written into the instance attributes, and the instance is initialized with a network slice.

[0078] Step S2: Based on the connection relationship and channel state of the initial network slice, derive the interference transfer function of the initial network slice, and iteratively solve the interference transfer function to obtain the optimal power spectral density function and time-frequency resource block allocation rule.

[0079] The specific steps of step S2 are as follows:

[0080] Step S201: Construct a dynamic evolution equation based on the topology connection parameters, state transition matrix, and channel response parameters of the initial network slice, where the state transition matrix is ​​calculated from the channel gain matrix and topology connection parameters.

[0081] In this embodiment, a topology connection matrix is ​​constructed based on the relationship between terminal devices and channel units in the initial network slice. A value of 1 indicates the existence of a connection, and a value of 0 indicates no connection. The gain coefficients of each channel unit in the channel gain matrix are multiplied by the corresponding connection weights in the topology connection matrix and then normalized to obtain the state transition matrix. Channel response parameters of the channel units in continuous time slots are extracted, including amplitude attenuation factors and phase offsets, to construct the channel response vector. The state transition matrix and the channel response vector are then subjected to a tensor product operation to obtain the state evolution kernel. Based on the state transition matrix, noise term, state evolution kernel, and linear transformation term of the channel response vector, a dynamic evolution equation is established. The state evolution kernel is divided into symmetric and antisymmetric components. The symmetric component describes the stable evolution characteristics of the system state, while the antisymmetric component characterizes the irreversibility of state transitions. The noise term represents environmental uncertainty, which is estimated using maximum likelihood estimation.

[0082] Step S202: Calculate the rate of change of the current channel unit output signal based on the dynamic evolution equation and the preset disturbance variable, and construct the interference transfer function based on the analytical expression of the ratio function of the rate of change to the disturbance variable in the frequency domain.

[0083] In this embodiment, a preset perturbation variable is introduced into the dynamic evolution equation. The perturbation variable is represented as a small power increment superimposed on the transmit power. The rate of change of the current channel unit output signal relative to its steady-state value caused by the perturbation variable is calculated. The ratio of the rate of change to the preset perturbation variable is defined as the incremental response function representing the local sensitivity. The incremental response function is subjected to a Fourier transform to transform it from the time domain to the frequency domain, and its analytical expression as a function of frequency is obtained. Based on the analytical expression, an interference transfer function describing the interference propagation characteristics is constructed. The perturbation variable is set by those skilled in the art according to the actual situation, for example, it is set to 1% of the maximum transmit power of the channel unit.

[0084] Step S203: Calculate the aggregated interference power of different channel units based on the interference transfer function, and establish an objective function with the minimum communication rate and the maximum transmission power as the boundaries.

[0085] In this embodiment, the aggregated interference power is calculated based on the interference transfer function and the transmit power of the channel unit. The quotient of the signal power of each channel unit and the aggregated interference power is summed with the background noise power to obtain the signal-to-interference-plus-noise ratio (SINR) parameter. Based on the SINR parameter, the achievable communication rate of each channel unit is determined through a preset spectral efficiency mapping table. The spectral efficiency mapping table defines the modulation and coding schemes and spectral efficiency values ​​corresponding to different SINR intervals. The communication rate of the channel unit is set to be greater than the minimum communication rate requirement corresponding to its service, and the transmit power of the channel unit is set to be less than the maximum transmit power limit corresponding to its service. An objective function is established. The spectral efficiency mapping table is set by those skilled in the art through simulation experiments. The statistical average value of the interference power spectral density when there is no service transmission is determined as the background noise power.

[0086] Step S204: By initializing the power spectral density function and the time-frequency resource block allocation rule, the objective function is solved iteratively with the power spectral density function and the time-frequency resource block allocation rule fixed respectively, until its difference norm is less than the preset difference value, and the optimal power spectral density function and time-frequency resource block allocation rule are output.

[0087] In this embodiment, the power spectral density function is initialized with uniform distribution, ensuring its power value equals the ratio of the maximum transmit power to the total bandwidth. A polling allocation rule is used to initialize the time-frequency resource block allocation, sequentially allocating available time-frequency resource blocks. The allocation rule is then fixed. The non-convex constraint term in the objective function with respect to the power spectral density function is transformed using a logarithmic barrier function, constructing an augmented Lagrangian function. Its first-order partial derivative with respect to the power spectral density function is calculated and set to zero. A power control equation system containing interference constraints is established. The conjugate gradient method is used to iteratively solve the power control equation system, updating the power spectral density function. The power spectral density function is then fixed. The power spectral density function is used to establish a three-dimensional tensor optimization model based on time, frequency blocks, and channel units. By calculating the expected value of the signal-to-interference-plus-noise ratio (SINR), the time-frequency resource block allocation rules are updated. The root mean square error of the power spectral density function and the Hamming distance of the time-frequency resource block allocation rules are calculated as the difference norm during the iteration process. When the difference norm is less than a preset difference value, the iteration process is terminated, and the optimal power spectral density function and the optimal time-frequency resource block allocation rule are output. The power spectral density function is not a measurement of interference, but an optimal transmitter power control strategy used to avoid and resist interference. The difference value is set by those skilled in the art according to the actual situation.

[0088] Step S3: Set the filtering parameters according to the optimal power spectral density function, perform interference component separation processing on the interfered initial network slice, and perform signal compensation in combination with the time-frequency resource block allocation rules to obtain the interference-suppressed initial network slice and its corresponding quality evaluation vector.

