A low-latency slice resource intelligent allocation method for a 5G network

Abstract

The application discloses a low-latency slice resource intelligent allocation method for a 5G network, which comprises the following steps: collecting the connection state and performance index of a network node in real time, dynamically adjusting the collection frequency according to network load fluctuation, and generating structured topology data; calculating the node degree distribution entropy value based on the topology data, generating a virtual resource topology graph, and predicting the resource demand of the network node; constructing a dynamic confidence interval by using the virtual resource topology graph and the entropy value change rate, designing a resource pre-allocation strategy; detecting network burst traffic, quickly adjusting resource allocation through incremental update of a graph neural network, and ensuring that the reconfiguration latency is lower than 1 millisecond; through topology perception driven by the graph neural network, prediction guided by the entropy value and the adaptive allocation strategy, the low-latency resource allocation of the uRLLC slice is realized, resource waste is avoided, and the strict performance requirements of the 5G network are met.

Description

Technical Field

[0001] This invention relates to the field of network slice resource management technology, and in particular to a method for intelligent allocation of low-latency slice resources for 5G networks. Background Technology

[0002] With the widespread deployment of fifth-generation (5G) mobile communication networks, network slicing technology has become a core technology supporting diverse service scenarios, such as enhanced mobile broadband (eMBB), ultra-reliable low-latency communication (uRLLC), and massive machine-type communication (mMTC). Network slicing achieves dynamic resource allocation through software-defined networking (SDN) and network function virtualization (NFV), providing customized Quality of Service (QoS) for different applications. Furthermore, the rapid development of artificial intelligence (AI) technology in recent years has demonstrated significant potential in network topology modeling and resource allocation optimization. By integrating communication technology with AI technology, positive progress has been made in improving network efficiency and supporting diverse services.

[0003] However, existing technologies still have some shortcomings, significantly limiting their application in low-latency uRLLC slicing scenarios. First, traditional resource allocation methods (mainly focused on static or semi-static scenarios) lack the ability to perceive real-time network topology changes, resulting in high reconfiguration latency during sudden traffic surges or topology changes (such as base station handover or VNF migration), making it difficult to meet the stringent latency requirements of uRLLC slicing. Second, while existing machine learning-based methods can handle complex network topologies, most rely on historical data prediction, neglecting the real-time quantification of network dynamic complexity, leading to an inability to accurately predict resource demands or quickly respond to traffic surges. Third, existing methods lack adaptability in resource pre-allocation, failing to dynamically adjust allocation strategies based on network conditions, resulting in low resource utilization and large latency fluctuations. Summary of the Invention

[0004] The purpose of this section is to outline some aspects of embodiments of the present invention and to briefly describe some preferred embodiments. Simplifications or omissions may be made in this section, as well as in the abstract and title of this application, to avoid obscuring the purpose of these documents; however, such simplifications or omissions should not be construed as limiting the scope of the invention.

[0005] In view of the aforementioned existing problems, this invention is proposed. Therefore, this invention provides a low-latency slice resource intelligent allocation method for 5G networks to solve the problems mentioned in the background art.

[0006] To address the aforementioned technical problems, this invention provides the following technical solution: a method for intelligent allocation of low-latency slice resources for 5G networks, comprising:

[0007] The system collects the connection status and performance indicators of network nodes in real time, and dynamically adjusts the collection frequency according to network load fluctuations to generate structured topology data.

[0008] Based on the topology data, calculate the degree distribution entropy value of the network nodes, generate a virtual resource topology map, and predict the resource requirements of the network nodes.

[0009] Based on the virtual resource topology map and the entropy change rate, a dynamic confidence interval is constructed, and a resource pre-allocation strategy is designed.

[0010] The system detects sudden traffic spikes in the network and rapidly adjusts resource allocation and reduces reconfiguration latency by incrementally updating the graph neural network.

