5G protocol and line short compatible intercommunication method and system

Abstract

The present application relates to the field of protocol switching technology, and discloses a method and system for 5G protocol and line short compatibility intercommunication. The method includes: collecting network data through a distributed sensor network and performing real-time collection and analysis to obtain spectrum demand prediction results; calculating the high and low frequency band allocation ratios based on the spectrum demand prediction results to obtain spectrum allocation strategies; performing protocol identification and analysis on communication data to obtain data information to be converted; reconstructing and encapsulating data based on adaptive protocol conversion rules to obtain communication data after cross-protocol conversion; performing MIMO mode selection and antenna configuration optimization on the communication data after cross-protocol conversion to obtain MIMO transmission parameters; dynamically adjusting and evaluating the performance of MIMO transmission parameters to obtain the optimal communication link configuration. The present application can intelligently perceive the network environment, dynamically adjust the spectrum allocation strategy, and achieve seamless conversion between different protocols.

Description

Technical Field

[0001] The present application relates to the field of protocol switching technology, and in particular to a 5G protocol and short line compatible intercommunication method and system. Background Art

[0002] With the rapid development and widespread application of 5G technology, mobile communication networks are facing unprecedented challenges and opportunities. 5G networks need to support a variety of application scenarios such as massive connections, ultra-high speeds, and ultra-low latency, which requires the network to have higher flexibility and adaptability. However, existing communication protocols and network architectures often find it difficult to meet the needs in the complex and changing 5G environment, especially in achieving interoperability with traditional industry-specific networks.

[0003] Currently, the protocol compatibility issues between 5G networks and industry-specific networks are becoming increasingly prominent. Different communication standards and protocol specifications make it difficult for networks to directly exchange data and collaborate on services, which seriously restricts the in-depth application of 5G technology in vertical industries. In addition, the contradiction between the limited spectrum resources and the growing communication needs is becoming increasingly prominent. How to achieve efficient spectrum allocation and utilization has become an urgent problem to be solved. Summary of the invention

[0004] The present application provides a 5G protocol and short line compatible intercommunication method and system for intelligently sensing the network environment, dynamically adjusting the spectrum allocation strategy, and realizing seamless conversion between different protocols.

[0005] In a first aspect, the present application provides a 5G protocol and line short compatible intercommunication method, the 5G protocol and line short compatible intercommunication method comprising:

[0006] Collecting network data through a distributed sensor network, and collecting and analyzing the network data in real time to obtain spectrum demand prediction results;

[0007] Calculate the high and low frequency band allocation ratios based on the spectrum demand prediction result to obtain a spectrum allocation strategy;

[0008] Based on the spectrum allocation strategy, protocol identification and analysis are performed on the communication data to obtain data information to be converted;

[0009] Reconstructing and encapsulating the data information to be converted based on the adaptive protocol conversion rules to obtain communication data after cross-protocol conversion;

[0010] Performing MIMO mode selection and antenna configuration optimization on the communication data after the cross-protocol conversion to obtain MIMO transmission parameters;

[0011] The MIMO transmission parameters are dynamically adjusted and performance evaluated to obtain an optimal communication link configuration.

[0012] In a second aspect, the present application provides a 5G protocol and line short compatible intercommunication system, and the 5G protocol and line short compatible intercommunication system includes:

[0013] A collection module, used to collect network data through a distributed sensor network, and to collect and analyze the network data in real time to obtain spectrum demand prediction results;

[0014] A calculation module, used to calculate the high and low frequency band allocation ratios based on the spectrum demand prediction result to obtain a spectrum allocation strategy;

[0015] A parsing module, used to perform protocol identification and parsing on the communication data based on the spectrum allocation strategy to obtain data information to be converted;

[0016] A reconstruction module, used to reconstruct and encapsulate the data information to be converted based on the adaptive protocol conversion rules to obtain communication data after cross-protocol conversion;

[0017] An optimization module, used to perform MIMO mode selection and antenna configuration optimization on the communication data after the cross-protocol conversion to obtain MIMO transmission parameters;

[0018] The adjustment module is used to dynamically adjust the MIMO transmission parameters and perform performance evaluation to obtain an optimal communication link configuration.

[0019] In the technical solution provided by the present application, through the distributed sensor network and the multi-layer perceptron neural network model, the accurate prediction of spectrum demand is achieved, and the utilization efficiency of spectrum resources is improved. Based on the adaptive protocol conversion rules, flexible conversion between different protocols is achieved, and the intercommunication problem between 5G network and industry-specific network is solved. Through MIMO mode selection and antenna configuration optimization, the transmission performance and energy efficiency of the system are improved. Real-time optimization of communication links is achieved, and the adaptability of the system to complex network environments is enhanced. The convolutional neural network model is used for protocol type identification, which improves the accuracy and efficiency of protocol identification. Through linear programming and genetic algorithms, dynamic optimization of spectrum allocation strategies is achieved, and the utilization efficiency of spectrum resources is improved. Through adaptive protocol conversion rules, accurate reconstruction and encapsulation of data are achieved, ensuring the reliability of cross-protocol communication. Through multi-objective optimization methods, performance evaluation and dynamic adjustment of communication links are achieved, and seamless conversion between different protocols is achieved. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0021] Figure 1 This is a schematic diagram of an embodiment of the 5G protocol and line short compatibility interoperability method in the embodiment of the present application;

[0022] Figure 2 This is a schematic diagram of an embodiment of the 5G protocol and line short compatible intercommunication system in the embodiment of the present application. DETAILED DESCRIPTION

[0023] The embodiments of the present application provide a 5G protocol and line short compatible intercommunication method and system. The terms "first", "second", "third", "fourth", etc. (if any) in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments described herein can be implemented in an order other than that illustrated or described herein. In addition, the terms "including" or "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0024] For ease of understanding, the specific process of the embodiment of the present application is described below. Figure 1 In the embodiment of the present application, an embodiment of the 5G protocol and line short compatible intercommunication method includes:

[0025] Step S101: Collect network data through a distributed sensor network, and collect and analyze the network data in real time to obtain spectrum demand prediction results;

[0026] It is understandable that the execution subject of this application can be a 5G protocol and line short compatible intercommunication system, or a terminal or a server, which is not limited here. The embodiment of this application is described by taking the server as the execution subject as an example.

[0027] Specifically, data collection tasks are assigned to each node in the distributed sensor network to obtain a node task list. Based on the node task list, the signal strength, user density and data flow of each node are monitored in real time to obtain comprehensive network data, including detailed operating conditions of each node and the communication environment data around it. These data reflect the operating status and user needs of the entire network. The network data is cleaned and standardized to remove outliers and noise data, ensure the consistency and accuracy of the data, and obtain a preprocessed network data set. The preprocessed network data set is time-series segmented to divide it into multiple time series data subsets. Time series segmentation can organize continuous data in time order, so that data analysis can be performed for a specific time period, which is convenient for observing the change trend and periodic characteristics of the data. Fast Fourier transform is performed on the time series data subset. Fast Fourier transform is a commonly used signal processing technology that converts time domain data into frequency domain data, extracts the spectrum features in the data, and obtains the spectrum feature vector. The spectrum feature vector can reflect the distribution and change of network data in the frequency domain. The spectrum feature vector is input into the preset multi-layer perceptron neural network model for spectrum demand prediction. The multi-layer perceptron neural network model is a deep learning model with nonlinear fitting and generalization capabilities. It can accurately predict future spectrum demand by learning from a large amount of historical data. Through the training and optimization of the neural network model, it has a high prediction accuracy and obtains the spectrum demand prediction results.