[0089] The specific steps of step S3 are as follows:

[0090] Step S301: Based on the amplitude-frequency characteristics of the optimal power spectral density function, identify the frequency points and bandwidths where the interference power is concentrated, and set the center frequency and stopband attenuation coefficient based on the frequency points and bandwidth respectively.

[0091] In this embodiment, the optimal power spectral density function is scanned in the frequency domain to calculate its power amplitude at each frequency point. By detecting local maxima and comparing their differences with the global average power value, frequency points with power amplitudes greater than the average power value are identified and used as the frequency points where interference power is concentrated. The bandwidth on the frequency axis is searched through the frequency points where interference power is concentrated. The center frequency of the corresponding filter channel is set according to the frequency value of the frequency points where interference power is concentrated. The stopband attenuation coefficient is calculated based on the reciprocal of the bandwidth, where the bandwidth is inversely proportional to the stopband attenuation coefficient.

[0092] Step S302: Construct a filter function based on the center frequency and stopband attenuation coefficient, and input the interfered signal in the initial network slice into the filter function for convolution operation to obtain the effective signal component.

[0093] In this embodiment, the center frequency is converted into a normalized angular frequency and its product with pi is calculated to obtain the center angular frequency. The stopband attenuation coefficient is divided by a preset attenuation normalization constant to obtain a normalized attenuation value. The natural exponential function of the normalized attenuation value is multiplied by a reference order to obtain an order selection parameter. Based on the center angular frequency and the order selection parameter, a filter transfer function with symmetric coefficients is constructed, wherein the polynomial coefficients of the filter transfer function are obtained through the window function method. The filter transfer function is converted into a time-domain difference equation form. By solving the eigenvalues ​​of the coefficient matrix of the time-domain difference equation, an impulse response sequence is obtained. The impulse response sequence is normalized to obtain the final time-domain representation of the filter function. The interfered signal in the initial network slice is segmented according to the time-frequency resource block allocation rule and discretely convolved with the time-domain representation of the filter function. The discrete convolution result is used as the effective signal component. The attenuation normalization constant and the reference order are set by those skilled in the art according to the actual situation.

[0094] Step S303: Compare the effective signal component with the ideal reference signal in the time-frequency domain, analyze and calculate the group delay and amplitude distortion generated by the effective signal component, and generate a compensation factor.

[0095] The specific steps of step S303 are as follows:

[0096] Step S3031: Extract the ideal reference signal according to the pilot position defined by the time-frequency resource block allocation rule, and perform time-frequency synchronization alignment by calculating the peak value of the cross-correlation function between the effective signal component and the ideal reference signal.

[0097] In this embodiment, the time slot index and subcarrier index occupied by the pilot symbols are read from the time-frequency resource block allocation rules. Complex symbols at corresponding positions are extracted from the preset signal sequence to form an ideal reference signal sequence. Complex symbols at the same time-frequency position are extracted from the effective signal components to form a signal sequence to be synchronized. The signal sequence to be synchronized is multiplied by the ideal reference signal sequence and accumulated. All time delay offsets are traversed to obtain a cross-correlation value sequence. The maximum magnitude value in the cross-correlation value sequence is searched, and the corresponding time delay offset is determined as the symbol timing deviation estimate. The time-domain sampling sequence of the effective signal components is cyclically shifted according to the symbol timing deviation estimate to complete the time-domain synchronization. The phase angle of the cross-correlation value near the maximum magnitude value point is calculated. Linear regression is performed on the phase angle, and the slope of the regression line is divided by a fixed time interval to obtain the carrier frequency offset estimate. A complex rotation sequence is generated according to the carrier frequency offset estimate. The effective signal component sequence after time-domain synchronization is multiplied point by point with the complex rotation sequence to complete the carrier frequency synchronization. The signal sequence and time interval are set by those skilled in the art through simulation experiments.

[0098] Step S3032: Calculate the complex value of the channel transfer function on each subcarrier in the frequency domain based on the aligned effective signal components and the ideal reference signal, and use the complex value of the channel transfer function as matrix elements to establish the channel response matrix.

[0099] In this embodiment, the effective signal component sequence and the ideal reference signal sequence after time-frequency synchronization and alignment are subjected to the same number of discrete Fourier transform operations. The time-domain sampling points of each sequence are multiplied by a set of pre-calculated complex rotation factors and summed to obtain the complex spectrum values ​​of the effective signal component and the ideal reference signal on each subcarrier in the frequency domain. For each subcarrier index, the complex spectrum value of the effective signal component on that subcarrier is divided by the complex spectrum value of the ideal reference signal on the same subcarrier index to obtain a complex quotient. This complex quotient is used as the channel transfer function estimate of that subcarrier at the current transmission time. According to the time-frequency resource block allocation rules, the subcarrier index included in the current transmission time and the time sequence number of the current time in the complete transmission frame are determined. Each subcarrier index is used as the row coordinate of a two-dimensional grid, and the time sequence number of the current time is used as the column coordinate of the two-dimensional grid. The channel transfer function estimate of that subcarrier is used as the matrix element to construct the channel response matrix.

[0100] Step S3033: Perform differential operations and polynomial fitting on the channel response matrix along the frequency dimension to establish the analytical expression of the group delay, and calculate the standard deviation of the amplitude value along the time dimension to obtain the amplitude distortion.

