5GCPE Dynamic Spectrum Allocation Method and System

Abstract

The application discloses a 5G CPE dynamic spectrum allocation method and system, and relates to the technical field of spectrum allocation. The method comprises the following steps: taking a CPE as a spectrum decision subject, collecting a multi-source state data set associated with current spectrum use in a preset time window; constructing a spectrum instantaneous state vector representing an operation situation; inputting the spectrum instantaneous state vector into a spectrum state evolution driving model to perform state evolution direction and evolution intensity analysis; constructing a spectrum state feasible evolution domain of an allowable evolution range; and performing CPE dynamic spectrum allocation management according to the optimal spectrum state migration path. The application solves the technical problems of insufficient adaptability and low operation reliability of the prior art in 5G dynamic spectrum allocation, and achieves the technical effects of taking a CPE as a decision subject, accurately selecting an optimal spectrum migration path, and improving the adaptability and reliability of 5G dynamic spectrum allocation.

Description

Technical Field

[0001] This invention relates to the field of spectrum allocation technology, specifically to a dynamic spectrum allocation method and system for 5G CPE. Background Technology

[0002] With the rapid development of 5G communication technology, the demand for spectrum resources from various terminal devices continues to grow. The contradiction between the scarcity of spectrum resources and the diversification of service needs is becoming increasingly prominent. Traditional spectrum allocation methods rely heavily on centralized decision-making on the network side, which has problems such as delayed response and difficulty in accurately adapting to the real-time operating status of terminals and dynamic changes in services. Furthermore, they do not fully consider the limitations of CPE hardware capabilities, network configuration requirements, and service continuity requirements, resulting in insufficient adaptability and reliability of spectrum allocation. This can easily lead to channel interference, service interruption, or low resource utilization, and cannot meet the service requirements of high bandwidth, low latency, and wide connectivity in 5G scenarios.

[0003] Existing technologies suffer from insufficient adaptability to 5G dynamic spectrum allocation and low operational reliability. Summary of the Invention

[0004] This application provides a 5G CPE dynamic spectrum allocation method and system to address the technical problems of insufficient adaptability and low operational reliability of 5G dynamic spectrum allocation in the prior art.

[0005] In view of the above problems, this application provides a method and system for dynamic spectrum allocation of 5G CPE.

[0006] A first aspect of this application provides a method for dynamic spectrum allocation in 5G CPE, the method comprising:

[0007] Using the CPE as the spectrum decision-making entity, a multi-source state dataset associated with the current spectrum usage is collected within a preset time window. This multi-source state dataset includes spectrum occupancy characteristics, channel quality stability characteristics, interference change trend characteristics, and service load continuity characteristics. Based on this multi-source state dataset, a spectrum instantaneous state vector representing the operational status is constructed. This instantaneous state vector is input into a spectrum state evolution driving model to analyze the evolution direction and intensity of the spectrum state caused by changes in the external environment, service behavior, and historical usage results. After obtaining the output results, constraint mapping of the output results is performed using 5G network configuration constraints, CPE hardware capability constraints, and service continuity constraints to construct a feasible evolution domain for the spectrum state with an allowable evolution range. Within this feasible evolution domain, multiple candidate spectrum state transition paths are evaluated, and the optimal spectrum state migration path is selected. Dynamic spectrum allocation management for the CPE is then performed based on this optimal spectrum state migration path.

[0008] A second aspect of this application provides a 5G CPE dynamic spectrum allocation system, the system comprising:

[0009] The system comprises the following modules: a dataset acquisition module, which collects multi-source state datasets related to current spectrum usage within a preset time window, with the CPE as the spectrum decision-making entity. These multi-source state datasets include spectrum occupancy characteristics, channel quality stability characteristics, interference change trend characteristics, and service load continuity characteristics. An instantaneous state vector construction module is used to construct an instantaneous spectrum state vector representing the operational status based on the multi-source state dataset. An evolution intensity analysis module inputs the instantaneous spectrum state vector into a spectrum state evolution driving model to analyze the evolution direction and intensity of the spectrum state caused by changes in the external environment, service behavior, and historical usage results. A feasible evolution domain construction module, after obtaining the output results, performs constraint mapping on the output results using 5G network configuration constraints, CPE hardware capability constraints, and service continuity constraints to construct a feasible evolution domain for the spectrum state with an allowable evolution range. A dynamic spectrum allocation management module evaluates multiple candidate spectrum state transition paths within the feasible evolution domain, selects the optimal spectrum state migration path, and performs dynamic spectrum allocation management for the CPE based on the optimal spectrum state migration path.

[0010] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0011] Using the CPE as the spectrum decision-making entity, a multi-source state dataset associated with current spectrum usage is collected within a preset time window. Based on this dataset, an instantaneous spectrum state vector representing the operational status is constructed. This vector is then input into a spectrum state evolution driving model to analyze the evolution direction and intensity of the spectrum state, influenced by changes in the external environment, service behavior, and historical usage results. A feasible evolution domain for the spectrum state with an allowable evolution range is constructed. Within this domain, multiple candidate spectrum state transition paths are evaluated, and the optimal path is selected. Dynamic spectrum allocation management for the CPE is then performed based on this optimal path. This approach achieves the technical effect of accurately selecting the optimal spectrum migration path with the CPE as the decision-making entity, thereby improving the adaptability and reliability of 5G dynamic spectrum allocation. Attached Figure Description

[0012] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0013] Figure 1 This is a schematic diagram of the 5G CPE dynamic spectrum allocation method provided in the embodiments of this application;

[0014] Figure 2 This is a schematic diagram of the 5G CPE dynamic spectrum allocation system provided in the embodiments of this application.

[0015] Figure labeling: Dataset acquisition module 10, instantaneous state vector construction module 20, evolution intensity analysis module 30, feasible evolution domain construction module 40, dynamic spectrum allocation management module 50. Detailed Implementation

[0016] This application provides a 5G CPE dynamic spectrum allocation method and system to address the technical problems of insufficient adaptability and low operational reliability of 5G dynamic spectrum allocation in the prior art.

[0017] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application.

[0018] Example 1, as Figure 1 As shown, this application provides a 5G CPE dynamic spectrum allocation method, the method comprising:

[0019] Step S100: Using CPE as the spectrum decision-making entity, collect multi-source state datasets associated with the current spectrum usage within a preset time window. The multi-source state datasets include spectrum occupancy characteristics, channel quality stability characteristics, interference change trend characteristics, and service load continuity characteristics.

[0020] Specifically, during the collection of multi-source state datasets, the CPE (Center Premises Equipment) serves as the core entity for spectrum decision-making. Relying on its built-in spectrum monitoring unit, channel quality detection module, and service load statistics component, it conducts comprehensive data collection within a pre-defined fixed time window. The collection process focuses on key characteristic dimensions directly related to current spectrum usage, including spectrum occupancy characteristics (accurately calculating the resource occupancy ratio, idle time distribution, and reuse status of each frequency band), channel quality stability characteristics (real-time monitoring of fluctuations in indicators such as signal transmission rate, bit error rate, and latency jitter), interference trends (continuously tracking changes in the intensity, frequency distribution, and impact range of external electromagnetic interference), and service load continuity characteristics (dynamically recording peak and trough traffic, connection duration, and load fluctuation frequency for various services). Through the synchronous collection and integration of multi-dimensional data, a comprehensive and accurate multi-source state dataset is formed, providing solid data support for the subsequent construction of instantaneous spectrum state vectors and spectrum evolution analysis.

[0021] Step S200: Construct a spectral instantaneous state vector representing the operational status based on the multi-source state dataset.