[0028] Normalize the spectral feature vectors to map the feature values ​​of different ranges into a unified range, usually [0, 1] or [-1, 1], to avoid the negative impact of the magnitude difference of the feature values ​​on the model training. After normalization, a normalized feature vector is obtained, and the normalized feature vector is used as the input data of the input layer in the multi-layer perceptron neural network model. According to the normalized feature vector, the activation function is calculated for the input layer neurons, usually using a linear activation function, to obtain the input layer output vector. The input layer output vector is matrix multiplied with the weight matrix of the first hidden layer to obtain the linear combination result of the first hidden layer. In order to introduce nonlinear characteristics, the linear combination result of the first hidden layer is activated by the ReLU (rectified linear unit) activation function. The ReLU activation function can effectively solve the gradient vanishing problem and improve the training efficiency of the model. After ReLU activation, the activation output of the first hidden layer is obtained. The activation output of the first hidden layer is matrix multiplied with the weight matrix of the second hidden layer to obtain the linear combination result of the second hidden layer. In order to introduce nonlinear characteristics and limit the output range, the Sigmoid activation function is operated on the linear combination result of the second hidden layer. The Sigmoid activation function can map the output to the interval (0, 1) to obtain the activation output of the second hidden layer. The activation output of the second hidden layer is matrix multiplied with the weight matrix of the output layer to obtain the linear combination result of the output layer. In order to convert the linear combination result into a probability distribution, the Softmax activation function is operated on the linear combination result of the output layer. The Softmax activation function maps the output value to the probability distribution of the demand of each frequency band. Through this process, the demand probability of each frequency band is obtained. According to the probability distribution of the demand of each frequency band, in order to determine the most likely frequency band demand, the high probability frequency band is selected by the threshold screening method. A probability threshold is set during the screening process. Only the frequency band exceeding the threshold is considered as a high demand frequency band. In order to smooth the prediction results and reduce the fluctuations caused by random factors, the selected high probability frequency band is smoothed by exponential moving average. Exponential moving average smoothing can smooth the short-term fluctuations of the data by assigning exponential decay weights to the time series data, and obtain more stable spectrum demand prediction results.

[0029] Step S102: Calculate the high and low frequency band allocation ratios based on the spectrum demand prediction result to obtain a spectrum allocation strategy;

[0030] Specifically, the spectrum demand prediction results are classified into frequency bands, and the demand results are divided into two parts: high frequency band and low frequency band, and the demand for high frequency band and low frequency band is obtained. According to the demand for high frequency band and low frequency band, the initial frequency band allocation ratio is calculated to obtain the initial allocation scheme. The objective function and constraint conditions of the initial allocation scheme are constructed based on the linear programming model to generate the optimization problem expression. The objective function aims to maximize the spectrum utilization efficiency or minimize the interference, while the constraint conditions take into account the actual spectrum resources and demand restrictions. The optimization problem expression is solved by the simplex method to obtain the basic feasible solution set. The simplex method is a classic linear programming solution algorithm that can effectively find the solution set of the optimization problem. The basic feasible solution set is crossover and mutation operations are performed to obtain the candidate solution population. The crossover operation generates a new solution by combining some features of different solutions, while the mutation operation introduces diversity and novelty by randomly changing certain features. These operations can simulate the biological evolution process to explore a better solution space. According to the preset fitness function, each solution in the candidate solution population is evaluated to obtain a fitness score. The fitness function usually comprehensively evaluates each solution based on indicators such as spectrum utilization efficiency, interference level, and service quality. The higher the score, the better the solution. Based on the fitness score, the solution with the highest fitness is selected from the candidate solution population to generate an optimized frequency band allocation plan. The optimized frequency band allocation plan is dynamically adjusted and discrete cosine transformed. Dynamic adjustment fine-tunes the plan through real-time monitoring and feedback mechanisms to adapt to changes in spectrum requirements. Discrete cosine transform is used to smooth the frequency band allocation data, reduce the impact of mutations, and improve the stability and robustness of the plan.

[0031] Step S103: Based on the spectrum allocation strategy, protocol identification and analysis are performed on the communication data to obtain data information to be converted;

[0032] Specifically, according to the spectrum allocation strategy, the received communication data is divided into frequency bands to obtain multiple frequency band data subsets, and the frequency domain features of the frequency band data subsets are extracted to obtain a feature vector set. By analyzing the data features of different frequency bands, the original data is converted into a vector set that can reflect the characteristics of the communication protocol. The feature vector set is input into the pre-trained convolutional neural network model for preliminary identification of the protocol type. The convolutional neural network model extracts the deep features of the data through layer-by-layer convolution and pooling operations, accurately identifies the protocol type, and obtains the probability distribution of the protocol type, that is, the probability that each feature vector corresponds to each protocol type. The maximum a posteriori probability criterion analysis is performed on the probability distribution of the protocol type, and the protocol type with the highest probability is selected as the initial identification result. According to the initial identification result, the corresponding protocol parsing rules are retrieved from the protocol template library to obtain a candidate parsing rule set. The protocol template library stores the parsing rules of various protocols, and the corresponding rules are retrieved according to different protocol types for data parsing. Based on each rule in the candidate parsing rule set, the communication data is parsed in parallel to obtain multiple parsing results. Parallel parsing can improve the parsing efficiency, and at the same time, the parsing results of different rules are compared and analyzed. Conduct credibility assessment on multiple parsing results to obtain credibility scores for each parsing result. Credibility assessment compares the rationality and consistency of parsing results, calculates the credibility score of each result, and ensures that the selected parsing result is the most accurate. Based on the credibility score, select the optimal parsing result and obtain the protocol field structure. The protocol field structure is a specific description of the communication data, including the name, type, and length of each field. Perform semantic analysis on the protocol field structure to obtain semantically annotated data. Semantic analysis clarifies the actual meaning and purpose of the data by understanding the meaning and relationship between each field. Associate and map the semantically annotated data with the communication data to construct a data dependency graph. The data dependency graph shows the dependencies and transmission paths between data, thereby revealing the overall structure and flow of the data.

[0033] Normalize the feature vector set, map the feature values ​​of different ranges to a unified standard range, usually [0, 1] or [-1, 1], eliminate the magnitude differences between features, and obtain a standardized feature vector set. Input the standardized feature vector set into the first convolution layer of the pre-trained convolutional neural network model. This convolution layer uses 32 3x3 convolution kernels for convolution operation, scans the input feature vector set one by one through a sliding window, and extracts local feature information. The convolution operation applies the ReLU activation function, which can effectively introduce nonlinear characteristics while avoiding the gradient vanishing problem. Perform a maximum pooling operation, downsample the convolution feature map through the pooling layer, reduce the dimension of the feature map, improve the computational efficiency of the model, and obtain the first layer of feature maps. Input the first layer of feature maps into the second convolution layer. The second convolution layer uses 64 3x3 convolution kernels for convolution operation to extract deeper feature information, and applies the ReLU activation function again to increase the nonlinear expression ability of the model, and continues to downsample through the maximum pooling operation to obtain the second layer of feature maps. Compared with the first-layer feature map, the second-layer feature map contains more high-level feature information, which enables the model to more accurately identify complex protocol type features. The second-layer feature map is input into the third convolutional layer. The third convolutional layer uses 128 3x3 convolution kernels for convolution operation. Through more convolution kernels and deeper convolutional layers, more abstract and high-level features are extracted. After applying the ReLU activation function, a global average pooling operation is performed to average all values ​​of each feature map to obtain a global feature vector. Global average pooling can not only reduce the dimension of the feature, but also retain global information and enhance the generalization ability of the model. The global feature vector is processed by the fully connected layer. The fully connected layer uses 256 neurons and performs nonlinear transformation through the ReLU activation function to obtain a high-dimensional feature representation. In order to prevent overfitting, the high-dimensional feature representation is input into the Dropout layer and randomly set to zero with a probability of 0.5, setting half of the neuron outputs to zero. This operation can increase the robustness of the model and avoid overfitting. The regularized features after the Dropout layer are processed by the final fully connected layer, and the number of neurons is equal to the number of protocol types to obtain the original output vector. The original output vector is input into the Softmax layer for exponential normalization. The Softmax layer can convert the original output into the probability distribution of each protocol type. Based on the maximum a posteriori probability criterion, the category with the highest probability in the probability distribution of the protocol type is selected as the initial recognition result. The maximum probability of the initial recognition result is compared with the preset confidence threshold. Only the results exceeding the threshold are accepted, and finally the initial recognition result of the protocol type is obtained.

[0034] Step S104, reconstructing and encapsulating the data information to be converted based on the adaptive protocol conversion rule to obtain communication data after cross-protocol conversion;

[0035] Specifically, the protocol field analysis is performed on the data information to be converted, and the name, type and meaning of each field are parsed to obtain the field feature set. According to the field feature set, the matching rules are retrieved from the preset protocol conversion rule library to obtain the initial conversion rule set. The rule library stores conversion rules for different protocols, and the conversion rules that may be applicable are screened out by matching the field features. The applicability of each rule in the initial conversion rule set is calculated to obtain the rule weight distribution. The applicability calculation is based on the matching degree between the field features and the conversion rules. By calculating the weight of each rule, its applicability in the specific conversion scenario is evaluated. Based on the rule weight distribution, the adaptive protocol conversion rule is generated. The adaptive protocol conversion rule is an optimization of the initial conversion rule set. Through weight distribution screening and adjustment, the most suitable rule for the current conversion task is selected to ensure the accuracy and efficiency of the conversion process. The adaptive protocol conversion rule is applied to the data information to be converted, and field mapping and conversion are performed to obtain a preliminary conversion result. In the field mapping process, according to the adaptive protocol conversion rule, the fields in the data information to be converted are converted to the corresponding fields in the target protocol to complete the conversion of the data format and structure. Perform a semantic consistency check on the preliminary conversion results. By comparing the meaning of the data before and after the conversion, possible semantic conflicts are identified and a conflict marker list is generated. According to the conflict marker list, the conflict fields are adjusted to obtain the corrected conversion data to eliminate the semantic inconsistency in the conversion process. The corrected conversion data is input into the encapsulation template of the target protocol, and the protocol header and checksum are generated to complete the data encapsulation process. The protocol header contains the control information required by the target protocol, and the checksum is used to verify the data integrity.