[0011] As a preferred embodiment of the intelligent allocation method for low-latency slice resources in 5G networks described in this invention, the method includes: real-time collection of network node connection status and performance indicators, dynamic adjustment of the collection frequency based on network load fluctuations, and generation of structured topology data, including:

[0012] By collecting connection status data of network nodes through distributed network probes, an adjacency matrix of connection strength between nodes is formed;

[0013] Collect performance metrics for each node, including bandwidth utilization, computational load, and latency, to form a node feature matrix;

[0014] The network load fluctuation is calculated based on the fluctuation rate of the bandwidth utilization, and the sampling frequency of the distributed network probe is dynamically adjusted.

[0015] The adjacency matrix and node feature matrix are denoised and normalized to generate structured topology data.

[0016] As a preferred embodiment of the intelligent allocation method for low-latency slice resources in 5G networks described in this invention, the following steps are taken: Calculating the degree distribution entropy value of network nodes based on the topology data includes:

[0017] Based on the adjacency matrix, calculate the degree of each node, which is the sum of the connection strengths between each node and its neighboring nodes;

[0018] Based on the degree of the nodes, a degree distribution function for smoothing the nodes is constructed, and the entropy value of the structured topology data is calculated.

[0019] By fusing the entropy values ​​of the current structured topology data with their historical entropy values ​​using a time-weighted method, a smooth and effective entropy value is generated.

[0020] Calculate the rate of change of the effective entropy value and store the effective entropy value and the rate of change. When the effective entropy value exceeds a preset threshold, trigger the generation operation of the virtual resource topology map.

[0021] As a preferred embodiment of the intelligent allocation method for low-latency slice resources in 5G networks described in this invention, the generation of a virtual resource topology map is triggered when the effective entropy value exceeds a preset threshold, including:

[0022] The preset threshold for the effective entropy value is obtained by calculating the mean and standard deviation of the effective entropy value and the network topology change coefficient.

[0023] As a preferred embodiment of the intelligent allocation method for low-latency slice resources in 5G networks described in this invention, the generation of the virtual resource topology map includes:

[0024] Based on the structured topology data, a network topology graph is constructed, where nodes represent network nodes, edges represent connection relationships, and node features are performance indicators.

[0025] A graph neural network is applied to process the network topology graph through a message passing mechanism to generate a virtual resource topology graph, which is then represented as a resource demand matrix. The resource demand matrix includes the bandwidth and computing resource requirements of each node.

[0026] As a preferred embodiment of the intelligent allocation method for low-latency slice resources in 5G networks described in this invention, the method includes: constructing a dynamic confidence interval and designing a resource pre-allocation strategy, including:

[0027] Based on the resource demand matrix, calculate the predicted resource demand and its fluctuation level for each node;

[0028] Based on the predicted resource demand and volatility, and combined with the rate of change of effective entropy, the confidence level is dynamically adjusted to construct the resource demand confidence interval for each node.

[0029] Design a resource pre-allocation strategy that prioritizes meeting the boundary requirements of the resource demand confidence interval, while constraining the total resource demand to not exceed the available network resources.

[0030] The allocation results obtained after executing the resource pre-allocation strategy will be stored, and the current network resource status will be updated.

[0031] As a preferred embodiment of the low-latency slice resource intelligent allocation method for 5G networks described in this invention, the method includes: detecting burst traffic in the network and rapidly adjusting resource allocation and reducing reconfiguration latency by incrementally updating the graph neural network, comprising:

[0032] Based on the time-series changes in bandwidth utilization, a traffic mutation index is calculated to determine whether there is a sudden traffic surge. When the traffic mutation index exceeds a preset threshold, the affected subgraph is identified.

[0033] By incrementally updating the graph neural network, the nodes of the affected subgraph are recalculated to generate a new resource demand matrix;

[0034] Based on the aforementioned resource demand confidence interval, resource reconfiguration is optimized to meet the high-priority node requirements of uRLLC slices while maintaining a reconfiguration latency of less than 1 millisecond.

[0035] The results of resource reconfiguration are fed back to the current network resource status.

[0036] As a preferred embodiment of the intelligent allocation method for low-latency slice resources in 5G networks according to the present invention, wherein: when the traffic mutation index exceeds a preset threshold, the following applies:

[0037] The preset threshold for the flow mutation index is obtained by calculating the mean and twice the standard deviation of the flow mutation index within a time window.