[0101] In this embodiment, the phase angle of each complex element in the channel response matrix is ​​extracted to obtain the phase matrix. Phase unwrapping is performed on the columns of the phase matrix, and the phase difference between adjacent rows is compared sequentially. If the difference exceeds a preset threshold, a dynamic value is added or subtracted from the subsequent phase values. The dynamic value is determined by... Multiplying by an integer dynamically determined based on the number of phase transitions yields the unwound phase matrix. For each column of the unwound phase matrix, the first-order difference between adjacent phases is calculated along the row direction. The difference value is divided by a fixed frequency interval to obtain the instantaneous group delay value at each frequency point. This operation is repeated for all columns to obtain a set of group delay curves. A group delay curve is selected based on the highest signal-to-noise ratio or optimal time stability. Using the frequency point as the independent variable and the group delay value as the dependent variable, a polynomial function of a preset order is fitted using the least squares method to obtain an analytical expression for the group delay as a function of frequency. The amplitude values ​​of the complex elements in the channel response matrix are extracted to obtain the amplitude matrix. The standard deviation is calculated along the rows of the amplitude matrix to obtain the amplitude fluctuation at each frequency point. The arithmetic mean of the amplitude fluctuations at all frequency points is calculated to obtain the global amplitude distortion. The threshold, polynomial function, and frequency interval are set by those skilled in the art through simulation experiments.

[0102] Step S3034: Calculate the phase compensation amount and amplitude compensation amount of the time-frequency resource block according to the group delay analytical expression and amplitude distortion, and convert the phase compensation amount and amplitude compensation amount into complex form compensation factors.

[0103] In this embodiment, according to the time-frequency resource block allocation rules, the center frequency value corresponding to the time-frequency resource block is obtained. The center frequency value is substituted into the group delay analytical expression to calculate the theoretical group delay value of the time-frequency resource block. The difference between the theoretical group delay value and the fixed reference group delay value is multiplied by the center angular frequency of the signal of the time-frequency resource block to obtain the phase compensation amount required for the time-frequency resource block. The reciprocal of the global amplitude distortion is multiplied by a nominal amplitude value to obtain the amplitude compensation coefficient. Adjustments are made according to the specific amplitude fluctuation of each time-frequency resource block at the corresponding position in the amplitude matrix to determine the amplitude compensation amount required for the time-frequency resource block. The amplitude compensation amount required for the time-frequency resource block is taken as the modulus, and the required phase compensation amount is taken as the phase angle, and converted into a complex form compensation factor using Euler's formula.

[0104] Step S304: Correct the time base deviation and phase rotation of the effective signal components according to the time-frequency resource block allocation rules, and generate the initial network slice after interference suppression through the compensation factor and the effective signal components.

[0105] In this embodiment, the time base deviation and phase rotation of the effective signal component are calculated according to the time-frequency resource block allocation rules. The time base deviation is obtained by detecting the difference between the peak position of the autocorrelation function of the effective signal component and the ideal timing position. The phase rotation is obtained by linear fitting of the effective signal component in the frequency domain. The sampling point position of the effective signal component is remapped using the time base deviation to eliminate inter-symbol interference. A phase compensation matrix is ​​constructed using the phase rotation to perform frequency domain phase correction on the effective signal component. The real and imaginary parts of the compensation factor are decomposed into amplitude adjustment coefficients and phase compensation. The amplitude adjustment coefficients are obtained by taking the modulus of the compensation factor, and the phase compensation is obtained by taking the argument of the compensation factor. The effective signal component is multiplied by the amplitude adjustment coefficient and then rotated with the phase compensation. According to the time window and frequency unit mapping relationship defined by the time-frequency resource block allocation rules, the processed effective signal component is reassembled into a data frame structure of the initial network slice transmission format to obtain the initial network slice after interference suppression.

[0106] Step S305: The initial network slice after interference suppression is segmented, the signal-to-noise ratio, bit error rate and time delay jitter parameters of each segment are calculated and normalized, and combined to generate a quality evaluation vector.

[0107] In this embodiment, the initial network slice after interference suppression is segmented according to the time window determined by the time-frequency resource block allocation rules. The signal-to-noise ratio (SNR) parameter is obtained by calculating the ratio of the mean square of the signal amplitude within a segment to the noise power and multiplying by 10 using the common logarithm. The bit error rate (BER) parameter is obtained by analyzing the Hamming distance between the preset pilot sequence and the received sequence within the segment and using the quotient of the number of erroneous bits to the total number of transmitted bits. The timestamp sequence of consecutive data packets within the segment is extracted, and the variance of the difference between adjacent timestamps is calculated to obtain the delay jitter parameter. The SNR is mapped to the [0, 1] interval; a higher SNR value indicates better quality. The BER is then mapped to the [0, 1] interval after taking the negative logarithm; a lower BER indicates a mapping result closer to 1 and better quality. The delay jitter parameter is then mapped to the [0, 1] interval. 1] After the interval, calculate the difference between 1 and it. The larger the difference, the better the quality. Combine the signal-to-noise ratio, bit error rate and delay jitter parameters to generate the corresponding segment quality evaluation vector. The pilot sequence and the received sequence are set by those skilled in the art through simulation experiments.

[0108] exist Figure 3 In this process, the initial network slice after interference suppression is taken as input and cut into multiple continuous segments. The signal-to-noise ratio (SNR) calculation, bit error rate (BER) calculation, and delay jitter analysis are performed in parallel to obtain the corresponding SNR subvector, BER subvector, and delay jitter parameter subvector. These three subvectors are combined and normalized to generate a quality assessment vector.

[0109] Step S4: Construct a dynamic equilibrium function using the quality evaluation vector as the independent variable. Quantify the deviation of the network resource state through the dynamic equilibrium function, update the time-frequency resource block allocation rules, and optimize the initial network slice to obtain the network slice.