[0036] Step S105: performing MIMO mode selection and antenna configuration optimization on the communication data after cross-protocol conversion to obtain MIMO transmission parameters;

[0037] Specifically, the communication data after cross-protocol conversion is analyzed in terms of data volume and service type. By analyzing the characteristics of the communication data, a communication demand feature vector is obtained, which includes key parameters such as the magnitude of the communication data, the transmission rate requirement, and the service quality requirement. According to the communication demand feature vector, a candidate mode set is extracted from a preset MIMO mode library to form an initial MIMO mode list. The MIMO mode library stores a variety of predefined MIMO transmission modes. By matching the communication demand feature vector, a mode suitable for the current communication task is screened out. The channel capacity is calculated for each MIMO mode in the initial MIMO mode list. The channel capacity is an important indicator for measuring the performance of the communication system, which indicates the maximum data rate that can be transmitted under given channel conditions. After calculating the theoretical throughput of each MIMO mode, the energy efficiency ratio of each MIMO mode is calculated based on the theoretical throughput. The energy efficiency ratio reflects the amount of data that can be transmitted per unit energy consumption and is an important indicator for optimizing the energy efficiency of the communication system. The optimal MIMO mode is generated according to the energy efficiency evaluation result. An exhaustive search is performed on the number of antennas under the optimal MIMO mode. By trying different antenna configurations, the signal-to-noise ratio of each configuration is calculated to obtain a candidate antenna configuration set. Signal-to-noise ratio is an important indicator for measuring communication quality. A higher signal-to-noise ratio means better communication quality and lower bit error rate. The candidate antenna configuration set is optimized, and the configuration scheme with the best signal-to-noise ratio is selected to obtain the optimized antenna configuration. Based on the optimized antenna configuration, the precoding matrix and the equalization matrix are calculated. The precoding matrix is ​​used to preprocess the signal at the transmitter to optimize the transmission effect of the signal in the multipath channel. The equalization matrix is ​​used to process the received signal at the receiver to compensate for the distortion and interference caused by the channel to obtain the MIMO system matrix. The MIMO system matrix is ​​characteristically decomposed to represent the complex MIMO channel as multiple independent sub-channels to obtain the spatial stream mapping scheme. The spatial stream mapping scheme determines the transmission mode and path of the signal on each sub-channel. The optimal MIMO mode, the optimized antenna configuration, the MIMO system matrix and the spatial stream mapping scheme are combined to form a complete MIMO transmission parameter.

[0038] Step S106: Dynamically adjust MIMO transmission parameters and evaluate performance to obtain an optimal communication link configuration.

[0039] Specifically, the MIMO transmission parameters are initialized and set to generate the initial state of the MIMO system. Based on the initial state of the MIMO system, the real-time channel state information is calculated to obtain the channel matrix. The channel matrix reflects the instantaneous state of the wireless channel and is an important basis for channel evaluation and optimization. The channel matrix is ​​decomposed and calculated to obtain the channel quality index. The channel quality index includes key parameters such as channel capacity and signal-to-noise ratio, which directly affect the performance of the communication system. According to the channel quality index, the best modulation mode and coding rate are selected to obtain the modulation and coding scheme. The selection of the modulation mode and coding rate needs to comprehensively consider the channel conditions and system performance to achieve the best transmission efficiency and reliability. The modulation and coding scheme is combined with the MIMO transmission parameters to generate the theoretical throughput prediction value. The theoretical throughput prediction value is an important indicator for measuring system performance, which indicates the maximum transmission rate that the system can achieve under the current configuration. The performance difference index of the theoretical throughput prediction value is calculated. The performance difference index refers to the difference between the actual throughput and the theoretical throughput, which is used to evaluate the effectiveness and performance potential of the current system configuration. Based on the performance difference index, the MIMO transmission parameters are adjusted and the performance is calculated. The adjustment process is a dynamic optimization process. According to the real-time channel status and performance feedback, the transmission parameters are continuously adjusted to approach the theoretical optimal performance. After adjusting the transmission parameters, the real-time performance indicators are obtained. The real-time performance indicators reflect the actual performance of the current system configuration, including indicators in multiple dimensions such as throughput, delay, and bit error rate. Multi-objective optimization is performed on the real-time performance indicators. Based on considering multiple performance indicators, a solution with the best comprehensive performance is found. Through multi-objective optimization, the relationship between different indicators is balanced, and a communication link configuration with the best overall performance is found, and the optimal communication link configuration is obtained.

[0040] In the embodiment of the present application, through the distributed sensor network and the multi-layer perceptron neural network model, the spectrum demand is accurately predicted and the utilization efficiency of the spectrum resources is improved. Based on the adaptive protocol conversion rules, flexible conversion between different protocols is realized, and the intercommunication problem between 5G network and industry-specific network is solved. Through MIMO mode selection and antenna configuration optimization, the transmission performance and energy efficiency of the system are improved. Real-time optimization of the communication link is achieved, and the adaptability of the system to complex network environments is enhanced. The convolutional neural network model is used for protocol type identification, which improves the accuracy and efficiency of protocol identification. Through linear programming and genetic algorithms, the spectrum allocation strategy is dynamically optimized and the utilization efficiency of spectrum resources is improved. Through adaptive protocol conversion rules, accurate reconstruction and encapsulation of data are achieved, ensuring the reliability of cross-protocol communication. Through multi-objective optimization methods, performance evaluation and dynamic adjustment of communication links are achieved, and seamless conversion between different protocols is achieved.

[0041] In a specific embodiment, the process of executing step S101 may specifically include the following steps:

[0042] Assign data collection tasks to each node in the distributed sensor network to obtain a node task list, and monitor the signal strength, user density and data flow of each node in real time based on the node task list to obtain network data;

[0043] Perform data cleaning and standardization on the network data to obtain a preprocessed network data set, and perform time series segmentation on the preprocessed network data set to obtain multiple time series data subsets;

[0044] Perform fast Fourier transform on the time series data subset to obtain the spectrum feature vector;

[0045] The spectrum feature vector is input into a preset multi-layer perceptron neural network model to perform spectrum demand prediction and obtain a spectrum demand prediction result.

[0046] Specifically, a task allocation strategy is formulated so that each node can efficiently complete the data collection task. Suppose there is a distributed sensor network consisting of Each node consists of Specific data collection tasks need to be performed according to network requirements. Define a task allocation matrix ,in Representation Node Execute the task The status of the node (1 means execution, 0 means not execution). It can be expressed as:

[0047] ;

[0048] Once the task allocation is completed, each node According to the task list Start collecting information such as signal strength, user density, and data traffic. Real-time monitoring of network data Contains each node at time Signal strength at the moment , user density and data traffic , its data form can be expressed as:

[0049] ;

[0050] Collect real-time data sets across the network over time ,in Contains real-time monitoring data of all nodes. After data collection is completed, the network data is cleaned and standardized. Data cleaning includes removing outliers and filling missing values ​​to ensure the accuracy and consistency of the data. Standardization makes the data have the same dimension and range. After standardization, the preprocessed network data set is obtained. The preprocessed data set is time-series segmented to divide the continuous data set into multiple time series data subsets to capture the changing characteristics of the data in different time periods, which is convenient for subsequent spectrum analysis. Fast Fourier transform is performed on each time series data subset to convert the time domain data into frequency domain data and extract the spectrum feature vector. The formula for fast Fourier transform is as follows:

[0051] ;

[0052] in, Indicates time domain samples, represents the total number of samples, Indicates Through fast Fourier transform, the spectral feature vector of each time series data subset is obtained . The spectral feature vectors of all time series data subsets are After summarizing, the input is fed into the preset multi-layer perceptron neural network model for spectrum demand prediction. Multi-layer perceptron is a common feedforward neural network, which consists of an input layer, one or more hidden layers, and an output layer. Assume that the input feature vector is , the activation function of the hidden layer is , the weight matrix and bias vector are and , then the hidden layer output It can be expressed as:

[0053] ;

[0054] The output layer uses a linear activation function to convert the hidden layer output into the final prediction result.

