Optimization method of high-performance 5G communication module

Abstract

The invention relates to the technical field of 5G communication, in particular to an optimization method of a high-performance 5G communication module. Comprising the following steps: acquiring key operation parameters in a multi-carrier concurrent state, and constructing a characteristic graph model reflecting a parameter evolution relationship; mining a state clustering mode through a graph polarization spectral clustering algorithm, and identifying key paths and core parameter nodes which influence communication performance in combination with a multi-hop path optimization algorithm of a graph attention mechanism; taking the key path strategy graph as input, combining target communication scene features, and adopting a graph structure alignment meta-learning algorithm to generate an optimal parameter configuration strategy; the strategy is applied to an actual module, a self-adaptive control model fusing fuzzy reasoning and reinforcement learning is constructed, and dynamic evaluation of the strategy execution effect and fine parameter tuning are achieved. According to the invention, the performance scheduling efficiency of the 5G communication module under the complex network condition is improved, and the intelligence, generalization ability and scene adaptability of the system are obviously enhanced.

Description

Technical Field

[0001] The present invention relates to the field of 5G communication technology, and in particular to an optimization method for a high-performance 5G communication module. Background Art

[0002] With the rapid development of 5G, driven by key features such as large-scale multi-input multi-output, ultra-dense networking, and multi-carrier concurrency, it is widely used in high-reliability, low-latency scenarios such as smart manufacturing, Internet of Vehicles, and telemedicine. As the core component for terminal access and data transmission, the performance optimization of 5G communication modules is of great significance to ensuring system stability and throughput efficiency. With the widespread application of multi-carrier aggregation, ultra-dense deployment, and heterogeneous network environments, 5G communication modules often face problems such as redundant parameter configuration, inefficient resource scheduling, and unpredictable performance fluctuations during operation. Traditional optimization methods often rely on rule-driven or static modeling, making it difficult to achieve efficient generalization and dynamic scheduling in complex and changing communication environments. The following problems still exist: in an environment of multi-carrier concurrency, high-frequency interference and rapid channel changes, the operating status of the communication module is highly complex, and traditional methods cannot effectively extract the potential synergistic relationship between parameters; existing optimization strategies mostly rely on static analysis, which makes it difficult to discover multi-hop paths and their key parameter nodes that have a significant impact on communication performance, limiting the potential for deep optimization of system performance; traditional optimization methods often rely on empirical rules or fixed models in specific communication scenarios, and cannot adapt to load changes or environmental switches, resulting in the failure of strategies in different scenarios. Summary of the Invention

[0003] To solve the above problems, the present invention provides an optimization method for a high-performance 5G communication module, which solves the problem of how to effectively extract the collaborative relationship between the operating parameters of the 5G communication module in an environment of multi-carrier concurrency, high-frequency interference and complex heterogeneous network, identify key paths and parameter nodes, and realize efficient generalized scheduling and dynamic optimization across scenarios, thereby improving the performance scheduling efficiency of the 5G communication module under complex network conditions.

[0004] To achieve the above object, the technical solution adopted by the present invention is: A method for optimizing a high-performance 5G communication module comprises the following steps: S1: Obtain the operating parameters of the 5G communication module in a multi-carrier concurrent state; S2: Based on the operating parameters, a graph polarization spectral clustering algorithm is used to cluster the state graph and construct a feature graph model. The nodes in the graph represent the operating parameter states, and the edge weights represent the degree of coordinated changes between the parameters. S3: Based on the feature graph model, a multi-hop path optimization algorithm with a graph attention mechanism is used to jointly model and score the node weights in the path, identify high-weight paths and key parameter nodes that affect communication performance, and generate a critical path strategy graph that includes node configuration relationships and path control logic; S4: Taking the critical path strategy graph as input, based on the scenario characteristics in a multi-communication environment, a meta-learning algorithm based on graph structure alignment is adopted to build a strategy graph migration model. Through the mapping function, generalized scheduling and optimal migration of parameter configuration under different load conditions are achieved, and the optimal parameter configuration strategy is generated. S5: Apply the optimal parameter configuration strategy to the actual operation of the 5G communication module, collect the real-time operating status of the module as the fuzzy input variable, build a reinforced fuzzy reasoning network model, evaluate the deviation and adaptability of the current execution strategy through fuzzy reasoning rules combined with the reinforcement learning reward mechanism, and perform parameter fine-tuning based on performance feedback.

[0005] Furthermore, the operating parameters include transmit power, intermodulation interference, crystal oscillator offset, voltage-current response, board-level temperature and instantaneous throughput.

[0006] Furthermore, step S2 includes the following steps: Based on the operating parameters, parameter normalization and time window segmentation strategies are adopted to construct a time series state diagram reflecting the parameter evolution process. The nodes represent the parameter states in different time slices, and the edge weights represent the coordinated change amplitude between adjacent parameter states. By calculating the correlation coefficient matrix for each parameter node pair in the time series state diagram and combining the maximum information coefficient to enhance the nonlinear collaborative characteristics, a weighted multi-attribute state map is generated; Based on the weighted multi-attribute state graph, a graph polarization spectral clustering algorithm is used to perform spectral decomposition and feature space mapping on the node set. By introducing a polarization operator to adjust the spectral clustering boundary, high-noise and weakly coupled parameter states are clustered, and parameter state cluster labels are output; Based on the parameter state clustering labels, the core nodes and edge weight distribution in each subgraph are extracted, a feature graph model including node features, edge collaboration weights and clustering labels is constructed, and a structured feature graph model is output.