[0022] Specifically, in constructing the instantaneous spectrum state vector, a multi-source state dataset is used as the core foundation. Accurate quantification of the operational status is achieved through multi-dimensional data integration and situational characterization modeling. First, the spectrum occupancy characteristics, channel quality stability characteristics, interference change trend characteristics, and service load continuity characteristics in the dataset are preprocessed. This includes data normalization to eliminate dimensional differences, outlier removal to ensure data reliability, and feature standardization to unify data distribution, ensuring the fusionability of different feature types. Then, a feature dimension mapping and weight allocation algorithm is used to assign corresponding weight coefficients to each feature based on its priority of impact on the spectrum operational status, such as the correlation between service load continuity and real-time spectrum demand, and the decisive role of channel quality stability in transmission efficiency. Finally, tensor fusion technology is used to integrate the multi-dimensional feature data into a unified high-dimensional data structure. Finally, considering the dynamic characteristics of spectrum operation, timestamp marking and situation feature extraction algorithms are introduced to extract core characterization parameters from the integrated high-dimensional data that can reflect the current spectrum resource utilization status, channel operation quality, interference impact level and service load requirements in real time. In the end, a comprehensive, accurate and dynamically adaptable instantaneous spectrum state vector is constructed, providing structured and computable core input data for subsequent spectrum state evolution analysis.

[0023] Step S300: Input the instantaneous spectrum state vector into the spectrum state evolution driving model, and perform state evolution direction and evolution intensity analysis of the spectrum state caused by the combined effects of changes in the external environment, changes in business behavior, and historical usage results.

[0024] Specifically, the constructed instantaneous spectrum state vector is first input into the spectrum state evolution driving model. The model transforms this vector into a multidimensional spectrum evolution state tensor through a mapping processing layer, which includes the evolution potential representations of spectrum occupancy, channel quality stability, interference change trends, and service load continuity at different time scales. Subsequently, historical spectrum state trajectories, external environmental gradients, and current service behavior changes, including service request volume, service type priority, and load continuity fluctuations, are read. Potential field mapping channels are configured, and each element of the multidimensional spectrum evolution state tensor is assigned an evolution potential value. Based on this potential value, state expansion processing is performed to construct a joint representation structure covering the time dimension, service behavior disturbance dimension, and environmental gradient dimension, forming the spectrum evolution state tensor space. Within this space, a historical evolution inertia descriptor is first constructed based on the stable holding time, recovery speed, and evolution backsliding probability of each component in the historical spectrum state trajectory. This descriptor is then coupled with current service behavior changes. Based on the coupling result, direction-dependent evolution trend modulation parameters are configured for each spectrum state component, constructing a non-uniform evolution potential distribution, so that different spectrum state components exhibit differentiated evolution attraction and suppression directions. Finally, the local evolution gradient vector of the spectrum state is calculated along this distribution. By the cumulative response intensity of the gradient vector at multiple time scales, the evolution direction and evolution intensity descriptors of the spectrum state under the combined effects of changes in the external environment, changes in business behavior, and historical usage results are constructed, thus completing a comprehensive analysis of the evolution direction and evolution intensity of the spectrum state.

[0025] Step S400: After obtaining the output results, use the 5G network configuration constraints, CPE hardware capability constraints and service continuity constraints to perform constraint mapping on the output results and construct a feasible evolution domain of spectrum state with an allowable evolution range.

[0026] Specifically, when constructing the feasible evolutionary domain of the spectrum state, the evolution direction descriptor and evolution intensity descriptor output by the spectrum state evolution driving model are first obtained, and then systematic constraint mapping is carried out based on the three types of core constraints. First, the evolution direction description is precisely matched with the pre-configured 5G network spectrum resource allocation rules to clarify the set of permissible evolution directions for spectrum status in dimensions such as band switching, bandwidth expansion, and carrier combination, ensuring that spectrum evolution complies with network resource allocation specifications. Next, based on the CPE's RF front-end performance, baseband processing capabilities, and RF switching response capabilities, hardware capability constraints are mapped to the evolution intensity description, strictly limiting the maximum evolution amplitude of the spectrum status per unit time, forming a clear evolution intensity boundary to avoid exceeding the equipment hardware's capacity. Subsequently, spectrum status evolution smoothness constraints are constructed based on service continuity constraints, ensuring service transmission stability by limiting the mutation rate and the number of consecutive jumps during the evolution process. Finally, by comprehensively considering the permissible evolution direction set, evolution intensity boundary, and spectrum status evolution smoothness constraints, a closed or semi-closed feasible evolution domain for spectrum status is generated within the status evolution description space, defining a legal and reasonable evolution range for the subsequent generation and evaluation of candidate spectrum status transition paths.

[0027] Step S500: Within the feasible evolution domain of the spectrum state, evaluate multiple candidate spectrum state transition paths, select the optimal spectrum state transition path, and perform CPE dynamic spectrum allocation management based on the optimal spectrum state transition path.

[0028] Specifically, taking the current spectrum state as the initial node, the candidate path search space is hierarchically pruned based on the geometric boundary structure of the feasible evolution domain of the spectrum state. Then, based on the hierarchical pruning results, the priority selection of the reachability density of the evolution domain is configured, generating multiple candidate spectrum state transition paths that satisfy directional constraints and intensity boundaries and consist of continuous spectrum state evolution nodes. Subsequently, along each candidate path, its stage-by-stage evolution cost in terms of evolution direction consistency, evolution intensity stability, and the degree of maintenance of business service continuity is evaluated. The disturbance impact of changes in business behavior on spectrum demand is mapped into a dynamic correction factor for the path weights, and the comprehensive evaluation weights are adjusted in real time during the path evolution process. This completes the process for all paths. After path evaluation, the optimal path is selected based on the joint evaluation results of cumulative evolution cost and service disturbance adaptability. If there are multiple candidate paths whose comprehensive evaluation results meet the preset deviation range, the length bias selection of the stable evolution segment is performed to determine the final optimal spectrum state migration path. Finally, CPE dynamic spectrum allocation management is performed based on the optimal path, and the actual spectrum state evolution is monitored and a monitoring dataset is constructed simultaneously. Deviation anomaly identification and deviation warning are performed by comparing the monitoring dataset with the optimal path. Response strategies are configured according to the warning and a tiered verification window is set. Consistency evaluation and feedback optimization of the response strategy are completed within the window to achieve precise control and dynamic adaptation of spectrum allocation.

[0029] In one possible implementation, step S300 further includes:

[0030] Step S310: Based on the mapping processing layer, the instantaneous spectrum state vector is mapped into a multidimensional spectrum evolution state tensor. The multidimensional spectrum evolution state tensor includes the evolution potential of spectrum occupancy, channel quality stability, interference change trend, and service load continuity at different time scales.

[0031] Step S320: After reading the historical spectrum status trajectory, external environment gradient and current business behavior changes, configure the potential field mapping channel. The current business behavior changes include business request volume, business type priority and load continuity fluctuation.

[0032] Step S330: Assign evolution potential values ​​to each element of the multidimensional spectral evolution state tensor using the potential field mapping channel.

[0033] Step S340: Based on the evolution potential value, perform state expansion processing on the multidimensional spectrum evolution state tensor to construct a joint representation structure containing each spectrum state component in the time dimension, service behavior disturbance dimension and environmental gradient dimension, forming a spectrum evolution state tensor space.

[0034] Step S350: Construct a non-uniform evolution potential distribution in the tensor space of the spectral evolution state, wherein the non-uniform evolution potential distribution causes different spectral state components to exhibit differentiated evolution attraction and inhibition directions under the drive of changes in business behavior.

[0035] Step S360: Calculate the local evolution gradient vector of the spectral state along the non-uniform evolution potential distribution, and construct the evolution direction descriptor and evolution intensity descriptor that characterize the evolution trend of the spectral state itself by the cumulative response intensity of the local evolution gradient vectors at multiple time scales.