[0055] ;

[0056] in, and are the weight matrix and bias vector of the output layer respectively. Through the trained multi-layer perceptron model, the input spectrum feature vector , and get the spectrum demand prediction result , that is, the spectrum demand in each time period.

[0057] In a specific embodiment, the execution step inputs the spectrum feature vector into a preset multi-layer perceptron neural network model to perform spectrum demand prediction, and the process of obtaining the spectrum demand prediction result may specifically include the following steps:

[0058] Normalizing the spectrum feature vector to obtain a normalized feature vector, and using the normalized feature vector as input data of an input layer in a multi-layer perceptron neural network model;

[0059] According to the normalized eigenvector, the activation function is calculated for the input layer neurons to obtain the input layer output vector, and the input layer output vector is matrix multiplied with the first hidden layer weight matrix to obtain the linear combination result of the first hidden layer;

[0060] Perform a ReLU activation function operation on the linear combination result of the first hidden layer to obtain the activation output of the first hidden layer, and perform a matrix multiplication operation on the activation output of the first hidden layer and the weight matrix of the second hidden layer to obtain the linear combination result of the second hidden layer;

[0061] Perform a Sigmoid activation function operation on the linear combination result of the second hidden layer to obtain the activation output of the second hidden layer, and perform a matrix multiplication operation on the activation output of the second hidden layer and the weight matrix of the output layer to obtain the linear combination result of the output layer;

[0062] Perform Softmax activation function operation on the linear combination results of the output layer to obtain the probability distribution of demand for each frequency band;

[0063] According to the probability distribution of demand in each frequency band, the high probability frequency band is selected through the threshold screening method and the exponential moving average smoothing is performed to obtain the spectrum demand prediction result.

[0064] Specifically, the original spectrum feature vector is normalized to map the feature values ​​of different ranges into a unified standard range, usually [0, 1] or [-1, 1], to eliminate the magnitude difference between the features and obtain a normalized feature vector. The normalized feature vector is used as the input data of the input layer in the multi-layer perceptron neural network model. The input layer calculates the activation function for each input neuron to obtain the input layer output vector . Assume that the input layer weight matrix is , the bias vector is , the activation function is , then the output of the input layer can be expressed as:

[0065] ;

[0066] in, represents the matrix multiplication of the weight matrix and the input vector, Represents the bias vector. Usually, the activation function of the input layer can choose linear activation or ReLU activation function. As the input of the first hidden layer, matrix multiplication is performed on it to obtain the linear combination result of the first hidden layer. Assume that the weight matrix of the first hidden layer is , the bias vector is , then the linear combination result of the first hidden layer is:

[0067] ;

[0068] right Perform ReLU activation function operation. The formula of ReLU activation function is:

[0069] ;

[0070] Get the activation output of the first hidden layer .Will As the input of the second hidden layer, matrix multiplication is performed on it to obtain the linear combination result of the second hidden layer. Assume that the weight matrix of the second hidden layer is , the bias vector is , then the linear combination result of the second hidden layer is:

[0071] ;

[0072] right Perform Sigmoid activation function operation. The formula of Sigmoid activation function is:

[0073] ;

[0074] Get the activation output of the second hidden layer .Will As the input of the output layer, matrix multiplication is performed on it to obtain the linear combination result of the output layer. Assume that the weight matrix of the output layer is , the bias vector is , then the linear combination result of the output layer is:

[0075] ;

[0076] right Perform Softmax activation function operation. The formula of Softmax activation function is:

[0077] ;

[0078] in, Indicates The demand probability of a frequency band, Represents the first linear combination result of the output layer elements, Represents the exponential sum of all output elements. Through the Softmax activation function, the linear combination result of the output layer is converted into the demand probability distribution of each frequency band. According to the demand probability distribution of each frequency band, the high probability frequency band is selected through the set threshold screening method. Assume that the threshold is , then the screening condition for high probability frequency bands is:

[0079] ;

[0080] The selected high-probability frequency bands are smoothed by exponential moving average. The smoothing formula is:

[0081] ;

[0082] in, Indicates The smoothed value at time, Indicates The demand probability at time represents the smoothing factor, Represents the smoothed value of the previous moment. Through exponential moving average smoothing, the fluctuation of demand probability is reduced, the stability of the prediction result is improved, and the spectrum demand prediction result is obtained.

[0083] In a specific embodiment, the process of executing step S102 may specifically include the following steps:

[0084] Classify the spectrum demand prediction results into frequency bands to obtain the demand for high-frequency bands and low-frequency bands;

[0085] According to the demand for high frequency bands and low frequency bands, the initial frequency band allocation ratio is calculated to obtain an initial allocation plan;

[0086] Construct the objective function and constraints of the initial allocation scheme based on the linear programming model and generate the optimization problem expression;

[0087] The simplex method is used to solve the optimization problem expression to obtain a basic feasible solution set, and the basic feasible solution set is subjected to crossover and mutation operations to obtain a candidate solution population;

[0088] According to the preset fitness function, each solution in the candidate solution population is evaluated to obtain a fitness score;

[0089] An optimized frequency band allocation scheme is generated based on the fitness score, and the optimized frequency band allocation scheme is dynamically adjusted and discrete cosine transformed to obtain a spectrum allocation strategy.

[0090] Specifically, the spectrum demand prediction results are classified according to the frequency bands to distinguish the demand for high-frequency bands and low-frequency bands. Assume that the spectrum demand prediction result is a vector , where each Indicates The demand for a frequency band. Define a threshold To distinguish high frequency band from low frequency band, that is, when When The frequency band is classified as high frequency band, and vice versa as low frequency band. By classifying the frequency band demand, we get the high frequency band demand vector and low-band demand vector . According to the demand for high and low frequency bands, the initial frequency band allocation ratio is calculated to obtain the initial allocation plan. Based on the above calculation results, a linear programming model for the initial allocation plan is constructed. The objective function is set to maximize spectrum utilization, and the constraints include total spectrum resource restrictions and demand restrictions for each frequency band. The objective function can be expressed as:

[0091] ;

[0092] in, Indicates The efficiency coefficient of each frequency band, Indicates The constraints include:

[0093] ;

[0094] , ;

[0095] in, represents the total spectrum resource constraint. Based on these conditions, an optimization problem expression is generated. The optimization problem expression is solved using the simplex method to obtain a basic feasible solution set. The simplex method is a classic linear programming solution algorithm that searches for the optimal solution through step-by-step iteration. The basic feasible solution set is subjected to crossover and mutation operations to generate a candidate solution population. The crossover operation generates a new solution by combining some features of different solutions, while the mutation operation introduces diversity and novelty by randomly changing certain features. According to the preset fitness function, each solution in the candidate solution population is evaluated to obtain a fitness score. The fitness function usually conducts a comprehensive evaluation of each solution based on indicators such as spectrum utilization efficiency, interference level, and service quality. The higher the score, the better the solution. Based on the fitness score, the solution with the highest fitness is selected to generate an optimized frequency band allocation scheme. The optimized frequency band allocation scheme is dynamically adjusted and discrete cosine transformed. Dynamic adjustment fine-tunes the scheme through real-time monitoring and feedback mechanism to adapt to changes in spectrum demand. Discrete cosine transform is used to smooth the frequency band allocation data, reduce the impact of mutations, and improve the stability and robustness of the scheme. The discrete cosine transform formula is:

[0096] ;

[0097] in, Indicates Discrete cosine transform coefficients, Represents the original frequency band allocation data, Indicates the data length. The allocation data is converted to the frequency domain through discrete cosine transform, smoothed, and then inversely transformed back to the time domain.

[0098] In a specific embodiment, the process of executing step S103 may specifically include the following steps:

[0099] According to the spectrum allocation strategy, the received communication data is divided into frequency bands to obtain multiple frequency band data subsets, and frequency domain features are extracted from the multiple frequency band data subsets to obtain a feature vector set;

[0100] The feature vector set is input into the pre-trained convolutional neural network model for preliminary identification of the protocol type, and the probability distribution of the protocol type is obtained. The maximum a posteriori probability criterion analysis is performed on the probability distribution of the protocol type to obtain the initial identification result.