[0110] The specific steps of step S4 are as follows:

[0111] Step S401: Map the quality assessment vector to coordinate points in three-dimensional space, calculate the extreme path between the coordinate points and the ideal coordinate points, and construct a dynamic equilibrium function by combining the vector field divergence of the coordinate points in each gradient direction. The ideal coordinate points are set according to the network service quality requirements.

[0112] The specific steps of step S401 are as follows:

[0113] Step S4011: Calculate the tangent function value, negative logarithm value, and reciprocal square root value of the components of the quality assessment vector, and orthogonalize them as basis vectors to generate an orthogonal feature set.

[0114] In this embodiment, the signal-to-noise ratio (SNR) value in the quality assessment vector is calculated using a tangent transform to compress extreme numerical fluctuations, yielding the tangent function value. The bit error rate (BER) value in the quality assessment vector is transformed using a negative natural logarithm transform to amplify its small variation characteristics, yielding a negative logarithmic value. The delay jitter parameter value in the quality assessment vector is transformed using a reciprocal square root transform to enhance its sensitivity to network state changes, yielding a reciprocal square root value. The tangent function value, negative logarithmic value, and reciprocal square root value are used as the first, second, and third basis vectors, respectively. The dot product of the first and second basis vectors is multiplied by the modulus of the first basis vector. The ratio of the squares of the lengths is used as the projection coefficient. The product of the projection coefficient and the first basis vector is subtracted from the second basis vector to obtain the first orthogonal vector, which is orthogonal to the first basis vector. The dot product of the third basis vector with the first and second basis vectors is taken, and the ratio of the squares of the lengths of the first and second basis vectors is used as the projection coefficient set. The third basis vector is subtracted from the product of the projection coefficient set and the corresponding basis vector in turn to obtain the second and third orthogonal vectors. The first, second, and third orthogonal vectors are normalized to their lengths to transform them into a unit orthogonal basis, which constitutes the orthogonal feature set.

[0115] Step S4012: Construct a metric tensor for three-dimensional space based on the orthogonal feature set, determine the coordinate points under the metric tensor, calculate the extreme path between the coordinate points and the ideal coordinate points, and take the length of the extreme path as the minimum path length.

[0116] In this embodiment, the inverse matrix of the covariance matrix of each feature dimension is calculated based on the orthogonal feature set, and it is used as the metric tensor in three-dimensional space. The orthogonal eigenvalues ​​of the orthogonal feature set are mapped to coordinate points. Ideal coordinate points are set according to network service quality requirements. A system of differential equations is established based on the second and first derivatives of the coordinate points. The system of differential equations is solved iteratively. In each iteration step, the derivative value of the coordinate point is calculated based on the current coordinate point position and tangent vector value. The derivative values ​​are weighted and averaged to obtain the update amount of the coordinate position and tangent vector. The coordinate position and tangent vector of the next iteration are generated based on the update amount until the coordinate point path trajectory converges to the neighborhood of the ideal coordinate point. The integral length of the extreme path under the metric tensor is obtained and used as the minimum path length. The ideal coordinate point is obtained by the optimal threshold of each performance dimension through the same orthogonalization process.

[0117] Step S4013: Construct a mass state vector field based on the three-dimensional space, connect it to the metric tensor, calculate the covariant derivative of the mass state vector field, and obtain the divergence of the mass state vector field by finding the trace of the covariant derivative. The components of the mass state vector field are obtained from the time derivative of the orthogonal feature set.

[0118] In this embodiment, based on the numerical changes of the orthogonal feature set within a preset continuous sampling period, its first-order difference in the time dimension is calculated. The difference between each feature component of the orthogonal feature set in adjacent time windows is divided by the sampling time interval to obtain the time derivative sequence of the orthogonal feature set. Using the coordinate points as reference positions, a continuous mass state vector field is constructed by combining the time derivative sequence. Each component of the metric tensor is condensed with the corresponding directional derivative of the mass state vector field to establish the connection between the metric tensor and the mass state vector field. The ordinary derivative of the mass state vector field is corrected using the correction parameters calculated by the metric tensor. The correction parameters are obtained by the partial derivatives of the metric tensor through index rise and fall and linear combination. The covariant derivative of the mass state vector field is obtained by summing the ordinary derivative of the mass state vector field component with the product of the correction parameters and the mass state vector field component. The diagonal components of the covariant derivative are summed to obtain the divergence of the mass state vector field. The continuous sampling period is set by those skilled in the art through simulation experiments.

[0119] Step S4014: Construct a dynamic equilibrium function based on the minimum path length, the divergence of the mass state vector field, and the stability constraint term. The stability constraint term is obtained based on the fluctuation amplitude of each component of the orthogonal feature set.

[0120] In this embodiment, the root mean square value of the difference between each component of the orthogonal feature set in adjacent time windows is calculated as the corresponding fluctuation amplitude index. The geometric mean of the fluctuation amplitude index of each component is used as the fluctuation coefficient. The ratio of the fluctuation coefficient to the preset benchmark threshold is used as the basic weight of the stability constraint term. According to the sign characteristics of the mass state vector field divergence, a diffusion suppression term is set when the mass state vector field divergence is detected to be positive. The diffusion suppression term is determined by the product of the tangent transformation function value of the mass state vector field divergence and the time constant. A dynamic balance function is constructed by the stability constraint term, the minimum path length, and the diffusion suppression term. The benchmark threshold is set by those skilled in the art according to the actual situation, and the time constant is set based on the dynamic response characteristics of the terminal.