[0036] Specifically, firstly, a feature separation algorithm is used to accurately decompose four core features in the instantaneous spectrum state vector: spectrum occupancy, channel quality stability, interference change trend, and service load continuity, extracting the basic parameters and dynamic correlation information of each feature. Then, through time-scale discretization, multiple time-scale dimensions (short-term, medium-term, long-term, etc.) are integrated into the feature representation. Combined with time-series prediction algorithms, such as LSTM time-series modeling, the potential change patterns and evolution potential of various features at different time scales are explored. Finally, tensor decomposition and reconstruction techniques are used to structurally integrate the decomposed single-feature multi-time-scale evolution potential data, constructing a multi-dimensional spectrum evolution state tensor with dimension matching and feature correlation. This tensor clearly distinguishes different feature types and time scales through tensor dimension indexing, achieving a comprehensive and structured representation of the evolution potential of the four core features.

[0037] By calling the historical spectrum state database through a distributed data storage interface, and using time-series data extraction algorithms, the core trajectory data such as the stable holding time, recovery speed, and evolution backsliding probability of each spectrum state component are read. At the same time, with the help of the environmental sensing unit built into the CPE and the network-side environmental monitoring node, external environmental gradients are collected through multi-dimensional data fusion technology, including real-time environmental parameters such as electromagnetic interference intensity, signal propagation attenuation coefficient, and interference source distribution density. Then, through the service traffic monitoring module and priority management unit, real-time stream processing algorithms are used to capture current service behavior change data, specifically covering the millisecond-level fluctuation value of service request volume, dynamic adjustment records of service type priority, and sliding window fluctuation coefficient of load continuity. Based on the above three types of data, the influence weights of historical state, external environment, and service changes on spectrum evolution are calculated through weight allocation algorithms, such as the analytic hierarchy process. Combined with the channel dimension matching algorithm, a potential field mapping channel corresponding one-to-one with the multi-dimensional spectrum evolution state tensor feature dimensions is dynamically constructed. This channel ensures the accuracy and efficiency of data transmission in the subsequent process of assigning evolution potential values ​​through preset feature mapping rules and data transmission protocols, realizing the adaptive association between multi-source inputs and tensor elements.

[0038] Based on the pre-defined feature association rules of the potential field mapping channel, a gradient descent algorithm is used to train historical spectrum state evolution data, external environment influence coefficients, and service behavior weights to construct an evolution potential value calculation model, clarifying the quantitative standards for evolution potential under different feature dimensions and time scales. Subsequently, through the channel data parsing interface, the feature parameters of spectrum occupancy, channel quality stability, interference change trends, and service load continuity in the multidimensional spectrum evolution state tensor at each time scale are matched one by one with the adaptation dimensions in the potential field mapping channel. Then, using a weighted summation algorithm, combined with historical evolution inertia weights, environmental disturbance coefficients, and service demand urgency, the matched tensor elements are quantified to generate the evolution potential value corresponding to each element. The magnitude of the evolution potential value is positively correlated with the evolution potential of the feature parameters; the higher the value, the more likely the spectrum state component corresponding to that element is to change in subsequent evolution. Finally, a data verification mechanism is used to verify the rationality of the assigned evolution potential value, eliminating outliers and correcting deviations to ensure that the evolution potential value can truly reflect the potential evolution trend of each tensor element.

[0039] Based on the quantitative distribution of evolutionary potential values, a tensor dimension expansion algorithm is used to extend the state of the multidimensional spectrum evolution state tensor, adding three core dimensions—time dimension, business behavior disturbance dimension, and environmental gradient dimension—to the original feature dimensions. The time dimension incorporates evolutionary correlation features across short-term, medium-term, and long-term time scales through temporal interpolation. The business behavior disturbance dimension integrates disturbance factors such as business request volume fluctuations and priority adjustments using a business demand disturbance modeling method. The environmental gradient dimension incorporates environmental impact features such as electromagnetic interference intensity gradient and propagation attenuation gradient using an environmental parameter gradient calculation model. Subsequently, a multi-dimensional feature coupling algorithm is used to deeply correlate and fuse the features of each spectrum state component in the three newly added dimensions with the original evolutionary potential values, constructing a three-dimensional joint representation structure of features, dimensions, and potential. Finally, using tensor space reconstruction technology, based on the dimensional correlation and data distribution patterns of the joint representation structure, a spectrum evolution state tensor space with high dimensional adaptability and evolutionary situation characterization capability is constructed. This space can accurately bear the evolutionary characteristics of each spectrum state component under multi-dimensional influence, providing a structured carrier for the subsequent construction of non-uniform evolutionary potential distribution.

[0040] Through time-series statistical analysis algorithms, the stability duration, recovery speed, and evolutionary regression probability of each spectral state component are extracted from historical spectral state trajectories. A weighted average model is used to calculate the weights of each parameter and construct a quantified historical evolutionary inertia descriptor, accurately representing the inherent evolutionary characteristics of different components. Subsequently, a multi-factor coupling algorithm is employed to nonlinearly correlate this historical evolutionary inertia descriptor with current business behavior changes, including the magnitude of sudden changes in business requests, the priority adjustment coefficient of business types, and the frequency of load continuity fluctuations. Matrix operations are used to quantify the disturbance intensity of business changes on the original evolutionary inertia of each spectral state component, forming... The perturbation intensity matrix is ​​generated. Finally, based on the perturbation intensity matrix, potential energy field modeling technology is used to dynamically configure direction-related evolution trend modulation parameters for each spectral state component in the spectral evolution state tensor space, including attraction direction coefficient and suppression direction threshold. Through parameter weighting and spatial potential superposition, a non-uniform evolution potential distribution is constructed, so that different spectral state components exhibit differentiated evolution attraction directions based on their own inertia and perturbation intensity under the drive of changes in business behavior. That is, the spectral state that is more easily tended and the suppression direction that is difficult to break through, which provides accurate potential energy field support for the subsequent calculation of local evolution gradient vectors.

[0041] Based on the spatial potential energy gradient characteristics of non-uniform evolution potential distribution, gradient descent optimization algorithms, such as stochastic gradient descent (SGD), are used to calculate the partial derivatives of the potential energy distribution of each spectral state component in the spectral evolution tensor space. This yields the local evolution gradient vector of each component in the current evolution potential field. The direction of this vector directly corresponds to the potential evolution trend of the spectral state component, and the vector magnitude reflects the initial intensity of local evolution. Subsequently, through time-scale discretization, multiple time scales, including short-term, medium-term, and long-term, are incorporated into the analysis. A sliding window integration algorithm is used to accumulate the local evolution gradient vectors at different time scales. Simultaneously, the accumulated weights are dynamically adjusted by combining historical evolution inertia weights and current service disturbance coefficients to obtain the accumulated response intensity at multiple time scales. Finally, feature quantization modeling technology is used to normalize the vector direction of the accumulated response intensity, constructing an evolution direction descriptor that can accurately characterize the overall development trend of the spectral state, presented in unit vector form. At the same time, through the quantization and normalization of the vector magnitude, an evolution intensity descriptor reflecting the amplitude and rate of state change is generated, ultimately forming core parameters of spectral evolution that combine directional guidance and intensity quantification.

[0042] In one possible implementation, step S350 further includes:

[0043] Step S351: Based on the stable holding time, recovery speed and evolutionary regression probability of each spectral state component in the historical spectral state trajectory, construct the corresponding historical evolution inertia description quantity.

[0044] Step S352: Couple the historical evolution inertia description quantity with the current business behavior change. The coupling process is used to describe the disturbance intensity of the original evolution inertia of different spectral state components caused by sudden changes in business requests, adjustment of business type priority, and load continuity fluctuations.

[0045] Step S353: Based on the coupling process, configure direction-dependent evolution trend modulation parameters for each spectral state component in the spectral evolution tensor space to construct a non-uniform evolution potential distribution.