[0101] According to the initial recognition result, the corresponding protocol parsing rules are retrieved from the protocol template library to obtain a candidate parsing rule set, and the communication data is parsed in parallel based on each rule in the candidate parsing rule set to obtain multiple parsing results;

[0102] Conduct credibility evaluation on multiple parsing results to obtain credibility scores of each parsing result, and select the best parsing result based on the credibility scores to obtain the protocol field structure;

[0103] The protocol field structure is semantically analyzed to obtain semantically annotated data, and the semantically annotated data is associated and mapped with the communication data and a data dependency graph is constructed to obtain the data information to be converted.

[0104] Specifically, the communication data is classified according to a predetermined spectrum allocation strategy to obtain multiple frequency band data subsets. Assume that the received communication data is , and the spectrum allocation strategy divides it into frequency bands, and the data subsets for each frequency band are Each of these data subsets corresponds to a specific frequency band. The frequency band division makes the subsequent processing more targeted and efficient. Frequency domain feature extraction is performed on the frequency band data subset to capture the characteristics of the data in the frequency domain. Frequency domain feature extraction is usually achieved through fast Fourier transform. Fast Fourier transform converts time domain data into frequency domain data. The formula is as follows:

[0105] ;

[0106] in, represents the frequency domain characteristics, Represents time domain data, samples, represents the total number of data points, Represents the frequency index. Through fast Fourier transform, each frequency band data subset Converted to its frequency domain feature vector , and finally obtain the feature vector set The feature vector set Input the pre-trained convolutional neural network model to perform preliminary identification of the protocol type. The convolutional neural network extracts the deep features of the input data through multiple convolutional layers and pooling layers. Assume that the feature vector set The input is I, and the output of the first convolutional layer is , the formula is as follows:

[0107] ;

[0108] in, is the convolution kernel of the first convolutional layer, * represents the convolution operation, is the bias, is an activation function (such as ReLU). After the convolution layer extracts features, it is downsampled through the pooling layer to reduce the dimension of the feature map. After multiple convolution layers and pooling layers, the final feature representation is obtained, and the probability distribution of the protocol type is calculated through the fully connected layer and the Softmax layer. The formula of the Softmax function is:

[0109] ;

[0110] in, For the The probability of a protocol type, is the output of the fully connected layer elements. Through the Softmax layer, we get the probability distribution of the protocol type . Probability distribution of protocol types Perform maximum a posteriori probability criterion analysis and select the protocol type with the highest probability as the initial recognition result. Assume that the initial recognition result is , that is, the protocol type corresponding to the maximum probability. According to the initial recognition result , retrieve the corresponding protocol parsing rules from the protocol template library to obtain the candidate parsing rule set The protocol template library contains parsing rules for different protocols. By matching the initial recognition results, the corresponding parsing rules are screened out. Based on each rule in the candidate parsing rule set, the communication data is parsed in parallel to obtain multiple parsing results. Each parsing rule Both are used to parse raw communication data , generate the corresponding analysis results . Parallel processing can improve parsing efficiency, allowing multiple rules to act on data at the same time. Conduct credibility assessment to obtain the credibility score of each analysis result. The credibility assessment is based on the degree of match between the analysis result and the expected result, and is usually evaluated by calculating the accuracy, completeness and other indicators of the analysis result. Assume that the credibility score is Based on credibility score Select the best parsing result, that is, the parsing result with the highest credibility score, as the final protocol field structure. Assume that the best parsing result is The result contains the structured information of the communication data, such as field name, field type, and field value. Perform semantic analysis to obtain semantically annotated data. Semantic analysis generates semantically annotated data by analyzing the meaning and context of the field. . Semantically annotated data Communication data Perform association mapping and build a data dependency graph. The data dependency graph shows the dependency relationship and transfer path between fields, making the structure and relationship of the data clearer. Get the data information to be converted , which contains the structured and semantic information of the communication data.

[0111] In a specific embodiment, the execution step inputs the feature vector set into the pre-trained convolutional neural network model to perform preliminary identification of the protocol type, obtains the protocol type probability distribution, and performs maximum a posteriori probability criterion analysis on the protocol type probability distribution. The process of obtaining the initial identification result can specifically include the following steps:

[0112] Normalizing the feature vector set to obtain a standardized feature vector set;

[0113] The standardized feature vector set is input into the first convolutional layer of the pre-trained convolutional neural network model, and 32 3x3 convolution kernels are used for convolution operation. The first layer feature map is obtained by ReLU activation function and maximum pooling.

[0114] The first-layer feature map is processed by the second convolution layer, using 64 3x3 convolution kernels for convolution operation, and the second-layer feature map is obtained through the ReLU activation function and maximum pooling;

[0115] The second layer feature map is input into the third convolution layer, and 128 3x3 convolution kernels are used for convolution operation. The global feature vector is obtained by ReLU activation function and global average pooling.

[0116] The global feature vector is processed by a fully connected layer, using 256 neurons and ReLU activation function to obtain a high-dimensional feature representation;

[0117] The high-dimensional feature representation is input into the Dropout layer, and randomly set to zero with a probability of 0.5 to obtain the regularized feature. The regularized feature is then fully connected in the last layer, with the number of neurons equal to the number of protocol types, to obtain the original output vector.

[0118] The original output vector is input into the Softmax layer for exponential normalization to obtain the protocol type probability distribution. Based on the maximum a posteriori probability criterion, the category with the highest probability of the protocol type probability distribution is selected and compared with the preset confidence threshold to obtain the initial recognition result.

[0119] Specifically, the original feature vector set is standardized so that each eigenvalue falls within the same range, thereby obtaining a standardized feature vector set. Input the first convolutional layer of the pre-trained convolutional neural network model. The first convolutional layer uses 32 3x3 convolution kernels for convolution operation, and the formula is as follows:

[0120] ;

[0121] in, represents the first layer feature map, is the convolution kernel of the first convolutional layer, * represents the convolution operation, is the bias, is the ReLU activation function. The formula of the ReLU activation function is:

[0122] ;

[0123] The maximum pooling operation is applied between the convolution operation and the activation function. The pooling window is usually 2x2, which reduces the size of the feature map while retaining important features. Assume that the result of the pooling operation is still expressed as , then the first layer feature map is processed by ReLU activation and maximum pooling to obtain the final output. As input, the second convolution layer is processed. The second convolution layer uses 64 3x3 convolution kernels for convolution operation, and the formula is as follows:

[0124] ;

[0125] in, represents the second layer feature map, is the convolution kernel of the second convolutional layer, is the bias, activation function The same is ReLU. The maximum pooling operation is also applied to the convolution operation and activation function to obtain the second layer feature map . The second layer feature map As input, the third convolution layer is processed. The third convolution layer uses 128 3x3 convolution kernels for convolution operation, and the formula is as follows:

[0126] ;

[0127] in, represents the third layer feature map, is the convolution kernel of the third convolutional layer, is the bias, activation function Still ReLU. A global average pooling operation is applied after the third convolutional layer to average all values ​​of each feature map to obtain a global feature vector The formula for global average pooling is as follows:

[0128] ;

[0129] in, Indicates global eigenvectors, Indicates the third layer feature map The feature map is The value of the position, and Represent the height and width of the feature map respectively. The global feature vector Input the fully connected layer, use 256 neurons and ReLU activation function to get high-dimensional feature representation The formula for the fully connected layer is as follows:

[0130] ;

[0131] in, is the weight matrix of the fully connected layer, is the bias vector, the activation function Still ReLU. Represent high-dimensional features Enter the Dropout layer and randomly set it to zero with a probability of 0.5 to obtain the regularized features The purpose of Dropout is to prevent the model from overfitting and improve the generalization ability of the model by randomly discarding half of the neuron outputs. Perform the final fully connected processing, the number of neurons in the output layer is equal to the number of protocol types, and the original output vector is obtained The formula for the output layer is as follows:

[0132] ;

[0133] in, is the weight matrix of the output layer, is the bias vector. Input the Softmax layer for exponential normalization to obtain the probability distribution of protocol types The formula of the Softmax function is:

[0134] ;

[0135] in, Indicates The probability of a protocol type, Represents the first elements. Through the Softmax layer, the original output is converted into the probability distribution of each protocol type. According to the maximum a posteriori probability criterion, the category with the largest probability in the probability distribution of the protocol type is selected as the initial recognition result. Assume that the initial recognition result is , and compare it with the preset confidence threshold For comparison, only when the maximum probability Greater than The initial recognition result is accepted.