[0007] Furthermore, the formula for constructing the feature graph model is as follows: ; in, Indicates the co-evolution strength between operating states in the constructed feature graph; and Represent the normalized operating parameter vectors in the i-th and j-th time windows respectively; represents the squared Euclidean distance of the multi-parameter states between the i-th and j-th time slices; represents the maximum information coefficient; Indicates the local curvature or abnormally drastic change points in the parameter change map; represents the global variance between the states of the operating parameters; represents the normalized scale factor for the maximum information coefficient; 、 and They represent the weight coefficients of the Euclidean distance term, nonlinear correlation term, and spectral perturbation term, respectively.

[0008] Furthermore, step S3 includes the following steps: Based on the feature graph model, a multi-hop path optimization algorithm based on the graph attention mechanism is adopted to jointly model the state characteristics and edge collaboration relationships of nodes in the multi-hop adjacency range of the graph, and construct a multi-hop path structure with cross-level information perception capabilities; In the multi-hop path structure, the attention mechanism is used to dynamically weight the edge weight distribution and state feature differences between nodes within different hop count ranges to generate a node attention distribution graph; Based on the node attention distribution graph, all reachable multi-hop paths are scored, and the attention weight, edge coordination strength, and hop count of each node in the path are comprehensively considered to screen out a set of high-weight paths with significant path scores. Based on the high-weight path set, the key parameter nodes in the path are extracted, their position roles in the path structure and their dependencies with other nodes are analyzed, and a key path strategy diagram integrating regulatory logic and priority relationships is constructed.

[0009] Furthermore, the formula of the multi-hop path optimization algorithm is as follows: ; in, represents the multi-hop path score from node p to node q, which is used to construct the critical path strategy graph; H represents the maximum hop limit, which controls the hierarchical depth of path modeling; represents the global adjustment factor of the path of the h-th hop; represents the set of h-th hop adjacent nodes of the current node p; represents the attention weight; represents the edge coordination weight, which indicates the degree of coordinated change between operating parameters; It represents the instantaneous throughput, which is used to measure the dynamic impact of the node on the communication performance in the h-th hop layer; Indicates the number of hops in the path topology; represents the board-level temperature standard deviation of the nodes associated with node p within the hth hop; Indicates the hop count attenuation factor.

[0010] Furthermore, step S4 includes the following steps: Based on the critical path strategy graph, extract the high-weight path sequence, key parameter node configuration combination and the control dependency relationship between the paths, construct a graph structure sample set associated with the specific communication scenario, and form a source scenario knowledge graph; In the target communication environment, the system combines the real-time characteristics of network load level, interference source distribution, and terminal connection density to construct a graph structure representation of the current scenario and initialize the scenario label and path control demand vector. A meta-learning algorithm based on graph structure alignment is used to perform structural mapping between the knowledge graph in the source scene and the target scene graph, and a fast parameter adaptation mechanism is used to generate the optimal path control mapping function in the target scene. Based on the control mapping function, the key path strategy graph in the source scenario is mapped to the target communication scenario, the structural migration of the key parameter configuration relationship is achieved, and the optimal parameter configuration strategy is output.

[0011] Furthermore, the formula of the optimal path control mapping function is as follows: ; in, represents the optimal path control mapping function under the target communication scenario R; N represents the number of source scenario samples participating in the control learning; represents the key path parameter vector extracted from the nth source scene; Indicates the optimal parameter configuration result expected to be generated by the nth sample in the target scenario; represents the critical path strategy graph structure constructed in the nth source communication scenario; R represents the scenario graph structure in the target communication environment; represents the path regulation mapping function; represents the regularization term of function J; represents the regularization coefficient.

[0012] Furthermore, the enhanced fuzzy reasoning network model is specifically based on the core parameter nodes in the critical path strategy diagram as input dimensions, collects status indicators of transmission power offset, voltage and current fluctuations, crystal oscillator frequency deviation and temperature change, constructs a fuzzy input vector, and completes fuzzification processing through membership functions; combines the node dependency relationship in the strategy diagram to generate a fuzzy rule base, and introduces a reinforcement learning mechanism to dynamically adjust the fuzzy rule weights and execution strategy priorities.

[0013] Furthermore, the reinforcement learning reward mechanism is specifically constructed based on the dynamic feedback of communication performance indicators, including the throughput improvement value, bit error rate reduction and switching delay convergence rate as core reward factors, and combining the weighted functions of each indicator to generate a real-time performance score.

[0014] The beneficial effects of the present invention are: The present invention clusters the operating status of the 5G communication module through the graph polarization spectrum clustering algorithm, effectively reveals the synergistic relationship between parameters, lays a structural foundation for subsequent optimization, and improves the interpretability and structural perception ability of the internal state of the communication module. By using the multi-hop path optimization algorithm of the graph attention mechanism, the weights of the parameter nodes in the path are jointly modeled, which can accurately identify the paths and key nodes that have the greatest impact on communication performance, provide a high-reliability decision-making basis for strategy generation, and significantly improve the accuracy and controllability of path regulation. By introducing a meta-learning algorithm for graph structure alignment, a strategy graph model that can be migrated in multiple communication environments is constructed, which solves the problems of poor strategy adaptability and inability to generalize in traditional methods, realizes efficient migration and dynamic scheduling of parameter configuration, and improves the robustness and adaptability of the system in complex network environments. In actual deployment, a feedback optimization model is jointly constructed based on fuzzy reasoning and reinforcement learning mechanisms, which can dynamically perceive slight changes in communication status and adjust parameters in real time to avoid performance degradation or strategy failure, and improve the stability and continuous optimization capability of the overall operation of the system. The present invention integrates intelligent technologies such as graph neural computing, meta-learning and enhanced fuzzy control to build a closed-loop optimization system from state perception, strategy generation to dynamic execution, significantly enhancing the high-performance operation capabilities of 5G communication modules in multi-carrier and multi-scenarios. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 It is a flow chart of an optimization method for a high-performance 5G communication module of the present invention.