[0046] Specifically, a historical evolutionary inertia descriptor is constructed using time-series data mining, multi-index quantification, and weighted fusion. Historical spectrum state trajectory data is retrieved through a distributed time-series database interface. A sliding window analysis algorithm is used to segment the data into subdivided state components corresponding to spectrum occupancy, channel quality stability, interference change trends, and service load continuity. For each component, a state threshold determination method is used to statistically analyze the length of time series continuously satisfying stability conditions and calculate the mean and variance to obtain the stability maintenance duration. An exponential regression model is used to fit the curve of regression to stability after state anomalies, and the regression rate parameter is solved to obtain the recovery speed. A Markov chain state transition matrix is ​​used to statistically analyze the frequency percentage of regression from the current state to the previous state to determine the evolutionary regression probability. The analytic hierarchy process (AHP) is used in conjunction with the characteristics of spectrum resource usage scenarios to determine the weight coefficients of the three types of indicators, with the stability maintenance duration weighted at 0.4, and the recovery speed and evolutionary regression probability each weighted at 0.3. Min-max standardization is used to map all index data to the interval between 0 and 1 for normalization. Then, a weighted summation formula is used to multiply the normalized value of stable holding time by the corresponding weight, the normalized value of recovery speed by the corresponding weight, and the normalized value of evolution backsliding probability by the corresponding weight after inversion. The sum of these three values ​​yields the historical evolution inertia descriptor for each spectral state component. The higher the value of this descriptor, the stronger the evolution inertia of the component and the more prominent its anti-disturbance ability.

[0047] A coupled processing approach is adopted, combining support vector regression (SVR) based on kernel functions with an attention mechanism. First, the historical evolutionary inertia descriptor is used as the basic feature vector. Quantitative data corresponding to sudden changes in business requests, adjustments in business type priorities, and fluctuations in load continuity are used to construct a business disturbance feature matrix. An embedding layer unifies and standardizes the dimensions of both types of features. Then, an SVR model with a radial basis function (RBF) kernel is introduced to learn the nonlinear mapping relationship between historical evolutionary inertia and business disturbance features. Simultaneously, a multi-head attention mechanism is embedded to automatically assign attention weights to the three types of business changes on different spectral state components, strengthening the representation of the impact of key business disturbance factors. The trained model is used to infer and calculate the input features, converting the magnitude of sudden changes in business requests, the adjustment level of business type priorities, and the fluctuation frequency of load continuity into corresponding quantified disturbance coefficients. These coefficients are then weighted and fused with the historical evolutionary inertia descriptor through a fully connected network in the model's output layer, accurately quantifying the disturbance intensity of each type of business change on the original evolutionary inertia of different spectral state components, ultimately generating a structured disturbance intensity matrix.

[0048] A non-uniform evolutionary potential distribution is constructed using specific implementation methods such as dynamic parameter adaptation, tensor space potential energy modeling, and iterative optimization of the distribution. First, based on the perturbation intensity matrix obtained through coupling processing, the correlation between the perturbation intensity and evolutionary characteristics of each spectral state component is analyzed using a gradient boosting decision tree algorithm. For each component, direction-related evolutionary trend modulation parameters are dynamically configured, including an evolutionary attraction direction coefficient and an evolutionary suppression direction threshold. The attraction direction coefficient is determined based on the positive correlation between the business-driven direction and the perturbation intensity, while the suppression direction threshold is set based on a comprehensive evaluation of historical evolutionary inertia and perturbation impact. Subsequently, tensor decomposition and reconstruction techniques are used to embed the modulation parameters into the corresponding dimensions of the spectral evolution tensor space. The potential energy value of each spatial node is calculated using the potential energy field equation, and a Gaussian mixture model is used to cluster and optimize the potential energy distribution, strengthening the differentiated representation of the evolutionary directions of different components. Finally, by iteratively adjusting the potential energy field parameters, a non-uniform evolutionary potential distribution is formed within the tensor space that precisely matches the modulation parameters of each component, ensuring that different spectral state components exhibit the expected evolutionary attraction and suppression directions driven by changes in business behavior.

[0049] In one possible implementation, step S400 further includes:

[0050] Step S410: Match the evolution direction description with the pre-configured 5G network spectrum resource allocation rules to determine the set of allowed evolution directions of the spectrum status in the band switching direction, bandwidth expansion direction and carrier combination direction.

[0051] Step S420: Based on the CPE RF front-end capabilities, baseband processing capabilities, and RF switching response capabilities, perform capability constraint mapping on the evolution intensity description quantity to limit the maximum evolution amplitude of the spectrum state per unit time, thus forming the evolution intensity boundary.

[0052] Step S430: Based on the service continuity constraint, construct the spectrum state evolution smoothness constraint, which is used to limit the mutation rate and the number of consecutive jumps in the spectrum state during the evolution process.

[0053] Step S440: Based on the set of allowed evolution directions, the evolution intensity boundary, and the spectral state evolution smoothness constraint, generate a closed or semi-closed spectral state feasible evolution domain in the state evolution description space.

[0054] Specifically, the pre-configured 5G network spectrum resource allocation rules are retrieved first. These rules clearly define the allocation range of available 5G network frequency bands, bandwidth allocation standards for different service scenarios, compliant carrier combination modes, and priority specifications for spectrum resource use, comprehensively defining the legal boundaries of spectrum evolution. Subsequently, core information such as frequency band switching intentions, bandwidth expansion demands, and carrier combination preferences, contained in the previously constructed evolution direction descriptors, is extracted. A direction consistency verification algorithm is used to match and compare this evolution direction information with the 5G network spectrum resource allocation rules dimension by dimension. In the frequency band switching direction, switching options that conform to the available frequency band range in the rules are selected; in the bandwidth expansion direction, expansion paths that do not exceed the bandwidth allocation limit stipulated in the rules are determined; and in the carrier combination direction, combination methods consistent with the compliant modes in the rules are retained. After multiple rounds of matching verification and selection, a set of permissible evolution directions for spectrum status in three core dimensions is finally integrated, ensuring that subsequent spectrum evolution always follows the 5G network spectrum resource planning requirements and avoids resource conflicts or illegal use.

[0055] First, the hardware performance parameters of the CPE are analyzed to clarify core capabilities such as the signal reception and transmission power range and channel bandwidth processing limit of the RF front-end; processing capabilities such as the data processing rate and spectrum resource scheduling efficiency of the baseband processing unit; and response capabilities such as response latency and state switching success rate during RF handover. A complete CPE hardware capability parameter library is then constructed. Next, key information such as the spectrum state evolution rate and amplitude change contained in the evolution intensity descriptor is extracted. A quantitative matching relationship between evolution intensity and CPE hardware capabilities is established through a capability constraint mapping algorithm. The bandwidth evolution amplitude is limited by the channel bandwidth processing limit of the RF front-end, the spectrum parameter adjustment rate is constrained by the processing rate of the baseband processing unit, and the state switching time interval is standardized by the RF handover response latency. Based on this matching relationship, the maximum evolution amplitude threshold that the spectrum state can achieve per unit time is accurately calculated, forming a clear evolution intensity boundary. This ensures that the evolution intensity remains within the CPE hardware's carrying capacity during subsequent spectrum evolution, avoiding problems such as abnormal spectrum allocation, equipment overload, or unstable signal transmission due to exceeding hardware capabilities.

[0056] First, the core requirements for service continuity constraints are clearly defined. Considering the transmission stability requirements of 5G services, the anti-interference capabilities, latency tolerance thresholds, and transmission quality assurance standards corresponding to different service types, such as high-definition video, real-time communication, and data transmission, are analyzed to determine the boundaries of fluctuation risks to be avoided during spectrum state evolution. Then, based on the characteristics of service scenarios and transmission quality requirements, a spectrum state evolution smoothness constraint model is constructed. This model sets two key constraint indicators: an upper limit for the rate of change and a threshold for the number of consecutive jumps. The upper limit for the rate of change is quantified based on the service latency tolerance threshold, limiting the maximum range of spectrum parameter changes per unit time to avoid large fluctuations affecting service transmission in a short period. The threshold for the number of consecutive jumps is set in conjunction with the service's anti-interference capabilities, clarifying the maximum allowed number of discontinuous spectrum state jumps within the same evolution cycle to prevent frequent jumps from causing transmission link instability. This constraint model monitors the evolution rate and jump frequency of the spectrum state in real time, restricting evolutionary behavior exceeding the thresholds to ensure a smooth and controllable spectrum evolution process, providing a solid guarantee for service continuity.