[0136] In a specific embodiment, the process of executing step S104 may specifically include the following steps:

[0137] Performing protocol field analysis on the data information to be converted to obtain a field feature set, and searching for matching rules from a preset protocol conversion rule library based on the field feature set to obtain an initial conversion rule set;

[0138] Calculate the applicability of each rule in the initial conversion rule set to obtain the rule weight distribution, and generate the adaptive protocol conversion rule based on the rule weight distribution;

[0139] Applying the adaptive protocol conversion rules to the data information to be converted, performing field mapping and conversion, obtaining a preliminary conversion result, and performing a semantic consistency check on the preliminary conversion result to obtain a conflict marker list;

[0140] According to the conflict marker list, the conflict field is adjusted to obtain the corrected conversion data, and the corrected conversion data is input into the encapsulation template of the target protocol to generate a protocol header and a checksum to obtain the communication data after cross-protocol conversion.

[0141] Specifically, the data is parsed to identify and extract the name, type and related characteristics of each protocol field to obtain a field feature set, which is recorded as , where each Represents the characteristics of a protocol field, including name, data type, length and other attributes. , retrieve matching rules from the preset protocol conversion rule library to obtain the initial conversion rule set The rule base contains a variety of protocol conversion rules, each rule They are defined for specific protocol fields and conversion requirements. By matching field characteristics, the initial conversion rule set applicable to the current data is selected. The suitability of each rule in the conversion scenario is calculated to determine the importance of each rule in the current conversion scenario. The suitability calculation can be evaluated based on factors such as field matching degree and historical application effect. Assume that the suitability score is , then the rule weight distribution Reflects the applicability and weight of each rule. Generates adaptive protocol conversion rules based on rule weight distribution , the formula is:

[0142] ;

[0143] in, Represents the adaptive protocol conversion rule. By weighting and summing each rule in the initial conversion rule set, an optimized rule set that adapts to the current data conversion requirements is obtained. Applied to the data information to be converted , perform field mapping and conversion to obtain preliminary conversion results . The process of field mapping and conversion involves mapping the fields of the source protocol to the fields of the target protocol, and performing necessary format and content conversion to obtain preliminary conversion results. For example, field A in the source protocol is converted to field B in the target protocol, and the data type is converted from integer to string. A semantic consistency check is performed on the preliminary conversion results to verify whether the converted data is semantically consistent with the original data, including field meaning, data range, contextual relationship, etc. After the check, a list of conflict markers is obtained , which lists all semantic conflicts found during the conversion process. Based on the conflict marker list , adjust the conflicting fields to resolve the semantic conflict and obtain the corrected conversion data The adjustment process may include remapping fields, modifying data content, adjusting data formats, and other operations. For example, if it is found that the data range of a field no longer meets the requirements of the target protocol after conversion, it is necessary to remap or adjust the data range. Input into the encapsulation template of the target protocol, generate the protocol header and checksum, and obtain the communication data after cross-protocol conversion The protocol header includes the control information required by the target protocol, such as source address, destination address, protocol version, etc. The checksum is used to verify the integrity of the data. The generation of the protocol header and checksum ensures that the converted data can be correctly identified and processed by the target protocol.

[0144] In a specific embodiment, the process of executing step S105 may specifically include the following steps:

[0145] Analyze the data volume and service type of the communication data after cross-protocol conversion to obtain a communication demand feature vector, and extract a candidate mode set from a preset MIMO mode library based on the communication demand feature vector to obtain an initial MIMO mode list;

[0146] Calculate the channel capacity of each MIMO mode in the initial MIMO mode list to obtain the theoretical throughput of each MIMO mode, and calculate the energy efficiency ratio of each MIMO mode based on the theoretical throughput to obtain the energy efficiency evaluation result;

[0147] Generate the optimal MIMO mode based on the energy efficiency evaluation results, perform an exhaustive search for the number of antennas in the optimal MIMO mode, calculate the signal-to-noise ratio of each configuration, and obtain a candidate antenna configuration set;

[0148] Optimizing the candidate antenna configuration set to obtain an optimized antenna configuration, and based on the optimized antenna configuration, calculating a precoding matrix and an equalization matrix to obtain a MIMO system matrix;

[0149] The MIMO system matrix is ​​eigen-decomposed to obtain a spatial stream mapping scheme, and the optimal MIMO mode, the optimized antenna configuration, the MIMO system matrix and the spatial stream mapping scheme are combined to obtain the MIMO transmission parameters.

[0150] Specifically, the communication data is decoded and analyzed to identify key features. Contains multiple data packets, each containing information such as the amount of data and service type. Through parsing, this information is extracted and a communication demand feature vector is generated. , which is in the form of .in, Indicates The characteristics of each data packet include data volume, service type, priority, etc. According to the communication demand feature vector , extract the candidate mode set from the preset MIMO mode library to obtain the initial MIMO mode list The MIMO mode library contains a variety of predefined MIMO modes, each of which is suitable for different communication requirements and channel conditions. By matching the communication requirement characteristics, the initial MIMO mode list suitable for the current communication requirements is screened out. The channel capacity is calculated for each MIMO mode in the initial MIMO mode list. The channel capacity indicates the maximum data transmission rate that the communication system can support under given channel conditions. The calculation formula for channel capacity is as follows:

[0151] ;

[0152] in, represents the channel capacity, represents the channel bandwidth, and SNR represents the signal-to-noise ratio. For each MIMO mode, calculate its theoretical throughput under the current channel conditions. , and calculate the energy efficiency ratio of each MIMO mode based on the theoretical throughput. The energy efficiency ratio indicates the amount of data that can be transmitted per unit energy consumption, and its calculation formula is as follows:

[0153] ;

[0154] in, It represents the energy efficiency ratio, represents the theoretical throughput, Represents energy consumption. By calculating the energy efficiency ratio of each MIMO mode, the energy efficiency evaluation result is obtained. According to the energy efficiency evaluation result E, the mode with the highest energy efficiency is selected to generate the optimal MIMO mode Perform an exhaustive search for the optimal number of antennas in MIIMO mode, calculate the signal-to-noise ratio (SNR) of each antenna configuration, and obtain the candidate antenna configuration set The signal-to-noise ratio is calculated as follows:

[0155] ;

[0156] in, represents the signal power, Represents the noise power. By exhaustively searching all possible antenna configurations, the signal-to-noise ratio under each configuration is calculated to screen out the candidate antenna configuration set. Optimize and select the configuration with the highest signal-to-noise ratio as the optimized antenna configuration Based on the optimized antenna configuration, the precoding matrix and equalization matrix are calculated to obtain the MIMO system matrix The precoding matrix is ​​used to preprocess the signal at the transmitting end to optimize the transmission effect of the signal in the multipath channel. The equalization matrix is ​​used to process the received signal at the receiving end to compensate for the distortion and interference caused by the channel. Assume that the precoding matrix is , the equilibrium matrix is , then the MIIMO system matrix It is expressed as:

[0157] ;

[0158] For the MIMO system matrix Perform eigendecomposition to obtain a spatial stream mapping scheme. Eigendecomposition represents a complex MIIMO channel as multiple independent subchannels, so that each subchannel can transmit data independently, thereby improving transmission efficiency. Assume that the result of eigendecomposition is:

[0159] ;

[0160] in, and are the left and right eigenvector matrices respectively, is the eigenvalue diagonal matrix. The spatial stream mapping scheme can be determined based on the eigenvectors and eigenvalues, and each eigenvalue corresponds to an independent spatial stream. , optimized antenna configuration , MIMO system matrix Combined with the spatial stream mapping scheme, the MIMO transmission parameters are obtained These parameters include optimized antenna configuration, channel precoding and equalization schemes, and spatial stream mapping strategies, which can ensure the best transmission performance under current communication requirements and channel conditions.

[0161] In a specific embodiment, the process of executing step S106 may specifically include the following steps:

[0162] Initialize the MIMO transmission parameters, generate the initial state of the MIMO system, and calculate the real-time channel state information based on the initial state of the MIMO system to obtain the channel matrix;

[0163] Decompose and calculate the channel matrix to obtain the channel quality index, and select the best modulation mode and coding rate according to the channel quality index to obtain the modulation and coding scheme;

[0164] Combining the modulation and coding scheme with the MIMO transmission parameters to generate a theoretical throughput prediction value, and calculating a performance difference index of the theoretical throughput prediction value;

[0165] Based on the performance difference index, the MIMO transmission parameters are adjusted and the performance is calculated to obtain the real-time performance index, and the real-time performance index is optimized by multiple objectives to obtain the optimal communication link configuration.