[0121] Step S402: Calculate the partial derivatives of each component of the quality assessment vector using the dynamic equilibrium function to generate the gradient vector. Calculate the product of the gradient vector and the rate of change of each component of the quality assessment vector and integrate it. Use the square root of the integration result as the deviation of the network resource state.

[0122] In this embodiment, the first-order partial derivatives of the components in the quality assessment vector are calculated based on the dynamic balance function. The calculation of the partial derivatives for the signal-to-noise ratio component needs to consider its exponential coefficient in the dynamic balance function; the calculation of the partial derivatives for the bit error rate component needs to consider its logarithmic characteristic in the dynamic balance function; and the calculation of the partial derivatives for the delay jitter parameter component needs to process its reciprocal term structure in the dynamic balance function. These components are sequentially combined to obtain the gradient vector. The quality assessment vectors for the current time and the previous two sampling periods are collected, and their changes within a unit sampling period are calculated to generate a rate of change vector. The gradient vector and the rate of change vector are then multiplied to obtain a scalar sequence. The scalar sequence is numerically integrated within a preset integration time window, and the square root of the integration result is taken to obtain the deviation of the network resource state. The integration time window is set by those skilled in the art through simulation experiments.

[0123] Step S403: Based on the magnitude and trend of the deviation of network resource status, construct a correction parameter generation function, find the adjustment coefficient that obtains the global minimum value of the correction parameter generation function through preset conditions, and adjust the time-frequency resource block allocation rules.

[0124] The specific steps of step S403 are as follows:

[0125] Step S4031: Calculate the first and second derivatives of the network resource state deviation, and establish a third-order feature vector using the network resource state deviation, its first and second derivatives.

[0126] In this embodiment, a time-domain analysis is performed on the network resource state deviation sequence. The first derivative is obtained by calculating the difference between the current deviation value and the deviation value of the previous period. The second derivative is obtained by performing a second difference operation between the current first derivative and the first derivative of the previous period. The network resource state deviation, the first derivative value, and the second derivative value at the current moment are combined in sequence to construct a third-order feature vector.

[0127] Step S4032: Using the third-order eigenvector as the independent variable, construct a modified parameter generation function containing an exponential decay term and a periodic oscillation term, wherein the exponential decay term is determined by the magnitude of the network resource state deviation, and the periodic oscillation term is determined by the frequency of the network resource state deviation change.

[0128] In this embodiment, the network resource state deviation component extracted from the third-order feature vector is used as the amplitude benchmark. The ratio of its square value to the unit constant is calculated, and the inverse of the natural exponent is taken to construct an exponential decay term. The arctangent function value of the product of the first and second derivative values ​​in the third-order feature vector is used as the phase angle input. The sine function value of the phase angle is calculated and multiplied by a preset oscillation amplitude coefficient to construct a periodic oscillation term. The sum of the exponential decay term and the periodic oscillation term is multiplied by a normalization coefficient determined by the magnitude of each component of the third-order feature vector. Combined with the basic bias, a correction parameter generation function is established. The basic bias is the smallest constant value that ensures the value of the correction parameter generation function is non-negative. The normalization coefficient is obtained by calculating the inverse of the square root of the sum of the squares of each component of the third-order feature vector.

[0129] Step S4033: Set the feasible range of the adjustment coefficient according to the number of time-frequency resource blocks, perform uniform sampling to generate an initial candidate solution set, calculate the square root of the sum of squares of the differences between the initial candidate solution and each component of the third-order feature vector to obtain the function value set, and establish a probability distribution function based on the function value set.

[0130] In this embodiment, the numerical boundary of the adjustment coefficient is determined based on the total number of time-frequency resource blocks. The range of the adjustment coefficient is set between zero and the reciprocal of the total number of time-frequency resource blocks. Within this range, values ​​are taken at equal intervals to generate an initial candidate solution set. Element-wise difference operations are performed on the corresponding components of each initial candidate solution and the third-order feature vector. The resulting differences are squared and summed. The square root of the sum is calculated to obtain the function value corresponding to the initial candidate solution, generating a function value set. A selection probability distribution is constructed based on the function value set. The function values ​​in the set are subjected to a negative exponential transformation. The transformed values ​​are normalized so that their sum is a unit value, obtaining the selection probability corresponding to each initial candidate solution. A discrete probability distribution function is constructed based on the selection probability value.

[0131] Step S4034: Generate a candidate solution set based on the probability distribution function, perform iterative calculation of the function value set until the average function value of the function value set is less than the preset convergence value, and take the candidate solution with the smallest function value as the global optimal adjustment coefficient.

[0132] In this embodiment, based on the mean vector and covariance matrix parameters of the probability distribution model, a candidate solution set of the same size as the initial candidate solution set is generated through multidimensional Gaussian sampling. For each candidate solution in the candidate solution set, the sum of squares of the differences between it and each component of the third-order eigenvector is calculated and the square root is taken to obtain a first function value set. The first function value set is merged with the function value set, and N candidate solutions with the smallest function values ​​are selected to form the first solution set, where N represents the number of candidate solutions selected. The mean vector and covariance matrix parameters of the probability distribution function are calculated based on the first solution set, where the mean vector is determined by the average value of the first solution set, and the covariance matrix is ​​determined by the product of the deviation matrix of the first solution set and the mean vector and its transpose. The calculation process of the first solution set is repeated until the change in the average value of the first solution set is less than a preset convergence threshold. The iteration process is then terminated, and the solution with the smallest function value in the current first solution set is taken as the global optimal adjustment coefficient. The convergence threshold and N are set by those skilled in the art according to the actual situation.