[0016] Figure 2 It is a flowchart of step S3 provided by one embodiment of the present invention.

[0017] Figure 3 It is a flowchart of step S4 provided in one embodiment of the present invention. DETAILED DESCRIPTION

[0018] See also Figure 1-3 As shown, the present invention relates to an optimization method for a high-performance 5G communication module.

[0019] Example A method for optimizing a high-performance 5G communication module comprises the following steps: S1: Obtain operating parameters of the 5G communication module in a multi-carrier concurrent state; the operating parameters include transmit power, intermodulation interference, crystal oscillator offset, voltage-current response, board-level temperature, and instantaneous throughput.

[0020] In one embodiment, a 5G communication module is deployed on an experimental test platform, and a base station simulator supporting NR (New Radio) multi-carrier aggregation is configured. A scheduling tool is used to trigger multiple carriers to be active simultaneously. During module operation, the following parameter data is collected: Transmit power: obtained by sampling through the power detection module of the RF transceiver chip; Intermodulation interference: Use spectrum analyzer to monitor in real time in the nonlinear mixing frequency band; Crystal oscillator offset: Use a high-precision clock offset measurement unit to collect the difference between the crystal oscillator frequency and the reference frequency; Voltage-current response: The PMU module (power management unit) collects the response curve of the supply voltage and current as the number of carriers enabled changes; Board-level temperature: Thermistor arrays are used to monitor the temperature distribution of key areas of the PCB; Instantaneous throughput: The actual data transmission volume per millisecond TTI is extracted through MAC layer scheduling feedback information.

[0021] The data cycle is 10ms, and the continuous acquisition time is 3 minutes, forming an operating parameter data set with high time resolution.

[0022] S2: Based on the operating parameters, a graph polarization spectral clustering algorithm is used to cluster the state graph and construct a feature graph model. The nodes in the graph represent the operating parameter states, and the edge weights represent the degree of coordinated changes between the parameters. Wherein, the step S2 includes the following steps: Based on the operating parameters, parameter normalization and time window segmentation strategies are adopted to construct a time series state diagram reflecting the parameter evolution process. The nodes represent the parameter states in different time slices, and the edge weights represent the coordinated change amplitude between adjacent parameter states. Specifically, all operating parameters are processed into uniform intervals, such as normalizing each indicator to the interval [0, 1]. A maximum-minimum scaling method is used, and the timestamps corresponding to the original values ​​are retained to facilitate restoration. Sliding time windows are set (for example, a 2-second window, sliding every 500ms). Within each window, the current value of each operating parameter is extracted to form a parameter state vector, which serves as a node in the graph. Each node in the graph represents the parameter state within a specific time window. Adjacent time windows are connected by edges, and the initial edge weight is the inverse Euclidean distance metric of the parameter vectors (i.e., the smaller the difference, the greater the edge weight).

[0023] By calculating the correlation coefficient matrix for each parameter node pair in the time series state diagram and combining the maximum information coefficient to enhance the nonlinear collaborative characteristics, a weighted multi-attribute state map is generated; Specifically, for all operating parameters within each time window, a state vector set is constructed; the Pearson correlation coefficient (a measure of linear synchronization) between different parameter pairs is calculated separately; for example, current fluctuations are often positively correlated with temperature rise trends; and a parameter correlation matrix is ​​generated to provide a basis for the initial edge weights of the state diagram.

[0024] For parameter relationships that cannot be explained by linear functions (such as crystal oscillator frequency deviation and bit error rate mutation), the maximum information coefficient (MIC) is used to measure their correlation; MIC can capture nonlinear patterns, such as step changes or exponential trends, and improve the state modeling capabilities of complex scenarios; ultimately, the edge weight between each pair of nodes is coupled by linear correlation and nonlinear information entropy, giving the edge a multi-attribute label.

[0025] Each node is bound to the following: parameter state value vector (such as the six operating parameter values ​​at a certain moment); timestamp label (for time sequence sorting); physical module ID (such as PA, LO, power management module); Each edge is bound to the following: synergy weight (calculated based on correlation and MIC); parameter type label pair (such as temperature-current); evolution direction label (whether it is in the rising or falling range of parameter change); Finally, a multi-dimensional weighted, time-directed state map is formed, laying a precise foundation for subsequent graph analysis.

[0026] Based on the weighted multi-attribute state graph, a graph polarization spectral clustering algorithm is used to perform spectral decomposition and feature space mapping on the node set. By introducing a polarization operator to adjust the spectral clustering boundary, high-noise and weakly coupled parameter states are clustered, and parameter state cluster labels are output; Specifically, the normalized Laplace matrix is ​​used to perform feature spectrum decomposition on the entire graph; the first K principal component vectors of the graph are extracted, and the high-dimensional node states are projected into the low-dimensional spectral space; the Euclidean distance between the low-dimensional vectors reflects the structural similarity between the original states.

[0027] In traditional spectral clustering, cluster boundaries are easily blurred due to high noise interference. The system introduces a polarization operator to adjust the cluster center and boundary weights; the polarization operator gives boundary nodes lower participation rights based on multiple indicators such as node local density, edge strength change rate, and state mutation index; for those state nodes with strong noise and violent fluctuations but no stable pattern, independent subcategories are separated or soft clustering is performed first.