[0166] Specifically, the initial state of the MIMO system is generated according to system requirements and physical resource configuration. Assuming that there are antennas and Represent the number of antennas at the transmitting and receiving ends respectively. The initial state of the MIMO system can be expressed as:

[0167] ;

[0168] in, Indicates The receiving antenna and Channel gain between the transmitting antennas. Initialization state is a The matrix represents the initial channel state. Based on the initial state of the MIMO system, the channel state information is calculated in real time to obtain the channel matrix The channel state information can be obtained through channel estimation technology, for example, using pilot signals to estimate the channel gain. The real-time channel state information reflects the dynamic changes of the channel and can be expressed as:

[0169] ;

[0170] in, Represents the change of the channel over time. Decompose and calculate to obtain the channel quality index. A common method is to perform singular value decomposition (SVD) to decompose the channel matrix into the product of three matrices:

[0171] in, and are the unit orthogonal matrices, It is a diagonal matrix whose diagonal elements are the singular values ​​of the channel. Channel quality indicators can be determined by singular values. For example, the channel capacity can be expressed as:

[0172] ;

[0173] in, It is singular values, SNR is the signal-to-noise ratio. According to the channel quality index, the optimal modulation mode and coding rate are selected to obtain the modulation and coding scheme. The modulation mode and coding rate directly affect the performance of the communication system. Common modulation modes include QPSK, 16QAM and 64QAM, etc. Different coding rates correspond to different error correction capabilities. The goal of selecting a modulation and coding scheme is to maximize the data transmission rate while ensuring that the bit error rate is within an acceptable range. The modulation and coding scheme is combined with the MIMO transmission parameters to generate a theoretical throughput prediction value. The theoretical throughput prediction value can be expressed as:

[0174] ;

[0175] in, is the bandwidth, is the modulation efficiency, is the channel capacity. The theoretical throughput prediction value reflects the maximum data transmission rate that the system can achieve under the current channel conditions. Calculate the performance difference index of the theoretical throughput prediction value. The performance difference index is used to measure the gap between the theoretical throughput and the actual throughput. The formula is as follows:

[0176] ;

[0177] in, is the actual throughput, obtained by real-time monitoring of system performance. The smaller the performance difference index, the closer the system performance is to the theoretical prediction value. Based on the performance difference index, the MIMO transmission parameters are adjusted and the performance is calculated. The adjustment process may include modifying the antenna configuration, modulation mode, and coding rate, etc. The goal is to minimize the performance difference index and improve the actual performance of the system. The adjusted MIMO transmission parameters can be expressed as , which includes the optimized antenna configuration, modulation and coding scheme, etc. After obtaining the adjusted MIMO transmission parameters, the real-time performance index calculation is performed. The real-time performance index includes throughput, delay, bit error rate, etc., through which the system performance can be comprehensively evaluated. The performance calculation formula is as follows:

[0178] ;

[0179] in, It is Packets at time The amount of data, is the total time. Multi-objective optimization is performed on real-time performance indicators to obtain the optimal communication link configuration. The goal of multi-objective optimization is to optimize multiple performance indicators at the same time to make the system reach the best state. The optimization process can use genetic algorithms, particle swarm optimization and other methods to finally obtain the optimal communication link configuration. .

[0180] The above describes the 5G protocol and line short compatible intercommunication method in the embodiment of the present application. The following describes the 5G protocol and line short compatible intercommunication system in the embodiment of the present application. Please refer to Figure 2 In the embodiment of the present application, an embodiment of the 5G protocol and line short compatible intercommunication system includes:

[0181] The collection module 201 is used to collect network data through a distributed sensor network, and to collect and analyze the network data in real time to obtain spectrum demand prediction results;

[0182] A calculation module 202 is used to calculate the high and low frequency band allocation ratios based on the spectrum demand prediction result to obtain a spectrum allocation strategy;

[0183] The parsing module 203 is used to perform protocol identification and parsing on the communication data based on the spectrum allocation strategy to obtain data information to be converted;

[0184] A reconstruction module 204 is used to reconstruct and encapsulate the data information to be converted based on the adaptive protocol conversion rules to obtain the communication data after cross-protocol conversion;

[0185] The optimization module 205 is used to select the MIMO mode and optimize the antenna configuration of the communication data after the cross-protocol conversion to obtain the MIMO transmission parameters;

[0186] The adjustment module 206 is used to dynamically adjust the MIMO transmission parameters and perform performance evaluation to obtain the optimal communication link configuration.

[0187] Through the collaboration of the above components, the distributed sensor network and the multi-layer perceptron neural network model can accurately predict the spectrum demand and improve the utilization efficiency of spectrum resources. Based on the adaptive protocol conversion rules, flexible conversion between different protocols is realized, solving the intercommunication problem between 5G network and industry-specific network. Through MIMO mode selection and antenna configuration optimization, the transmission performance and energy efficiency of the system are improved. Real-time optimization of communication links is achieved, and the system's adaptability to complex network environments is enhanced. The convolutional neural network model is used for protocol type identification, which improves the accuracy and efficiency of protocol identification. Through linear programming and genetic algorithms, the spectrum allocation strategy is dynamically optimized and the utilization efficiency of spectrum resources is improved. Through adaptive protocol conversion rules, accurate reconstruction and encapsulation of data are achieved, ensuring the reliability of cross-protocol communication. Through multi-objective optimization methods, performance evaluation and dynamic adjustment of communication links are achieved, and seamless conversion between different protocols is achieved.

[0188] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described systems, systems and units can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0189] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions to enable a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk, etc., various media that can store program codes.

[0190] As described above, the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A 5G protocol and line short compatibility interoperability method, characterized in that: The method comprises: The network data is collected through a distributed sensor network, and the network data is collected and analyzed in real time to obtain a spectrum demand prediction result; specifically, the method includes: allocating data collection tasks to each node in the distributed sensor network to obtain a node task list, and according to the node task list, real-time monitoring of the signal strength, user density and data flow of each node to obtain network data; performing data cleaning and standardization processing on the network data to obtain a pre-processed network data set, and performing time series segmentation on the pre-processed network data set to obtain multiple time series data subsets; performing fast Fourier transform on the time series data subset to obtain a spectrum feature vector; normalizing the spectrum feature vector to obtain a normalized feature vector, and using the normalized feature vector as input data of an input layer in a multi-layer perceptron neural network model; performing activation function calculation on input layer neurons according to the normalized feature vector to obtain an input layer Output vector, and perform matrix multiplication operation on the input layer output vector and the first hidden layer weight matrix to obtain the linear combination result of the first hidden layer; perform ReLU activation function operation on the linear combination result of the first hidden layer to obtain the activation output of the first hidden layer, and perform matrix multiplication operation on the activation output of the first hidden layer and the second hidden layer weight matrix to obtain the linear combination result of the second hidden layer; perform Sigmoid activation function operation on the linear combination result of the second hidden layer to obtain the activation output of the second hidden layer, and perform matrix multiplication operation on the activation output of the second hidden layer and the output layer weight matrix to obtain the linear combination result of the output layer; perform Softmax activation function operation on the linear combination result of the output layer to obtain the probability distribution of demand for each frequency band; according to the probability distribution of demand for each frequency band, select high probability frequency bands through threshold screening method and perform exponential moving average smoothing processing to obtain spectrum demand prediction results; Calculate the high and low frequency band allocation ratios based on the spectrum demand prediction result to obtain a spectrum allocation strategy; Based on the spectrum allocation strategy, protocol identification and analysis are performed on the communication data to obtain data information to be converted; Reconstructing and encapsulating the data information to be converted based on the adaptive protocol conversion rules to obtain communication data after cross-protocol conversion; Performing MIMO mode selection and antenna configuration optimization on the communication data after the cross-protocol conversion to obtain MIMO transmission parameters; The MIMO transmission parameters are dynamically adjusted and performance evaluated to obtain an optimal communication link configuration.

2. The 5G protocol and line short compatibility intercommunication method according to claim 1 is characterized in that: The calculating the high and low frequency band allocation ratios based on the spectrum demand prediction result to obtain a spectrum allocation strategy includes: Classifying the spectrum demand prediction result by frequency bands to obtain demand for high frequency bands and low frequency bands; According to the demand for the high frequency band and the low frequency band, an initial frequency band allocation ratio is calculated to obtain an initial allocation plan; Constructing the objective function and constraint conditions of the initial allocation scheme based on the linear programming model, and generating an optimization problem expression; Solving the optimization problem expression by using the simplex method to obtain a basic feasible solution set, and performing crossover and mutation operations on the basic feasible solution set to obtain a candidate solution population; According to a preset fitness function, each solution in the candidate solution population is evaluated to obtain a fitness score; An optimized frequency band allocation scheme is generated based on the fitness score, and the optimized frequency band allocation scheme is dynamically adjusted and discrete cosine transformed to obtain a spectrum allocation strategy.