[0057] The defined set of permissible evolution directions is transformed into spatial directional constraint boundaries, limiting the spectrum state to evolve only in compliant directions. The resulting evolution intensity boundaries are transformed into spatial amplitude limit thresholds, defining the maximum amplitude range of spectrum state evolution per unit time. The constructed spectrum state evolution smoothness constraints are transformed into spatial fluctuation control conditions, limiting the mutation rate and the number of consecutive jumps during the evolution process. These three types of constraints are fused using spatial modeling techniques, forming an interwoven constraint network within the state evolution description space. This ultimately generates a closed or semi-closed feasible evolution domain for the spectrum state. This domain clearly defines the legal range of subsequent spectrum state evolution, ensuring that all possible evolution paths simultaneously meet 5G network configuration requirements, CPE hardware capability limitations, and service continuity requirements.

[0058] In one possible implementation, step S500 further includes:

[0059] Step S510: Using the current spectrum state as the initial node, generate candidate spectrum state transition paths that satisfy the direction constraints and intensity boundaries within the feasible evolution domain of the spectrum state, wherein each candidate spectrum state transition path is composed of consecutive spectrum state evolution nodes.

[0060] Step S520: Along each candidate spectrum state migration path, evaluate the stage-by-stage evolution cost of the corresponding path in terms of consistency of evolution direction, stability of evolution intensity, and degree of maintenance of service continuity.

[0061] Step S530: Map the disturbance impact of changes in business behavior on spectrum demand into a dynamic correction factor for path weights, and adjust the comprehensive evaluation weight of the corresponding candidate path in real time during the path evolution process.

[0062] Step S540: After completing the full-process evaluation of all candidate spectrum state migration paths, construct the optimal spectrum state migration path based on the joint evaluation results of cumulative evolution cost and service disturbance adaptability.

[0063] Specifically, the process begins by defining the parameters of the current spectrum state and anchoring them as the initial nodes for path generation within the feasible evolution domain. Then, based on the defined set of permissible evolution directions and the evolution intensity boundaries within the feasible evolution domain, a path search algorithm is used to screen potential evolution nodes that meet the directional constraints. This means only nodes whose directions (band switching, bandwidth expansion, carrier combination, etc.) are within the permissible range are retained, while invalid nodes whose evolution amplitude exceeds the maximum limit per unit time are removed. Through neighborhood node association analysis, nodes that meet the constraints are linked together according to evolutionary logic, forming multiple continuous and compliant spectrum state evolution trajectories. Each trajectory consists of a series of interconnected spectrum state evolution nodes, ultimately generating multiple candidate spectrum state migration paths that meet the directional constraints and intensity boundary requirements, laying the foundation for subsequent path evaluation and optimal selection.

[0064] A three-dimensional cost evaluation index system is constructed, encompassing evolution direction consistency, evolution intensity stability, and the degree of business service continuity maintenance. Quantitative scoring standards and weighting ratios are set for each index. For each candidate spectrum state transition path, it is divided into several stages according to the time nodes or state transition nodes of the evolution process. Spectrum state parameter change data for each stage are extracted using a path trajectory analysis algorithm. In the dimension of evolution direction consistency, the direction cosine similarity algorithm is used to calculate the matching degree between the actual evolution direction and the set of allowed evolution directions for that stage, converting the degree of mismatch into a compliance cost score. In the dimension of evolution intensity stability, the sliding window statistical method is used to monitor the difference between the evolution amplitude and the evolution intensity boundary for that stage, quantifying the performance loss cost based on the excess amplitude and duration. In the dimension of business service continuity maintenance, a mutation detection algorithm is used to identify the mutation rate and the number of consecutive jumps for that stage. Combined with the stability requirements in the business service level agreement, the excess situation is converted into a business impact cost score. Finally, the cost scores of the three dimensions are fused using a weighted summation formula to obtain the stage-specific evolution cost for each candidate path, forming a detailed path evaluation dataset.

[0065] By capturing data on changes in business behavior, such as sudden changes in service requests, adjustments in service type priority, and fluctuations in load continuity, real-time data stream acquisition technology is used. A time-series prediction model based on Long Short-Term Memory (LSTM) networks is then used to analyze the correlation between these changes and spectrum requirements (bandwidth, frequency bands, and carrier combinations), transforming multi-dimensional service disturbances into standardized quantitative indicators. Through normalization, these quantitative indicators are mapped to dynamic correction factors for path weights. The sign and magnitude of these correction factors are dynamically set based on the direction and intensity of the impact of service disturbances on path adaptability; positive correction factors are generated when adaptability improves, and negative correction factors are generated when adaptability decreases. During path evolution evaluation, a real-time weight adjustment algorithm is embedded. After each stage of evolution cost evaluation, the dynamic correction factors are updated based on the latest service behavior change data. A weighted iterative formula is used to adjust the comprehensive evaluation weights of each candidate path in real time, ensuring that the weights accurately track the dynamic changes in service disturbances and that the path evaluation results remain highly aligned with actual business needs.

[0066] This paper employs cumulative cost accounting, adaptive quantification modeling, and multi-objective optimization ranking to construct the optimal spectrum state transition path. First, a path cost aggregation algorithm is used to accumulate the stage-by-stage evolution costs of each candidate path, yielding a cumulative evolution cost reflecting the overall path consumption. Simultaneously, a cosine similarity algorithm is used to quantify the degree of response and matching of each path to changes in business behavior during evolution, generating a business disturbance adaptability score. A weighted joint evaluation model is constructed, setting weight coefficients for cumulative evolution cost and business disturbance adaptability based on actual application scenario requirements. A linear weighted summation formula is used to calculate the joint evaluation score of each candidate path. The non-dominated sorting genetic algorithm NSGA-II is introduced to perform multi-objective optimization ranking of the joint evaluation scores of all candidate paths, selecting paths from the Pareto optimal solution set. If multiple paths are in a preset deviation range, a stable evolution segment detection algorithm is used to extract the stable evolution duration of each path, prioritizing paths with longer stable evolution segments. Finally, the optimal spectrum state transition path is determined, ensuring a balance between low evolution cost and high business adaptability.

[0067] In one possible implementation, step S510 further includes:

[0068] Step S511: Based on the geometric boundary structure of the feasible evolution domain of the spectrum state, perform hierarchical pruning on the search space of the candidate spectrum state transition path.

[0069] Step S512: Based on the hierarchical pruning results, configure the priority selection of the reachability density of the evolutionary domain to construct candidate spectral state transition paths.

[0070] Specifically, the geometric boundaries of the feasible evolution domain of the spectrum state are scanned using 3D point cloud modeling technology, and core feature parameters such as boundary vertices, normal vectors, and constraint thresholds are extracted to construct a complete boundary feature model. Subsequently, based on boundary features and constraint compliance, the K-means clustering algorithm is used to divide the search space into three levels: the core layer (the region that fully meets the triple constraints of 5G network configuration, CPE hardware capabilities, and service continuity), the transition layer (the region that is close to the constraint boundary but still compliant), and the edge layer (the critical compliance zone that is close to the constraint threshold). Finally, a spatial voxelization pruning algorithm is used to perform differentiated pruning on each level, retaining all effective space in the core layer. Redundant regions in the transition layer that are prone to constraint conflicts are filtered through boundary buffer thresholds, and invalid spaces in the edge layer that exceed the constraint threshold and pose compliance risks are directly eliminated. This achieves accurate compression and efficient screening of the search space, laying a lightweight foundation for the subsequent candidate path construction.