[0028] Each node is accurately labeled as a cluster (e.g., Cluster-3, Cluster-7). Nodes in the same state exhibit a typical change pattern (e.g., "low power + high temperature + increased frequency deviation"). The clustering results are visualized as multiple color-coded subgraphs to distinguish different operating behavior patterns and abnormal operating condition trajectories.

[0029] Based on the parameter state clustering labels, the core nodes and edge weight distribution in each subgraph are extracted, a feature graph model including node features, edge collaboration weights and clustering labels is constructed, and a structured feature graph model is output.

[0030] Specifically, in each cluster subgraph, representative core nodes are selected based on indicators such as node degree centrality, edge weight mean, and time stability; these core nodes correspond to key control point parameters in actual operation, such as "temperature peak" and "power jump point".

[0031] Statistics on the collaborative weight distribution of all edges in each subgraph include the following: Edge weight mean: represents the overall coupling strength within this type of state; Edge weight variance: reflects the stability of the relationship between parameters within this type of state; Maximum path weight: Identify the most coupled paths in the subgraph for subsequent path optimization strategies.

[0032] Each cluster subgraph is encapsulated as a structure, which includes the following: Cluster labels (identifying subgraph categories); Core node index (such as nodes 3, 8, and 15); Node vector (parameter value + module location information); Edge connectivity matrix (recording the existence of edges and their weights); All subgraphs jointly constitute the final feature graph model library, which serves as the input basis for subsequent graph attention mechanism modeling and path recognition.

[0033] Furthermore, the formula for constructing the feature graph model is as follows: ; in, Indicates the co-evolution strength between operating states in the constructed feature graph; and Represents the normalized operating parameter vectors in the i-th and j-th time windows, respectively. The parameters include transmit power P, intermodulation interference I, and crystal oscillator offset. , voltage-current response V / I, board-level temperature T, instantaneous throughput R; represents the squared Euclidean distance of the multi-parameter states between the i-th and j-th time slices; represents the maximum information coefficient, which is used to capture the degree of nonlinear coupling between state pairs; Indicates the local curvature or abnormally drastic change points in the parameter change map; represents the global variance between the states of the operating parameters; represents the normalized scale factor for the maximum information coefficient; 、 and They represent the weight coefficients of the Euclidean distance term, nonlinear correlation term, and spectral perturbation term, respectively.

[0034] The calculation formula is as follows: ; in, and Represents a discrete distribution where data is divided into different grids; represents the mutual information of the joint grid.

[0035] S3: Based on the feature graph model, a multi-hop path optimization algorithm with a graph attention mechanism is used to jointly model and score the node weights in the path, identify high-weight paths and key parameter nodes that affect communication performance, and generate a critical path strategy graph that includes node configuration relationships and path control logic; Wherein, the step S3 includes the following steps: Based on the feature graph model, a multi-hop path optimization algorithm based on the graph attention mechanism is adopted to jointly model the state characteristics and edge collaboration relationships of nodes in the multi-hop adjacency range of the graph, and construct a multi-hop path structure with cross-level information perception capabilities; In the multi-hop path structure, the attention mechanism is used to dynamically weight the edge weight distribution and state feature differences between nodes within different hop count ranges to generate a node attention distribution graph; Among them, for each target node, the system extracts its 1- to 3-hop adjacent nodes from the feature graph model; the feature combination of each hop neighbor node includes: node state value (such as frequency offset, current anomaly, etc.); edge weight (co-variation strength) between the target node and the target node; hop number label (used to guide hierarchical sensitive modeling); all features are spliced ​​and input into the attention mechanism of the graph neural network to form a feature interaction vector between node pairs.

[0036] The system incorporates a graph attention mechanism (e.g., the GATv2 architecture): each target node calculates an attention coefficient for each of its neighboring nodes. The attention coefficient reflects the weight of the neighboring node in the target node's decision-making. Low-hop neighbors (e.g., one hop) are initially given a higher weight, but two- or three-hop neighbors with strong edge coordination can also have their weight increased. When a node's state changes dramatically (e.g., a sudden power surge or a sharp temperature fluctuation), its attention weight is amplified. Attention weights are updated over time, and the system refreshes the attention distribution of all nodes every second.

[0037] Each node outputs a weight vector corresponding to the attention coefficient of its adjacent nodes; the system organizes the attention weights of all nodes in the entire graph in a matrix manner to form a node attention distribution map; this distribution map is not only used for subsequent path scoring, but also can be used to analyze the clustering areas of communication performance bottlenecks.

[0038] Based on the node attention distribution graph, all reachable multi-hop paths are scored, and the attention weight, edge coordination strength, and hop count of each node in the path are comprehensively considered to screen out a set of high-weight paths with significant path scores. Specifically, for all reachable paths (limited to 3 hops or less) starting from key nodes in the graph, the system calculates the following three key metrics in sequence: Node average attention score: the average weight of all nodes in the path in the current attention graph; Total edge synergy strength: the sum of the synergy weights of all edges in the path; Hop penalty factor: The longer the path, the larger the factor, which reduces the priority of redundant paths.

[0039] The system calculates a comprehensive path score based on a weighted combination of the three factors mentioned above. A high-scoring path indicates that it is highly important in state evolution and that there is close coordination between nodes. Shorter paths are more conducive to real-time control.

[0040] For each key node (such as a node with abnormal indicators), the system starts from it and performs a depth-first path search (DFS); for each 2-3 hop path, it extracts the node set and edge set in sequence; based on the current attention distribution map and edge weight map, it completes the score calculation in real time; all path scores are uniformly sorted, and the top N paths (such as the top 10) are selected as the high-weight path set.

[0041] High-weight paths are stored in a graph structure, with nodes identified by numbers and edges stored by attributes. The path structure and score values ​​can be presented in a visual interface to facilitate subsequent model debugging and policy rule optimization. The system periodically caches high-weight path collections and tracks their evolution trends to discover new key evolution paths.