3. The 5G protocol and line short compatibility intercommunication method according to claim 1 is characterized in that: The step of performing protocol identification and parsing on the communication data based on the spectrum allocation strategy to obtain data information to be converted includes: According to the spectrum allocation strategy, the received communication data is divided into frequency bands to obtain a plurality of frequency band data subsets, and frequency domain features are extracted from the plurality of frequency band data subsets to obtain a feature vector set; Inputting the feature vector set into a pre-trained convolutional neural network model to perform preliminary identification of the protocol type, obtaining a probability distribution of the protocol type, and performing a maximum a posteriori probability criterion analysis on the probability distribution of the protocol type to obtain an initial identification result; According to the initial recognition result, the corresponding protocol parsing rule is retrieved from the protocol template library to obtain a candidate parsing rule set, and the communication data is parsed in parallel based on each rule in the candidate parsing rule set to obtain multiple parsing results; Performing credibility evaluation on the multiple parsing results to obtain a credibility score for each parsing result, and selecting an optimal parsing result based on the credibility score to obtain a protocol field structure; The protocol field structure is semantically analyzed to obtain semantically annotated data, and the semantically annotated data is associated and mapped with the communication data and a data dependency graph is constructed to obtain data information to be converted.

4. The 5G protocol and line short compatibility intercommunication method according to claim 3 is characterized in that: The feature vector set is input into a pre-trained convolutional neural network model to perform preliminary identification of the protocol type, obtain a probability distribution of the protocol type, and perform maximum a posteriori probability criterion analysis on the probability distribution of the protocol type to obtain an initial identification result, including: Normalizing the feature vector set to obtain a standardized feature vector set; Input the standardized feature vector set into the first convolution layer of the pre-trained convolutional neural network model, perform convolution operation using 32 3x3 convolution kernels, and obtain the first layer feature map through ReLU activation function and maximum pooling; Perform the second convolutional layer processing on the first layer feature map, use 64 3x3 convolution kernels to perform convolution operation, and use ReLU activation function and maximum pooling to obtain the second layer feature map; The second layer feature map is input into the third convolutional layer, and 128 3x3 convolution kernels are used for convolution operation, and the global feature vector is obtained through ReLU activation function and global average pooling. Performing fully connected layer processing on the global feature vector, using 256 neurons and ReLU activation function, to obtain a high-dimensional feature representation; The high-dimensional feature representation is input into the Dropout layer, and randomly set to zero with a probability of 0.5 to obtain a regularized feature, and the regularized feature is subjected to a final layer of full connection processing, with the number of neurons being equal to the number of protocol types, to obtain an original output vector; The original output vector is input into the Softmax layer for exponential normalization to obtain the protocol type probability distribution, and the category with the highest probability of the protocol type probability distribution is selected based on the maximum a posteriori probability criterion, and compared with the preset confidence threshold to obtain the initial recognition result.

5. The 5G protocol and line short compatibility intercommunication method according to claim 1 is characterized in that: The data information to be converted is reconstructed and encapsulated based on the adaptive protocol conversion rule to obtain the communication data after cross-protocol conversion, including: Performing protocol field analysis on the data information to be converted to obtain a field feature set, and retrieving matching rules from a preset protocol conversion rule library based on the field feature set to obtain an initial conversion rule set; Calculating the applicability of each rule in the initial conversion rule set to obtain a rule weight distribution, and generating an adaptive protocol conversion rule based on the rule weight distribution; Applying the adaptive protocol conversion rule to the data information to be converted, performing field mapping and conversion, obtaining a preliminary conversion result, and performing a semantic consistency check on the preliminary conversion result to obtain a conflict marker list; According to the conflict marker list, the conflict field is adjusted to obtain the corrected conversion data, and the corrected conversion data is input into the encapsulation template of the target protocol to generate a protocol header and a checksum to obtain the communication data after cross-protocol conversion.

6. The 5G protocol and line short compatibility intercommunication method according to claim 1 is characterized in that: The performing MIMO mode selection and antenna configuration optimization on the communication data after the cross-protocol conversion to obtain MIMO transmission parameters includes: Performing data volume and service type analysis on the communication data after the cross-protocol conversion to obtain a communication demand feature vector, and extracting a candidate mode set from a preset MIMO mode library according to the communication demand feature vector to obtain an initial MIMO mode list; Calculating the channel capacity of each MIMO mode in the initial MIMO mode list to obtain a theoretical throughput of each MIMO mode, and calculating an energy efficiency ratio of each MIMO mode based on the theoretical throughput to obtain an energy efficiency evaluation result; Generate an optimal MIMO mode according to the energy efficiency evaluation result, perform an exhaustive search on the number of antennas in the optimal MIMO mode, calculate the signal-to-noise ratio of each configuration, and obtain a candidate antenna configuration set; Optimizing the candidate antenna configuration set to obtain an optimized antenna configuration, and calculating a precoding matrix and an equalization matrix based on the optimized antenna configuration to obtain a MIMO system matrix; The MIMO system matrix is ​​eigen-decomposed to obtain a spatial stream mapping scheme, and the optimal MIMO mode, the optimized antenna configuration, the MIMO system matrix and the spatial stream mapping scheme are combined to obtain MIMO transmission parameters.

7. The 5G protocol and line short compatibility intercommunication method according to claim 1 is characterized in that: The dynamically adjusting and performance evaluating the MIMO transmission parameters to obtain an optimal communication link configuration includes: Initializing the MIMO transmission parameters to generate an initial state of the MIMO system, and based on the initial state of the MIMO system, calculating real-time channel state information to obtain a channel matrix; Decomposing and calculating the channel matrix to obtain a channel quality index, and selecting an optimal modulation mode and coding rate according to the channel quality index to obtain a modulation and coding scheme; Combining the modulation and coding scheme with the MIMO transmission parameters to generate a theoretical throughput prediction value, and calculating a performance difference indicator of the theoretical throughput prediction value; Based on the performance difference index, the MIMO transmission parameters are adjusted and the performance is calculated to obtain a real-time performance index, and the real-time performance index is multi-objective optimized to obtain an optimal communication link configuration.

8. A 5G protocol and line short compatible intercommunication system, characterized in that: A system for executing the 5G protocol and line short compatibility interworking method according to any one of claims 1 to 7, comprising: The acquisition module is used to collect network data through a distributed sensor network, and to collect and analyze the network data in real time to obtain spectrum demand prediction results; specifically, it includes: allocating data acquisition tasks to each node in the distributed sensor network to obtain a node task list, and according to the node task list, real-time monitoring of the signal strength, user density and data flow of each node to obtain network data; performing data cleaning and standardization processing on the network data to obtain a preprocessed network data set, and performing time series segmentation on the preprocessed network data set to obtain multiple time series data subsets; performing fast Fourier transform on the time series data subsets to obtain a spectrum feature vector; normalizing the spectrum feature vector to obtain a normalized feature vector, and using the normalized feature vector as input data of the input layer in the multi-layer perceptron neural network model; according to the normalized feature vector, performing activation function calculation on the input layer neurons to obtain Output vector to the input layer, and perform matrix multiplication operation on the input layer output vector and the first hidden layer weight matrix to obtain the linear combination result of the first hidden layer; perform ReLU activation function operation on the linear combination result of the first hidden layer to obtain the activation output of the first hidden layer, and perform matrix multiplication operation on the activation output of the first hidden layer and the second hidden layer weight matrix to obtain the linear combination result of the second hidden layer; perform Sigmoid activation function operation on the linear combination result of the second hidden layer to obtain the activation output of the second hidden layer, and perform matrix multiplication operation on the activation output of the second hidden layer and the output layer weight matrix to obtain the linear combination result of the output layer; perform Softmax activation function operation on the linear combination result of the output layer to obtain the probability distribution of demand for each frequency band; according to the probability distribution of demand for each frequency band, select high probability frequency bands through threshold screening method and perform exponential moving average smoothing processing to obtain spectrum demand prediction results; A calculation module, used to calculate the high and low frequency band allocation ratios based on the spectrum demand prediction result to obtain a spectrum allocation strategy; A parsing module, used to perform protocol identification and parsing on the communication data based on the spectrum allocation strategy to obtain data information to be converted; A reconstruction module, used to reconstruct and encapsulate the data information to be converted based on the adaptive protocol conversion rules to obtain communication data after cross-protocol conversion; An optimization module, used to perform MIMO mode selection and antenna configuration optimization on the communication data after the cross-protocol conversion to obtain MIMO transmission parameters; The adjustment module is used to dynamically adjust the MIMO transmission parameters and perform performance evaluation to obtain the optimal communication link configuration.

Images