[0071] This paper employs reachability density quantization modeling, priority weight configuration, and path generation optimization to construct candidate spectrum state transition paths. First, using a kernel density estimation algorithm, the reachability density of evolutionary nodes in the effective space of the core, transition, and edge layers after hierarchical pruning is calculated. This quantifies the distribution density and connectivity feasibility of spectrum state evolution nodes within each layer, generating reachability density heatmaps for each layer. Based on hierarchical compliance and stability priorities, reachability density weight coefficients are configured for the core layer > transition layer > edge layer. Simultaneously, local weights are dynamically adjusted based on the constraint satisfaction of nodes within each layer, such as directional compliance rate and intensity compliance rate. An improved A / B algorithm is used. The path search algorithm takes the current spectrum state as the initial node, prioritizes screening connecting nodes from the core layer nodes with high reachability density, and then gradually expands to the transition layer and edge layer according to the weight coefficient. Through the constraint compatibility verification between nodes and the path smoothness optimization, multiple continuous and compliant candidate spectrum state migration paths are formed in series, ensuring that the path takes into account both search efficiency and meets the directional constraints and strength boundary requirements.

[0072] In one possible implementation, step S500 further includes:

[0073] Step S550: Perform monitoring and management of the actual spectrum state evolution and construct a monitoring dataset.

[0074] Step S560: Use the monitoring dataset and the optimal spectrum state migration path to identify deviation anomalies, establish deviation early warning, and perform early warning management based on the deviation early warning.

[0075] Specifically, after initiating dynamic spectrum allocation for CPE based on the optimal spectrum state migration path, a comprehensive real-time monitoring system is established. Following a preset time sampling interval, multi-source operational data related to the actual spectrum state evolution is continuously collected. The core spectrum parameters collected include real-time spectrum occupancy, frequency band switching execution status, actual bandwidth expansion, and carrier group matching status. Channel operational status data covers channel quality fluctuations, signal transmission rates, and real-time changes in interference intensity. CPE hardware operational data includes RF front-end operating power, baseband processing unit load rate, and RF switching response latency. Service adaptation data involves real-time feedback of service requests, transmission stability indicators, and service interruption frequency. The collected multi-dimensional data is standardized to eliminate dimensional differences, and then all data is systematically correlated and integrated according to timestamps. This ultimately constructs a well-structured and complete monitoring dataset, comprehensively and accurately capturing the complete evolution trajectory of the actual spectrum state from the initial node to the target state, providing reliable data support for subsequent deviation and anomaly identification and early warning management.

[0076] This system employs specific implementation methods such as data time-series alignment, multi-dimensional deviation quantification, tiered early warning triggering, and step-by-step verification optimization to perform deviation anomaly identification and early warning management. First, a dynamic time warping algorithm is used to precisely align the monitoring dataset with the preset evolution parameters of the optimal spectrum state migration path along the time dimension. Corresponding data for core dimensions such as spectrum state nodes, evolution time sequence, intensity changes, and service adaptability performance are extracted. Mean squared error combined with a dynamic threshold algorithm is used to calculate the deviation values ​​for each dimension. Based on a preset tiered deviation threshold system, a logical judgment model identifies deviation anomalies. When a single-dimensional deviation value exceeds the threshold or multiple-dimensional deviations exhibit coordinated anomalies, a tiered deviation early warning is immediately triggered, and the anomaly level, impact range, and potential risk points are marked. Based on the anomaly level, differentiated early warning response strategies are configured through the rule engine. Minor anomalies trigger a spectrum parameter fine-tuning mechanism, moderate anomalies execute local path optimization, and severe anomalies trigger optimal path reselection preparation. At the same time, a tiered verification window is set up. Within the window, a consistency evaluation algorithm is used to monitor the execution effect of the response strategy, and tiered verification feedback data such as the degree of improvement in service adaptation and spectrum status stability are collected. The response strategy parameters are dynamically adjusted through a feedback iterative algorithm to achieve closed-loop optimization of the early warning response and ensure that the spectrum status quickly returns to the optimal evolution trajectory.

[0077] In one possible implementation, step S560 further includes:

[0078] Step S561: Configure the warning response strategy according to the deviation warning and set the tiered verification window.

[0079] Step S562: Perform a consistency evaluation of the early warning response strategy in the tiered verification window, configure tiered verification feedback, and perform early warning response optimization based on the tiered verification feedback.

[0080] Specifically, after identifying deviations and establishing early warnings, a comprehensive assessment is conducted on the anomaly level indicated by the warning (e.g., minor, moderate, severe), the scope of impact (e.g., specific frequency bands, service types, or CPE hardware modules), and potential risk points (e.g., service interruption risks, hardware overload risks). Differentiated early warning response strategies are then configured accordingly. For minor anomalies, a spectrum parameter fine-tuning plan is developed, quickly correcting deviations by slightly adjusting parameters such as frequency band occupancy ratios and bandwidth allocation. For moderate anomalies, a local path optimization strategy is initiated, adjusting some evolution nodes in the optimal spectrum state migration path to adapt to constraints changes in actual operation. For severe anomalies, an optimal path reselection preparation process is triggered, while emergency spectrum resources are reserved to ensure the continuity of core services. Combining anomaly types (e.g., directional deviation, intensity exceeding limits, smoothness violation) with spectrum evolution cycle characteristics, multi-stage tiered verification windows are set up, clearly defining the verification duration and core monitoring indicators for each window, such as spectrum state return rate, service adaptation compliance rate, and compliance threshold. This establishes clear standards for subsequent accurate evaluation of the early warning response effectiveness.

[0081] Within each tiered verification window, core data such as spectrum state regression rate, business adaptation improvement rate, and constraint satisfaction are acquired in real time using multi-dimensional indicator collection technology. A consistency evaluation model is constructed using the analytic hierarchy process (AHP) to quantitatively score the execution effect of the early warning response strategy. Based on the evaluation results, tiered verification feedback is generated, including the achievement of the target, the trend of deviation changes, and potential optimization points. An intuitive feedback report is formed using data visualization technology. The verification feedback data is analyzed in depth using a feedback iteration algorithm. If the strategy execution effect does not reach the preset threshold of the window, the response strategy parameters are dynamically adjusted through a parameter tuning model, taking into account dynamic changes in business behavior and fluctuations in constraints. These adjustments include the fine-tuning of spectrum parameters, local path optimization logic, and re-selection path screening conditions. If the threshold is reached, the system enters the next stage of the verification window for continuous monitoring. The early warning response strategy is continuously iterated and improved through a rolling optimization mechanism to ensure that the spectrum state quickly and stably conforms to the optimal evolution trajectory.

[0082] In one possible implementation, step S500 further includes:

[0083] Step S570: When selecting the optimal spectrum state transition path, determine whether the comprehensive evaluation results of multiple candidate spectrum state transition paths meet the preset deviation range.

[0084] Step S580: If it exists, perform length bias selection of the stable evolution segment to construct the optimal spectral state transition path.

[0085] Specifically, after completing the full-process evaluation of all candidate spectrum state migration paths and obtaining the joint evaluation score of the cumulative evolution cost and service disturbance adaptability of each path, the final joint evaluation result of all candidate paths is extracted and compared one by one with a preset deviation interval to determine whether there are two or more candidate paths whose joint evaluation scores fall within this interval. The preset deviation interval is set according to the acceptable range of path performance differences in actual application scenarios and is used to screen out high-quality candidate paths with similar overall performance. If no such path is found, the path with the best joint evaluation score is directly selected as the final optimal spectrum state migration path.

[0086] When the comprehensive evaluation results of multiple candidate spectrum state transition paths fall within a preset deviation range, the complete evolution node sequence of each candidate path is first extracted using path trajectory analysis technology. Combined with the spectrum state evolution smoothness constraint requirements, stable evolution segments in each path that satisfy the requirements of consistent evolution direction, stable evolution intensity, no excessive mutation rate, and no violation of consecutive jump number rules are identified. A time-series interval statistical algorithm is used to accurately calculate the cumulative length of the stable evolution segments of each candidate path, while simultaneously assisting in verifying the average constraint satisfaction within the stable segments and the service adaptability stability index. The candidate path with the longest stable evolution segment is prioritized. If multiple paths have the same stable evolution segment length, the average constraint satisfaction within the stable segments is further compared, and the path with the higher average value is selected. Finally, the optimal spectrum state transition path with similar comprehensive performance and better operational stability is constructed.