[0133] Step S4035: Update the current time-frequency resource block allocation rule according to the global optimal adjustment coefficient to obtain the updated time-frequency resource block allocation rule.

[0134] In this embodiment, an adjustment vector is constructed based on the globally optimal adjustment coefficient. The dimension of the adjustment vector is consistent with the parameter dimension of the time-frequency resource block allocation rule. The adjustment amount of each dimension is determined based on the difference between the globally optimal adjustment coefficient and the current resource allocation state. The time-frequency resource block allocation rule is converted into a parameter matrix. The parameter matrix is ​​updated by the adjustment vector to obtain the updated parameter matrix. The updated parameter matrix is ​​then scanned column by column to convert the element values ​​into time-frequency resource block allocation rules.

[0135] Step S404: Reconstruct the mapping relationship between time-frequency resource blocks and terminals based on adjustment coefficients, and calculate the interference coupling matrix of the initial network slice in combination with transmit power.

[0136] In this embodiment, a resource block weight vector is generated based on the adjustment coefficient and the time-frequency resource block allocation rule. A mapping table between time-frequency resource blocks and terminals is generated using the resource block weight vector. Based on the mapping table, the transmit power value configured for each terminal on the time-frequency resource block is extracted. The projection component of the transmit terminal signal to the receive terminal signal is calculated by combining the channel gain matrix. The transmit power value and the square of the modulus of the projection component are multiplied by a scalar to obtain the basic interference intensity of the transmit terminal and the receive terminal on the time-frequency resource block. The reciprocal of the frequency interval between time-frequency resource blocks is calculated as the orthogonality compensation factor. The basic interference intensity is multiplied by the orthogonality compensation factor to obtain the normalized unit interference quantity. The unit interference quantities are arranged according to the transmit terminal index and the receive terminal index to finally generate the interference coupling matrix of the initial network slice.

[0137] Step S405: When the maximum eigenvalue of the interference coupling matrix is ​​greater than the stability threshold, the tolerance threshold is updated by the network resource state deviation, and the network slice is obtained by solving the planning function. The stability threshold is determined by the network resource state deviation.

[0138] In this embodiment, based on the deviation of network resource status, the dynamic adjustment amount of the stability threshold and the dynamic adjustment amount of the tolerance threshold are calculated through a piecewise linear mapping function, where the piecewise linear mapping function includes linear scaling and tangent function compression. The basic stability threshold is added to the dynamic adjustment amount of the stability threshold to obtain the stability threshold. The basic stability threshold is determined according to the statistical distribution of the eigenvalues ​​of the interference coupling matrix. The interference coupling matrix is ​​decomposed into eigenvalues, and the sum of squares of the magnitudes of the eigenvalues ​​is calculated. An approximate solution of the largest eigenvalue is calculated through power iteration. This approximate solution is compared with the stability threshold. If it is greater, the network slice is reconstructed. The weight coefficients of the penalty term in the planning function are updated according to the magnitudes of the off-diagonal elements of the current interference coupling matrix. The weight of the penalty term for high interference links is proportional to the square of the magnitude of the corresponding off-diagonal element. Based on the ratio of the largest eigenvalue to the dynamic stability threshold, the slack variable boundary of the planning function is adjusted. The dynamic adjustment amount of the tolerance threshold is summed with the original tolerance threshold to obtain the updated tolerance threshold. The planning function is solved by recursive search of the solution space to obtain the final network slice.

[0139] Example 2:

[0140] Please see Figure 4 The present invention provides an embodiment of a network slicing and interference suppression system for surface substations, comprising a slice initialization module, an interference analysis module, a signal processing module, and a dynamic optimization module.

[0141] The slice initialization module is used to collect terminal request data and network channel data to establish a planning function, and to solve the planning function by constructing a branch tree structure to obtain the initial network slice.

[0142] The interference analysis module is used to derive the interference transfer function of the initial network slice based on the connection relationship and channel state of the initial network slice, and iteratively solve the interference transfer function to obtain the optimal power spectral density function and time-frequency resource block allocation rules.

[0143] The signal processing module is used to set filtering parameters according to the optimal power spectral density function, perform interference component separation processing on the interfered initial network slice, and perform signal compensation in combination with the time-frequency resource block allocation rules to obtain the interference-suppressed initial network slice and its corresponding quality evaluation vector.

[0144] The dynamic optimization module is used to construct a dynamic equilibrium function with the quality evaluation vector as the independent variable, quantify the deviation of the network resource state through the dynamic equilibrium function, update the time-frequency resource block allocation rules and optimize the initial network slice to obtain the network slice.

[0145] The signal processing module includes a filter control unit and a quality assessment unit.

[0146] The filter control unit is used to identify interference frequency points and bandwidths based on the optimal power spectral density function, and to calculate and set filter parameters.

[0147] The quality assessment unit is used for segmented quality monitoring, calculating signal-to-noise ratio, bit error rate, and delay jitter parameters, and performing numerical normalization to generate a quality assessment vector.

[0148] In addition, the parts of the technical solutions provided in the embodiments of this application that are consistent with the implementation principles of the corresponding technical solutions in the prior art have not been described in detail, so as to avoid excessive elaboration.

[0149] The embodiments of the present invention have been described above with reference to the accompanying drawings. However, the present invention is not limited to the specific embodiments described above. The specific embodiments described above are merely illustrative and not restrictive. Any improvements and modifications made by those skilled in the art under the guidance of the present invention without departing from the principles of the present invention should also be considered within the scope of protection of the present invention.