[0042] Based on the high-weight path set, the key parameter nodes in the path are extracted, their position roles in the path structure and their dependencies with other nodes are analyzed, and a key path strategy diagram integrating regulatory logic and priority relationships is constructed.

[0043] Specifically, in each high-weight path, the system identifies the following role nodes: Source node: The location where the state change first occurs, usually the node where the temperature or power parameter mutation occurs; Conducting node: The intermediate node connecting the source and the target, with a high edge weight, and the state change has a great impact on the terminal; Target node: a parameter node that exhibits significant changes in communication performance (such as decreased throughput and increased bit errors); The system records the above nodes as key parameter nodes and classifies them into a "cause-transmission-result" model structure.

[0044] Perform graph traversal and causal chain analysis on the path structure between key nodes to identify the following relationships: whether there is a strong dependency path (for example, temperature increase is always accompanied by crystal oscillator drift); whether there is a state feedback loop (for example, parameter adjustment reversely affects itself); based on the analysis, generate control rules: for example, "If A increases and causes B to be abnormal, and B abnormality affects C performance", the system generates a control logic chain: "Monitor A, adjust B, protect C".

[0045] A strategy map consists of the following elements: Node collection: contains the unique identifier, parameter type, and logical role (cause, result) of key nodes; Edge set: includes edge start and end nodes, control direction, and dependency level; Path priority: Assign high, medium, or low levels based on the path scoring results; Strategy labels: such as "power regulation", "temperature control protection", "frequency deviation recovery", etc. The strategy graph will serve as input for subsequent strategy migration models (such as meta-learning models) or execution engines to perform intelligent parameter scheduling.

[0046] Furthermore, the formula of the multi-hop path optimization algorithm is as follows: ; in, represents the multi-hop path score from node p to node q, which is used to construct the critical path strategy graph; H represents the maximum hop limit, which controls the hierarchical depth of path modeling; The global adjustment factor of the path of the h-th hop reflects the importance of path aggregation in the current hop layer; represents the set of h-th hop adjacent nodes of the current node p; represents the attention weight; represents the edge coordination weight, which indicates the degree of coordinated change between operating parameters; It represents the instantaneous throughput, which is used to measure the dynamic impact of the node on the communication performance in the h-th hop layer; Indicates the number of hops in the path topology; represents the board-level temperature standard deviation of the nodes associated with node p within the hth hop; Indicates the hop count attenuation factor.

[0047] The calculation formula is as follows: ; in, represents the instantaneous throughput of node k in the hth time window; and Represents adjacent time windows and cooperates with the time window segmentation strategy; The calculation formula is as follows: ; in, represents the board-level temperature sampling value of the h-th hop neighbor node k; Represents the average board-level temperature of the h-th hop neighbor of node p.

[0048] S4: Taking the critical path strategy graph as input, based on the scenario characteristics in a multi-communication environment, a meta-learning algorithm based on graph structure alignment is adopted to build a strategy graph migration model. Through the mapping function, generalized scheduling and optimal migration of parameter configuration under different load conditions are achieved, and the optimal parameter configuration strategy is generated. Wherein, the step S4 includes the following steps: Based on the critical path strategy graph, extract the high-weight path sequence, key parameter node configuration combination and the control dependency relationship between the paths, construct a graph structure sample set associated with the specific communication scenario, and form a source scenario knowledge graph; Specifically, the following core structural information is extracted from the critical path strategy graph generated by S3: High-weight path sequence: such as "transmit power fluctuation → crystal oscillator frequency deviation → throughput reduction"; Key parameter node configuration combination: Node attributes include parameter value range, adjustment direction, and device module ID; Inter-path control dependency: If two paths share a key node, they need to be jointly controlled.

[0049] The above structural information is organized into graph structure samples, where nodes represent parameter configuration entities and edges represent dependency control paths; each sample is bound to the label of the current communication scenario, such as "high-interference urban area" and "medium user density rural area". The set of strategy graph structure samples collected under multiple different typical communication environments is encapsulated into a source scenario knowledge graph library for subsequent alignment learning.

[0050] In the target communication environment, the system combines the real-time characteristics of network load level, interference source distribution, and terminal connection density to construct a graph structure representation of the current scenario and initialize the scenario label and path control demand vector. Specifically, the system is based on a monitoring module deployed on the base station side, which periodically collects the following communication environment indicators to form the input feature set of the target scenario: Network load level: calculated by resource block (PRB) occupancy rate, such as idle <30%, medium 30-70%, overload >70%; Interference source distribution: By measuring the signal strength and SINR value of neighboring cells, the number and strength of interference sources are clustered and divided; Terminal connection density: Counts the number of UEs connected to the current cell and identifies the density level based on the UE mobility speed.

[0051] The process of constructing the target scene graph structure is as follows: Node construction: one node for each type of environmental factor, for example: Node A: Load level = high; Node B: Interference = medium; Node C: Connection density = high.

[0052] Edge definition: Based on historical data analysis, add edges that represent interactive effects, for example, high loads tend to cause increased interference; Node attributes: Each node carries information such as the corresponding factor value range, change trend, mutation probability, etc. Image label: The entire image is assigned a unique scenario label (such as "dense urban area high load scenario") and scenario objective (such as "improve spectrum utilization and reduce bit error rate").

[0053] Based on the current environmental analysis results and performance requirements, the system automatically generates a set of parameter control priorities: such as frequency stability > power efficiency > temperature suppression; this priority vector will serve as the target guidance value for subsequent mapping functions to adjust the configuration of key nodes.