[0087] Example 2, based on the same inventive concept as the 5G CPE dynamic spectrum allocation method in the previous examples, such as... Figure 2 As shown, this application provides a 5G CPE dynamic spectrum allocation system. The system and method embodiments in this application are based on the same inventive concept. The system includes:

[0088] The dataset acquisition module 10 is used to collect multi-source state datasets associated with the current spectrum usage within a preset time window, with the CPE as the spectrum decision-making entity. The multi-source state datasets include spectrum occupancy characteristics, channel quality stability characteristics, interference change trend characteristics, and service load continuity characteristics.

[0089] The instantaneous state vector construction module 20 is used to construct a spectral instantaneous state vector representing the operating status based on the multi-source state dataset.

[0090] The evolution intensity analysis module 30 is used to input the instantaneous state vector of the spectrum into the spectrum state evolution driving model and perform state evolution direction and evolution intensity analysis of the spectrum state caused by the combined effects of changes in the external environment, changes in business behavior, and historical usage results.

[0091] The feasible evolution domain construction module 40 is used to obtain the output results and then perform constraint mapping on the output results using 5G network configuration constraints, CPE hardware capability constraints and service continuity constraints to construct a feasible evolution domain of spectrum state with an allowable evolution range.

[0092] The dynamic spectrum allocation management module 50 is used to evaluate multiple candidate spectrum state transition paths within the feasible evolution domain of the spectrum state, select the optimal spectrum state transition path, and perform CPE dynamic spectrum allocation management based on the optimal spectrum state transition path.

[0093] Furthermore, the system is also used to implement the following functions:

[0094] The instantaneous spectrum state vector is mapped to a multidimensional spectrum evolution state tensor based on the mapping processing layer. This multidimensional spectrum evolution state tensor includes representations of spectrum occupancy, channel quality stability, interference variation trends, and service load continuity evolution potential at different time scales. After reading historical spectrum state trajectories, external environment gradients, and current service behavior changes, a potential field mapping channel is configured. These current service behavior changes include service request volume, service type priority, and load continuity fluctuations. The potential field mapping channel is used to assign evolution potential values ​​to each element of the multidimensional spectrum evolution state tensor. Based on these evolution potential values, the multidimensional spectrum evolution state... Tensor execution state extension processing is used to construct a joint representation structure of each spectral state component in the time dimension, business behavior disturbance dimension, and environmental gradient dimension, forming a spectral evolution state tensor space. A non-uniform evolution potential distribution is constructed within this tensor space, causing different spectral state components to exhibit differentiated evolutionary attraction and inhibition directions driven by changes in business behavior. Local evolution gradient vectors of the spectral state are calculated along the non-uniform evolution potential distribution, and evolutionary direction and intensity descriptors characterizing the evolutionary trend of the spectral state itself are constructed by accumulating the response intensity of local evolution gradient vectors at multiple time scales.

[0095] Furthermore, the system is also used to implement the following functions:

[0096] Based on the stable holding time, recovery speed, and evolutionary rollback probability of each spectrum state component in the historical spectrum state trajectory, a corresponding historical evolutionary inertia descriptor is constructed. The historical evolutionary inertia descriptor is coupled with the current business behavior changes. The coupling process is used to describe the disturbance intensity of business request mutations, business type priority adjustments, and load continuity fluctuations on the original evolutionary inertia of different spectrum state components. According to the coupling process, direction-related evolutionary trend modulation parameters are configured for each spectrum state component in the spectrum evolution tensor space to construct a non-uniform evolutionary potential distribution.

[0097] Furthermore, the system is also used to implement the following functions:

[0098] The evolution direction description is matched with the pre-configured 5G network spectrum resource allocation rules to determine the set of permissible evolution directions for the spectrum state in the band switching direction, bandwidth expansion direction, and carrier combination direction. Based on the CPE RF front-end capabilities, baseband processing capabilities, and RF switching response capabilities, the evolution intensity description is mapped with capability constraints to limit the maximum evolution amplitude of the spectrum state per unit time, forming an evolution intensity boundary. According to the service continuity constraints, a spectrum state evolution smoothness constraint is constructed, which is used to limit the mutation rate and the number of consecutive jumps of the spectrum state during the evolution process. Based on the permissible evolution direction set, the evolution intensity boundary, and the spectrum state evolution smoothness constraint, a closed or semi-closed feasible evolution domain of the spectrum state is generated in the state evolution description space.

[0099] Furthermore, the system is also used to implement the following functions:

[0100] Using the current spectrum state as the initial node, candidate spectrum state migration paths that satisfy directional constraints and intensity boundaries are generated within the feasible evolution domain of the spectrum state. Each candidate spectrum state migration path consists of consecutive spectrum state evolution nodes. Along each candidate spectrum state migration path, the stage-by-stage evolution cost of the corresponding path in terms of evolution direction consistency, evolution intensity stability, and the degree of maintenance of service continuity is evaluated. The disturbance impact of changes in service behavior on spectrum demand is mapped as a dynamic correction factor for path weights, and the comprehensive evaluation weight of the corresponding candidate path is adjusted in real time during the path evolution process. After completing the full-process evaluation of all candidate spectrum state migration paths, the optimal spectrum state migration path is constructed based on the joint evaluation results of cumulative evolution cost and service disturbance adaptability.

[0101] Furthermore, the system is also used to implement the following functions:

[0102] Based on the geometric boundary structure of the feasible evolutionary domain of the spectrum state, the search space of the candidate spectrum state transition path is hierarchically pruned; according to the hierarchical pruning result, the priority selection of the reachability density of the evolutionary domain is configured to construct the candidate spectrum state transition path.

[0103] Furthermore, the system is also used to implement the following functions:

[0104] Perform monitoring and management of the actual spectrum state evolution, and construct a monitoring dataset; use the monitoring dataset and the optimal spectrum state migration path to identify deviations and anomalies, establish deviation warnings, and perform warning management based on the deviation warnings.

[0105] Furthermore, the system is also used to implement the following functions:

[0106] Configure an early warning response strategy based on the deviation early warning and set a tiered verification window; perform a consistency evaluation of the early warning response strategy in the tiered verification window, configure tiered verification feedback, and perform early warning response optimization based on the tiered verification feedback.

[0107] Furthermore, the system is also used to implement the following functions:

[0108] When selecting the optimal spectrum state migration path, it is determined whether the comprehensive evaluation results of multiple candidate spectrum state migration paths meet the preset deviation interval; if so, the length bias selection of the stable evolution segment is performed to construct the optimal spectrum state migration path.

[0109] It should be noted that the order of the embodiments described above is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. Furthermore, the above description focuses on specific embodiments of this specification. Additionally, the processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired results. In some implementations, multitasking and parallel processing are possible or may be advantageous.

[0110] The above description is only a preferred embodiment of this application and is not intended to limit this application. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

[0111] This specification and accompanying drawings are merely illustrative examples of this application and are intended to cover any and all modifications, variations, combinations, or equivalents within the scope of this application. Clearly, those skilled in the art can make various alterations and modifications to this application without departing from its scope. Therefore, if such modifications and variations fall within the scope of this application and its equivalents, this application intends to include such modifications and variations.

Claims

A 1.5G CPE dynamic spectrum allocation method, characterized in that, The method includes: Using CPE as the spectrum decision-making entity, a multi-source state dataset associated with the current spectrum usage is collected within a preset time window. The multi-source state dataset includes spectrum occupancy characteristics, channel quality stability characteristics, interference change trend characteristics, and service load continuity characteristics. Construct a spectral instantaneous state vector representing the operational status based on the multi-source state dataset; The instantaneous state vector of the spectrum is input into the spectrum state evolution driving model to perform state evolution direction and intensity analysis of the spectrum state caused by the combined effects of changes in the external environment, changes in business behavior, and historical usage results; After obtaining the output results, the constraint mapping of the output results is performed using 5G network configuration constraints, CPE hardware capability constraints, and service continuity constraints to construct a spectrum state feasible evolution domain with an allowable evolution range. Within the feasible evolution domain of the spectrum state, multiple candidate spectrum state transition paths are evaluated, the optimal spectrum state transition path is selected, and CPE dynamic spectrum allocation management is performed based on the optimal spectrum state transition path.