Claims

1. A method for substation network slicing and interference suppression, characterized in that, include: A planning function is established by collecting terminal request data and network channel data. The planning function is solved by constructing a branch tree structure to obtain the initial network slice. Based on the connection relationship and channel state of the initial network slice, the interference transfer function of the initial network slice is derived, and the optimal power spectral density function and time-frequency resource block allocation rule are obtained by iteratively solving the interference transfer function. The filtering parameters are set according to the optimal power spectral density function, and the interference components are separated on the initial network slice affected by the interference. The signal compensation is performed in combination with the time-frequency resource block allocation rules to obtain the initial network slice after interference suppression and its corresponding quality evaluation vector. A dynamic equilibrium function is constructed using the quality evaluation vector as the independent variable. The deviation of the network resource state is quantified by the dynamic equilibrium function. The time-frequency resource block allocation rule is updated and the initial network slice is optimized to obtain the network slice. The process of constructing a dynamic equilibrium function using the quality evaluation vector as the independent variable, quantifying the deviation of network resource states through the dynamic equilibrium function, updating the time-frequency resource block allocation rules, and optimizing the initial network slice to obtain the network slice includes: The quality assessment vector is mapped to coordinate points in three-dimensional space. The extreme path between the coordinate points and the ideal coordinate points is calculated. The dynamic equilibrium function is constructed by combining the vector field divergence of the coordinate points in each gradient direction. The ideal coordinate points are set according to the network service quality requirements. The partial derivatives of each component of the quality assessment vector are calculated using the dynamic equilibrium function to generate the gradient vector. The product of the gradient vector and the rate of change of each component of the quality assessment vector is calculated and integrated. The square root of the integral result is taken as the deviation of the network resource state. Based on the magnitude and trend of network resource status deviation, a correction parameter generation function is constructed. The adjustment coefficient that obtains the global minimum value of the correction parameter generation function is found through preset conditions, and the time-frequency resource block allocation rules are adjusted. The mapping relationship between time-frequency resource blocks and terminals is reconstructed based on the adjustment coefficients, and the interference coupling matrix of the initial network slice is calculated in combination with the transmit power. When the maximum eigenvalue of the interference coupling matrix is ​​greater than the stability threshold, the tolerance threshold is updated by the network resource state deviation, and the network slice is obtained by solving the planning function. The stability threshold is determined by the network resource state deviation.

2. The substation network slicing and interference suppression method according to claim 1, characterized in that, The acquisition terminal requests data and network channel data to establish a planning function. The planning function is solved by constructing a branch tree structure to obtain an initial network slice, including: By analyzing the service characteristics of terminal request data, it is mapped to bandwidth demand level and latency sensitivity coefficient, and channel gain matrix and interference power spectral density are constructed by combining channel state information in network channel data. A planning function is constructed based on the channel gain matrix and the interference power spectral density, combined with a penalty term. The penalty term is determined by measuring the product of the transmit power of the interference link and the power coupling coefficient. By initializing the branch tree with the root node resource allocation state of zero, the solution space of the planning function is recursively searched, child nodes are generated and their function upper bounds are calculated. When the cumulative interference of the link corresponding to the child node exceeds the preset tolerance threshold, the path expansion is terminated, and the solution that makes the planning function obtain the global minimum value is selected. The demapping is converted into a logical slice structure, and the priority scheduling rules of the logical slice structure are set based on the latency sensitivity coefficient. The initial network slice is then instantiated and generated.

3. The substation network slicing and interference suppression method according to claim 2, characterized in that, The process of deriving the interference transfer function of the initial network slice based on the connection relationship and channel state, and iteratively solving the interference transfer function to obtain the optimal power spectral density function and time-frequency resource block allocation rules includes: Based on the topology connection parameters, state transition matrix, and channel response parameters of the initial network slice, a dynamic evolution equation is constructed, where the state transition matrix is ​​calculated from the channel gain matrix and topology connection parameters. The rate of change of the current channel unit output signal is calculated based on the dynamic evolution equation and the preset perturbation variables. Based on the analytical expression of the ratio function of the rate of change to the perturbation variables in the frequency domain, the interference transfer function is constructed. The aggregated interference power of different channel units is calculated based on the interference transfer function, and an objective function is established with the minimum communication rate and the maximum transmit power as the boundary. By initializing the power spectral density function and the time-frequency resource block allocation rule, the objective function is solved iteratively with the power spectral density function and the time-frequency resource block allocation rule fixed respectively, until its difference norm is less than the preset difference value, and the optimal power spectral density function and time-frequency resource block allocation rule are output.

4. The substation network slicing and interference suppression method according to claim 3, characterized in that, The process involves setting filtering parameters based on the optimal power spectral density function, performing interference component separation on the interfered initial network slice, and combining this with time-frequency resource block allocation rules for signal compensation. This yields the interference-suppressed initial network slice and its corresponding quality evaluation vector, including: Based on the amplitude-frequency characteristics of the optimal power spectral density function, identify the frequency points and bandwidths where interference power is concentrated, and set the center frequency and stopband attenuation coefficient based on the frequency points and bandwidth respectively. A filter function is constructed based on the center frequency and the stopband attenuation coefficient. The interference signal in the initial network slice is input into the filter function for convolution operation to obtain the effective signal component. The effective signal components are compared with the ideal reference signal in the time and frequency domains. The group delay and amplitude distortion generated by the effective signal components are analyzed and calculated, and a compensation factor is generated. The time base deviation and phase rotation of the effective signal components are corrected according to the time-frequency resource block allocation rules, and the initial network slice after interference suppression is generated by the compensation factor and the effective signal components. The initial network slices after interference suppression are segmented, and the signal-to-noise ratio, bit error rate, and time delay jitter parameters of each segment are calculated and normalized. The segments are then combined to generate a quality assessment vector.