[0054] A meta-learning algorithm based on graph structure alignment is used to perform structural mapping between the knowledge graph in the source scene and the target scene graph, and a fast parameter adaptation mechanism is used to generate the optimal path control mapping function in the target scene. Specifically, a graph embedding algorithm (such as GraphSAGE or GINE) is used to generate graph structure vectors for the source scene graph and the target scene graph respectively; similarity calculation is performed on the graph structure vectors to measure the structural differences between the two graphs (such as the difference in the number of nodes and the difference in edge connectivity); and a graph edit distance metric is introduced to quantify the minimum set of operations required to "transform from the source graph to the target graph."

[0055] Use the MAML algorithm framework to build a cross-scenario migration strategy: Meta-task definition: Each source scene graph structure and its control results are defined as a training task; Inner loop update: Each task trains the policy model in a small number of iterations and records the parameter update direction; Outer loop optimization: learning how to initialize a policy network so that it can converge quickly on new tasks; After the meta-model is trained, it can accept any new scene graph structure as input and quickly adjust the policy network to generate an adaptive mapping function.

[0056] Finally, the meta-learning model outputs a structure alignment mapping function, which takes as input the structural features of the target scene graph and outputs the policy configuration adjustment instructions. This function has the following functions: adjust the node parameter value range in the source strategy graph according to the node characteristics of the target scenario; modify the control intensity and trigger conditions of the path according to the structure mapping results; and output a set of efficient path combinations that are suitable for the target scenario.

[0057] Based on the control mapping function, the key path strategy graph in the source scenario is mapped to the target communication scenario, the structural migration of the key parameter configuration relationship is achieved, and the optimal parameter configuration strategy is output.

[0058] Specifically, after the graph structure alignment is completed, the system automatically performs the following strategy graph migration operations: Map the key paths in the source graph to the functional areas in the target graph (such as projecting the "frequency compensation path" to the "interference-intensive node subgraph"); update the parameter configuration range of key nodes (such as adjusting the original power limit of 2025dBm to 1822dBm); dynamically replace low-fitness paths or nodes, and add supplementary paths based on the target scenario.

[0059] The output optimal parameter configuration strategies include: Parameter dimension table: lists the parameter names, optimal setting values, adjustment ranges, and limit boundaries of key nodes; Control logic diagram: Generates "IF-THEN" control logic based on the path, such as: IF interference level is high & connection density is high THEN reduce the transmit power by 2dBm; Policy application conditions: The system automatically marks the load intervals, time periods, or device types to which the policy applies; Set up real-time monitoring feedback. If the performance drops by more than a threshold after policy execution, enable an alternative policy or roll back the old policy. Add a policy scoring mechanism to dynamically evaluate the control effect for subsequent policy updates and iterations.

[0060] After the strategy is pushed to the 5G communication module, the control engine dynamically adjusts the module parameters based on the control strategy. The module operation results are uploaded in real time, and the following data is fed back for strategy backtracking and optimization: throughput rate change trend; BER change curve; parameter response delay and stabilization time; the feedback data is compared and analyzed with the initial strategy effect to provide reference for the next fine-tuning of the meta-model.

[0061] Furthermore, the formula of the optimal path control mapping function is as follows: ; in, represents the optimal path control mapping function under the target communication scenario R; N represents the number of source scenario samples participating in the control learning; represents the key path parameter vector extracted from the nth source scene; Indicates the optimal parameter configuration result expected to be generated by the nth sample in the target scenario; represents the critical path strategy graph structure constructed in the nth source communication scenario; R represents the scenario graph structure in the target communication environment; represents the path regulation mapping function; represents the regularization term of function J; represents the regularization coefficient.

[0062] S5: Apply the optimal parameter configuration strategy to the actual operation of the 5G communication module, collect the real-time operating status of the module as the fuzzy input variable, build a reinforced fuzzy reasoning network model, evaluate the deviation and adaptability of the current execution strategy through fuzzy reasoning rules combined with the reinforcement learning reward mechanism, and perform parameter fine-tuning based on performance feedback.

[0063] The enhanced fuzzy reasoning network model is specifically based on the core parameter nodes in the critical path strategy graph as input dimensions, collecting status indicators such as transmission power offset, voltage and current fluctuations, crystal oscillator frequency deviation, and temperature change, constructing a fuzzy input vector, and completing fuzzification processing through membership functions. It also generates a fuzzy rule base based on the node dependency relationship in the strategy graph, and introduces a reinforcement learning mechanism to dynamically adjust the fuzzy rule weights and execution strategy priorities. The reinforcement learning reward mechanism is specifically constructed based on the dynamic feedback of communication performance indicators, including the throughput improvement value, bit error rate reduction and switching delay convergence rate as core reward factors, and combining the weighted functions of each indicator to generate a real-time performance score.

[0064] Specifically, the generated optimal parameter configuration strategy is deployed in the operation control unit of the 5G communication module. After the system is started, the module's operating status data in the multi-carrier concurrent communication state is collected in real time, mainly including: Transmit power offset: Indicates the difference between the current transmit power and the target transmit power.

[0065] Voltage and current fluctuations: reflect the stability of the power supply system and capture instantaneous changes caused by environmental or load fluctuations.

[0066] Crystal oscillator frequency deviation: used to monitor the accuracy of the clock system and has a significant impact on frequency synchronization stability.

[0067] Temperature changes: Record the heat accumulation status of key components and evaluate their thermal impact on module performance.

[0068] The above four types of state indicators are mapped into fuzzy input variables respectively, and each variable is fuzzified according to the empirical rules or the membership function set by the expert system (such as the "three-stage" low-medium-high membership level) to form a fuzzy input vector.