2. The 5G CPE dynamic spectrum allocation method as described in claim 1, characterized in that, After inputting the instantaneous spectral state vector into the spectral state evolution driving model, the following steps are included: The instantaneous spectrum state vector is mapped to a multidimensional spectrum evolution state tensor based on the mapping processing layer. The multidimensional spectrum evolution state tensor includes the evolution potential of spectrum occupancy, channel quality stability, interference change trend and service load continuity at different time scales. After reading the historical spectrum status trajectory, external environment gradient, and current business behavior changes, configure the potential field mapping channel. The current business behavior changes include business request volume, business type priority, and load continuity fluctuations. The potential energy value is assigned to each element of the multidimensional spectral evolution state tensor using the potential energy field mapping channel; Based on the evolution potential value, state expansion processing is performed on the multidimensional spectrum evolution state tensor to construct a joint representation structure containing each spectrum state component in the time dimension, service behavior disturbance dimension and environmental gradient dimension, forming a spectrum evolution state tensor space. A non-uniform evolution potential distribution is constructed in the tensor space of the spectral evolution state. The non-uniform evolution potential distribution causes different spectral state components to exhibit differentiated evolution attraction and inhibition directions under the drive of changes in business behavior. The local evolution gradient vector of the spectral state is calculated along the non-uniform evolution potential distribution, and the evolution direction descriptor and evolution intensity descriptor, which characterize the evolution trend of the spectral state itself, are constructed by the cumulative response intensity of the local evolution gradient vectors at multiple time scales.

3. The 5G CPE dynamic spectrum allocation method as described in claim 2, characterized in that, Constructing a non-uniform evolution potential distribution within the tensor space of the spectral evolution state includes: Based on the stable holding time, recovery speed and evolutionary regression probability of each spectral state component in the historical spectral state trajectory, the corresponding historical evolutionary inertial description quantity is constructed. The historical evolution inertia descriptor is coupled with the current business behavior change. The coupling process is used to describe the disturbance intensity of the original evolution inertia of different spectral state components caused by sudden changes in business requests, adjustments to business type priorities, and load continuity fluctuations. Based on the coupling process, direction-dependent evolution trend modulation parameters are configured for each spectral state component in the spectral evolution tensor space to construct a non-uniform evolution potential distribution.

4. The 5G CPE dynamic spectrum allocation method as described in claim 3, characterized in that, By leveraging 5G network configuration constraints, CPE hardware capability constraints, and service continuity constraints, constraint mapping of the output results is performed to construct a feasible evolution domain for the spectrum state with an allowable evolution range, including: The evolution direction description is matched with the pre-configured 5G network spectrum resource allocation rules to determine the set of allowed evolution directions of the spectrum status in the band switching direction, bandwidth expansion direction, and carrier combination direction. Based on the CPE RF front-end capabilities, baseband processing capabilities, and RF switching response capabilities, capability constraint mapping is performed on the evolution intensity description quantity to limit the maximum evolution amplitude of the spectrum state per unit time, thus forming the evolution intensity boundary. Based on the business service continuity constraint, a spectrum state evolution smoothness constraint is constructed. The spectrum state evolution smoothness constraint is used to limit the mutation rate and the number of consecutive jumps in the spectrum state during the evolution process. Based on the set of allowed evolution directions, the boundary of evolution intensity, and the constraint of spectral state evolution smoothness, a closed or semi-closed spectral state feasible evolution domain is generated in the state evolution description space.

5. The 5G CPE dynamic spectrum allocation method as described in claim 1, characterized in that, Within the feasible evolution domain of the spectral state, multiple candidate spectral state transition paths are evaluated, and the optimal spectral state transition path is selected, including: Using the current spectrum state as the initial node, candidate spectrum state transition paths that satisfy the direction constraints and intensity boundaries are generated within the feasible evolution domain of the spectrum state. Each candidate spectrum state transition path consists of consecutive spectrum state evolution nodes. Along each candidate spectrum state migration path, the stage-by-stage evolution costs of the corresponding path in terms of consistency of evolution direction, stability of evolution intensity, and degree of maintenance of business service continuity are evaluated. The impact of changes in business behavior on spectrum demand is mapped as a dynamic correction factor for path weights, and the comprehensive evaluation weight of the corresponding candidate path is adjusted in real time during the path evolution process. After completing the full-process evaluation of all candidate spectrum state migration paths, the optimal spectrum state migration path is constructed based on the joint evaluation results of cumulative evolution cost and service disturbance adaptability.

6. The 5G CPE dynamic spectrum allocation method as described in claim 5, characterized in that, Using the current spectral state as the initial node, candidate spectral state transition paths satisfying direction constraints and intensity boundaries are generated within the feasible evolution domain of the spectral state, including: Based on the geometric boundary structure of the feasible evolution domain of the spectrum state, the search space of the candidate spectrum state transition path is hierarchically pruned. Based on the hierarchical pruning results, the priority of the reachability density of the evolutionary domain is configured to construct candidate spectral state transition paths.

7. The 5G CPE dynamic spectrum allocation method as described in claim 1, characterized in that, CPE dynamic spectrum allocation management based on the optimal spectrum state transition path includes: Perform monitoring and management of actual spectrum state evolution, and construct monitoring datasets; The monitoring dataset and the optimal spectrum state migration path are used to identify deviation anomalies, establish deviation early warning, and perform early warning management based on the deviation early warning.

8. The 5G CPE dynamic spectrum allocation method as described in claim 7, characterized in that, Based on the aforementioned deviation warning, early warning management is performed, including: Configure the early warning response strategy according to the aforementioned deviation early warning, and set a tiered verification window; The consistency evaluation of the early warning response strategy is performed in the tiered verification window, the tiered verification feedback is configured, and the early warning response is optimized based on the tiered verification feedback.

9. The 5G CPE dynamic spectrum allocation method as described in claim 1, characterized in that, Selecting the optimal spectrum state transition path includes: When selecting the optimal spectrum state transition path, it is determined whether the comprehensive evaluation results of multiple candidate spectrum state transition paths meet the preset deviation range. If it exists, then perform length bias selection of the stable evolution segment to construct the optimal spectral state transition path. The 10.5G CPE dynamic spectrum allocation system is characterized by: The system is used to implement the 5G CPE dynamic spectrum allocation method according to any one of claims 1-9, and the system comprises: The dataset acquisition module is used to collect multi-source state datasets related to the current spectrum usage within a preset time window, with CPE as the spectrum decision-making entity. The multi-source state datasets include spectrum occupancy characteristics, channel quality stability characteristics, interference change trend characteristics, and service load continuity characteristics. The instantaneous state vector construction module is used to construct a spectral instantaneous state vector representing the operating status based on the multi-source state dataset; The evolution intensity analysis module is used to input the instantaneous state vector of the spectrum into the spectrum state evolution driving model, and perform state evolution direction and evolution intensity analysis of the spectrum state caused by the combined effects of changes in the external environment, changes in business behavior, and historical usage results. The feasible evolution domain construction module is used to obtain the output results and then perform constraint mapping on the output results using 5G network configuration constraints, CPE hardware capability constraints, and service continuity constraints to construct a spectrum state feasible evolution domain with an allowable evolution range. The dynamic spectrum allocation management module is used to evaluate multiple candidate spectrum state transition paths within the feasible evolution domain of the spectrum state, select the optimal spectrum state transition path, and perform CPE dynamic spectrum allocation management based on the optimal spectrum state transition path.

Images