5. The substation network slicing and interference suppression method according to claim 4, characterized in that, The step of comparing the effective signal component with the ideal reference signal in the time-frequency domain, analyzing and calculating the group delay and amplitude distortion generated by the effective signal component, and generating a compensation factor includes: The ideal reference signal is extracted based on the pilot position defined by the time-frequency resource block allocation rules, and time-frequency synchronization alignment is performed by calculating the peak value of the cross-correlation function between the effective signal component and the ideal reference signal. Based on the aligned effective signal components and the ideal reference signal, the complex value of the channel transfer function on each subcarrier is calculated in the frequency domain. The complex value of the channel transfer function is used as matrix elements to establish the channel response matrix. A group delay analytical expression is established by performing difference operations and polynomial fitting on the channel response matrix along the frequency dimension, and the standard deviation of the amplitude value is calculated along the time dimension to obtain the amplitude distortion. Based on the analytical expression of group delay and amplitude distortion, the phase compensation and amplitude compensation of the time-frequency resource block are calculated, and the phase compensation and amplitude compensation are converted into compensation factors in complex form.

6. The substation network slicing and interference suppression method according to claim 5, characterized in that, The process involves mapping the quality assessment vector to coordinate points in three-dimensional space, calculating the extreme path between the coordinate points and the ideal coordinate points, and constructing a dynamic equilibrium function by combining the vector field divergence of the coordinate points in each gradient direction. The ideal coordinate points are set according to network service quality requirements, including: The tangent, negative logarithm, and reciprocal square root values ​​of the components of the quality assessment vector are calculated, and these are used as basis vectors for orthogonalization to generate an orthogonal feature set. Construct a metric tensor for three-dimensional space based on the orthogonal feature set, determine the coordinate points under the metric tensor, calculate the extreme path between the coordinate points and the ideal coordinate points, and take the length of the extreme path as the minimum path length. Construct a mass state vector field in three-dimensional space, connect it to a metric tensor, calculate the covariant derivative of the mass state vector field, and obtain the divergence of the mass state vector field by finding the trace of the covariant derivative. The components of the mass state vector field are obtained from the time derivative of the orthogonal feature set. A dynamic equilibrium function is constructed based on the minimum path length, the divergence of the quality state vector field, and the stability constraint term, wherein the stability constraint term is obtained based on the fluctuation amplitude of each component of the orthogonal feature set.

7. The substation network slicing and interference suppression method according to claim 6, characterized in that, The process involves constructing a correction parameter generation function based on the magnitude and trend of network resource status deviation, finding the adjustment coefficient that yields the global minimum value of the correction parameter generation function through preset conditions, and adjusting the time-frequency resource block allocation rules, including: The first and second derivatives of the network resource state deviation are calculated, and a third-order eigenvector is established using the network resource state deviation, its first and second derivatives. Using the third-order eigenvector as the independent variable, a modified parameter generation function containing an exponential decay term and a periodic oscillation term is constructed. The exponential decay term is determined by the magnitude of the network resource state deviation, and the periodic oscillation term is determined by the frequency of the network resource state deviation. The feasible range of the adjustment coefficient is set according to the number of time-frequency resource blocks. The initial candidate solution set is generated by uniform sampling. The square root of the sum of squares of the differences between the initial candidate solution and each component of the third-order feature vector is calculated to obtain the function value set. The probability distribution function is established based on the function value set. A candidate solution set is generated based on the probability distribution function. The function value set is iteratively calculated until the average function value of the function value set is less than the preset convergence value. The candidate solution with the smallest function value is taken as the global optimal adjustment coefficient. The current time-frequency resource block allocation rule is updated and calculated based on the globally optimal adjustment coefficient to obtain the updated time-frequency resource block allocation rule.

8. A substation network slicing and interference suppression system, used to implement the substation network slicing and interference suppression method according to any one of claims 1-7, characterized in that, It includes a slice initialization module, an interference analysis module, a signal processing module, and a dynamic optimization module: The slice initialization module is used to collect terminal request data and network channel data to establish a planning function, and solve the planning function by constructing a branch tree structure to obtain the initial network slice; The interference analysis module is used to derive the interference transfer function of the initial network slice based on the connection relationship and channel state of the initial network slice, and iteratively solve the interference transfer function to obtain the optimal power spectral density function and time-frequency resource block allocation rules. The signal processing module is used to set filtering parameters according to the optimal power spectral density function, perform interference component separation processing on the interfered initial network slice, and perform signal compensation in combination with the time-frequency resource block allocation rules to obtain the interference-suppressed initial network slice and its corresponding quality evaluation vector. The dynamic optimization module is used to construct a dynamic equilibrium function with the quality evaluation vector as the independent variable, quantify the deviation of the network resource state through the dynamic equilibrium function, update the time-frequency resource block allocation rules and optimize the initial network slice to obtain the network slice.

9. The substation network slicing and interference suppression system according to claim 8, characterized in that, The signal processing module includes a filter control unit and a quality evaluation unit: The filter control unit is used to identify the interference frequency and bandwidth according to the optimal power spectral density function, and to calculate and set the filter parameters. The quality assessment unit is used for segmented quality monitoring, calculating signal-to-noise ratio, bit error rate, and delay jitter parameters, and performing numerical normalization to generate a quality assessment vector.

Images