[0069] A fuzzy rule base is constructed based on the dependencies between core parameter nodes in the critical path strategy graph. Each rule describes the optimal control behavior for a specific input combination, for example, "If the transmit power offset is high and the voltage fluctuation is medium, then reduce the power control weight." These rules serve as the foundation for the inference engine, enabling real-time policy decisions.

[0070] A reinforcement learning module based on sequential decision-making is embedded in the inference engine. This module continuously tracks the deviation between fuzzy inference results and actual performance. By comparing the system performance after executing the fuzzy output strategy with the expected benefits, it dynamically updates the weight of each fuzzy rule and adjusts the priority of strategy selection during the inference process.

[0071] Build a performance feedback channel to collect three key communication performance indicators in real time after the system is running: Throughput improvement: measures the extent to which the current policy enhances data transmission capabilities.

[0072] Bit error rate reduction: reflects the improvement in link quality.

[0073] Handover delay convergence rate: evaluates the response efficiency during multi-band handover or carrier aggregation.

[0074] Each indicator is combined by setting weighted coefficients to generate a comprehensive performance score, which serves as an immediate reward signal in reinforcement learning and drives the strategy model to converge towards a better direction.

[0075] The system compares the above comprehensive performance score with the preset threshold. If there is a significant deviation in the current strategy execution effect, the module will trigger the adaptive parameter fine-tuning process to automatically adjust the core parameter configuration in the critical path until the system runs stably in the optimal state area, achieving closed-loop optimization control.

[0076] In summary, this invention, by introducing a high-temporal-resolution data acquisition mechanism and integrating the monitoring of multiple heterogeneous parameters (such as power, frequency deviation, temperature, and current), constructs a state diagram of operating parameters under realistic and complex operating conditions. This significantly improves the system's perception of dynamic communication states, providing solid data support for subsequent control. The combined mapping of the maximum information coefficient (MIC) and linear correlation indices not only enhances the ability to mine nonlinear parameter coupling relationships, but also enables highly robust modeling of complex physical properties, effectively identifying key state change patterns that are difficult to detect using traditional linear methods.

[0077] By introducing a polarization operator, the present invention can effectively filter abnormal nodes and achieve soft clustering in the face of high noise and high disturbance states, ensuring that the state graph clustering results have higher stability and resolution, and effectively supporting the accuracy and reliability of subsequent key path construction. By combining graph neural networks with multi-hop attention mechanisms, the dependency modeling capabilities between state nodes across levels and hops are enhanced, effectively identifying the core paths and key parameter nodes that affect communication performance, and providing a structured, high-value strategy graph for control strategy generation. By constructing a source scenario knowledge graph that aligns with the graph structure of the target communication environment, and combining graph embedding with the MAML meta-learning algorithm, the system has the ability to "rapidly adapt to a small number of samples" and can quickly generate adaptive optimal parameter configuration strategies under rapid changes in the communication environment (such as carrier switching and sudden changes in network load).

[0078] This invention maps the structure and migrates parameters of the critical path strategy graph, effectively enabling the transfer and reuse of control knowledge across different scenarios and devices. This reduces system development and tuning costs, and improves the engineering versatility and reuse efficiency of strategy design. Fuzzy control and reinforcement learning are integrated to construct an adaptive inference network model. Dynamic parameter fine-tuning is performed based on communication performance feedback (throughput, bit error rate, and handover latency). This enables continuous optimization of the optimal strategy and online self-learning capabilities, ensuring the system's long-term performance stability and intelligent responsiveness.

[0079] The above embodiments are merely descriptions of preferred embodiments of the present invention and are not intended to limit the scope of the present invention. Without departing from the design spirit of the present invention, various modifications and improvements made to the technical solutions of the present invention by ordinary engineering technicians in this field should fall within the scope of protection determined by the claims of the present invention.

Claims

1. A method for optimizing a high-performance 5G communication module, characterized in that: The following steps are involved: S1: Obtain the operating parameters of the 5G communication module in a multi-carrier concurrent state; S2: Based on the operating parameters, a graph polarization spectral clustering algorithm is used to cluster the state graph and construct a feature graph model. The nodes in the graph represent the operating parameter states, and the edge weights represent the degree of coordinated changes between the parameters. S3: Based on the feature graph model, a multi-hop path optimization algorithm with a graph attention mechanism is used to jointly model and score the node weights in the path, identify high-weight paths and key parameter nodes that affect communication performance, and generate a critical path strategy graph that includes node configuration relationships and path control logic; S4: Taking the critical path strategy graph as input, based on the scenario characteristics in a multi-communication environment, a meta-learning algorithm based on graph structure alignment is adopted to build a strategy graph migration model. Through the mapping function, generalized scheduling and optimal migration of parameter configuration under different load conditions are achieved, and the optimal parameter configuration strategy is generated. S5: Apply the optimal parameter configuration strategy to the actual operation of the 5G communication module, collect the real-time operating status of the module as the fuzzy input variable, build a reinforced fuzzy reasoning network model, evaluate the deviation and adaptability of the current execution strategy through fuzzy reasoning rules combined with the reinforcement learning reward mechanism, and perform parameter fine-tuning based on performance feedback.

2. The optimization method of a high-performance 5G communication module according to claim 1, characterized in that: The operating parameters include transmit power, intermodulation interference, crystal oscillator offset, voltage-current response, board temperature, and instantaneous throughput.

3. The optimization method of a high-performance 5G communication module according to claim 1, characterized in that: The step S2 comprises the following steps: Based on the operating parameters, parameter normalization and time window segmentation strategies are adopted to construct a time series state diagram reflecting the parameter evolution process. The nodes represent the parameter states in different time slices, and the edge weights represent the coordinated change amplitude between adjacent parameter states. By calculating the correlation coefficient matrix for each parameter node pair in the time series state diagram and combining the maximum information coefficient to enhance the nonlinear collaborative characteristics, a weighted multi-attribute state map is generated; Based on the weighted multi-attribute state graph, a graph polarization spectral clustering algorithm is used to perform spectral decomposition and feature space mapping on the node set. By introducing a polarization operator to adjust the spectral clustering boundary, high-noise and weakly coupled parameter states are clustered, and parameter state cluster labels are output; Based on the parameter state clustering labels, the core nodes and edge weight distribution in each subgraph are extracted, a feature graph model including node features, edge collaboration weights and clustering labels is constructed, and a structured feature graph model is output.

4. The optimization method for a high-performance 5G communication module according to claim 3, characterized in that: The formula for constructing the feature graph model is as follows: ； in, Indicates the co-evolution strength between operating states in the constructed feature graph; and Represent the normalized operating parameter vectors in the i-th and j-th time windows respectively; represents the squared Euclidean distance of the multi-parameter states between the i-th and j-th time slices; represents the maximum information coefficient; Indicates the local curvature or abnormally drastic change points in the parameter change map; represents the global variance between the states of the operating parameters; The normalized scale factor representing the maximum information coefficient; 、 and They represent the weight coefficients of the Euclidean distance term, nonlinear correlation term, and spectral perturbation term, respectively.

5. The optimization method of a high-performance 5G communication module according to claim 1, characterized in that: The step S3 comprises the following steps: Based on the feature graph model, a multi-hop path optimization algorithm based on the graph attention mechanism is adopted to jointly model the state characteristics and edge collaboration relationships of nodes in the multi-hop adjacency range of the graph, and construct a multi-hop path structure with cross-level information perception capabilities; In the multi-hop path structure, the attention mechanism is used to dynamically weight the edge weight distribution and state feature differences between nodes within different hop count ranges to generate a node attention distribution graph; Based on the node attention distribution graph, all reachable multi-hop paths are scored, and the attention weight, edge coordination strength, and hop count of each node in the path are comprehensively considered to screen out a set of high-weight paths with significant path scores. Based on the high-weight path set, the key parameter nodes in the path are extracted, their position roles in the path structure and their dependencies with other nodes are analyzed, and a key path strategy diagram integrating regulatory logic and priority relationships is constructed.

6. The method for optimizing a high-performance 5G communication module according to claim 5, wherein: The formula of the multi-hop path optimization algorithm is as follows: ； in, represents the multi-hop path score from node p to node q, which is used to construct the critical path strategy graph; H represents the maximum hop limit, which controls the hierarchical depth of path modeling; represents the global adjustment factor of the path of the h-th hop; represents the set of h-th hop adjacent nodes of the current node p; represents the attention weight; represents the edge coordination weight, which indicates the degree of coordinated change between operating parameters; It represents the instantaneous throughput, which is used to measure the dynamic impact of the node on the communication performance in the h-th hop layer; Indicates the number of hops in the path topology; represents the board-level temperature standard deviation of the nodes associated with node p within the hth hop; Indicates the hop count attenuation factor.

7. The method for optimizing a high-performance 5G communication module according to claim 1, wherein: The step S4 comprises the following steps: Based on the critical path strategy graph, extract the high-weight path sequence, key parameter node configuration combination and the control dependency relationship between the paths, construct a graph structure sample set associated with the specific communication scenario, and form a source scenario knowledge graph; In the target communication environment, the system combines the real-time characteristics of network load level, interference source distribution, and terminal connection density to construct a graph structure representation of the current scenario and initialize the scenario label and path control demand vector. A meta-learning algorithm based on graph structure alignment is used to perform structural mapping between the knowledge graph in the source scene and the target scene graph, and a fast parameter adaptation mechanism is used to generate the optimal path control mapping function in the target scene. Based on the control mapping function, the key path strategy graph in the source scenario is mapped to the target communication scenario, the structural migration of the key parameter configuration relationship is achieved, and the optimal parameter configuration strategy is output.

8. The method for optimizing a high-performance 5G communication module according to claim 7, wherein: The formula of the optimal path control mapping function is as follows: ； in, represents the optimal path control mapping function under the target communication scenario R; N represents the number of source scenario samples participating in the control learning; represents the key path parameter vector extracted from the nth source scene; Indicates the optimal parameter configuration result expected to be generated by the nth sample in the target scenario; represents the critical path strategy graph structure constructed in the nth source communication scenario; R represents the scenario graph structure in the target communication environment; represents the path regulation mapping function; represents the regularization term of function J; represents the regularization coefficient.

9. The method for optimizing a high-performance 5G communication module according to claim 1, wherein: The enhanced fuzzy reasoning network model is specifically based on the core parameter nodes in the critical path strategy graph as input dimensions, collecting status indicators such as transmission power offset, voltage and current fluctuations, crystal oscillator frequency deviation and temperature change, constructing a fuzzy input vector, and completing fuzzification processing through membership functions; combining the node dependency relationship in the strategy graph to generate a fuzzy rule base, and introducing a reinforcement learning mechanism to dynamically adjust the fuzzy rule weights and execution strategy priorities.

10. The method for optimizing a high-performance 5G communication module according to claim 1, wherein: The reinforcement learning reward mechanism is specifically constructed based on the dynamic feedback of communication performance indicators, including the throughput improvement value, bit error rate reduction and switching delay convergence rate as core reward factors, and combining the weighted functions of each indicator to generate a real-time performance score.

Images