[0038] Compared with existing technologies, the beneficial effects of the invention are:

[0039] 1. Through the real-time topology awareness and incremental update mechanism driven by graph neural networks, this invention can quickly respond to sudden traffic and network topology changes, ensuring that the resource reconfiguration latency is less than 1 millisecond, meeting the strict latency requirements of 5G ultra-reliable low-latency communication (uRLLC) slicing, and overcoming the limitation of high reconfiguration latency in dynamic scenarios by traditional static or semi-static resource allocation methods.

[0040] 2. By using the node degree distribution entropy value to quantify the network topology complexity and combining it with the dynamic confidence interval strategy, resources are predicted and pre-allocated in real time, giving priority to meeting the needs of high-priority nodes (such as uRLLC slice VNF), which improves the accuracy of resource allocation and adapts to the dynamic complexity of 5G networks.

[0041] 3. By using adaptive sampling frequency adjustment and entropy-guided graph neural network resource prediction, this invention ensures low latency while dynamically constraining the total resource allocation to not exceed the available network resources, thus avoiding resource waste. Attached Figure Description

[0042] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the 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. Wherein:

[0043] Figure 1 This is a flowchart illustrating the overall process of a low-latency slice resource intelligent allocation method for 5G networks according to an embodiment of the present invention. Detailed Implementation

[0044] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the protection scope of the present invention.

[0045] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0046] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0047] This invention is described in detail with reference to the schematic diagrams. When describing the embodiments of this invention, for ease of explanation, the cross-sectional views illustrating the device structure may be partially enlarged, not according to the usual scale. Furthermore, the schematic diagrams are merely examples and should not be construed as limiting the scope of protection of this invention. In actual fabrication, the three-dimensional spatial dimensions of length, width, and depth should be included.

[0048] Furthermore, in the description of this invention, it should be noted that the terms "upper," "lower," "inner," and "outer," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. These terms are used solely for the convenience of describing the invention and for simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on the invention. In addition, the terms "first," "second," or "third" are used for descriptive purposes only and should not be construed as indicating or implying relative importance.

[0049] Unless otherwise explicitly specified and limited, the terms "installation," "connection," and "joining" in this invention should be interpreted broadly. For example, they can refer to fixed connections, detachable connections, or integral connections; similarly, they can refer to mechanical connections, electrical connections, or direct connections, or indirect connections through an intermediate medium, or internal connections between two components. Those skilled in the art can understand the specific meaning of the above terms in this invention based on the specific circumstances.

[0050] Example 1

[0051] Reference Figure 1 This is the first embodiment of the present invention, which provides a method for intelligent allocation of low-latency slice resources for 5G networks, including:

[0052] S1. Real-time collection of network node connection status and performance indicators, and dynamic adjustment of collection frequency according to network load fluctuations to generate structured topology data;

[0053] Furthermore, deploy distributed network probes (such as SNMP-based sensors) and integrate them into the 5G core network's Network Function Virtualization (NFV) infrastructure, collecting node connection status and node performance metrics every time step Δt (initially 100ms).

[0054] Furthermore, an adjacency matrix is ​​formed based on the connection status of the data acquisition nodes (the logical connections between nodes). N is the number of nodes;

[0055] Specifically, the elements in this adjacency matrix contain the connection strength between nodes:

[0056]

[0057] Among them, A ij (t) represents the connection strength between nodes i and j; B ij (t) represents the link bandwidth utilization; B max Represented as maximum bandwidth; L ij (t) represents the link delay; L ref For reference latency (default 1ms), to support uRLLC slicing, the weight of low-latency links is prioritized (by reducing L). ref up to 0.5ms);

[0058] Furthermore, the collected node performance metrics are stored as node feature vectors, where the node performance metrics include bandwidth utilization, computational load, and latency. These node feature vectors are then used to form a node feature matrix. 3 represents the dimension of the node feature vector;

[0059] Specifically, the node feature vector x i (t) is represented as:

[0060] x i (t)=[B i (t),C i (t),D i (t)]

[0061] Among them, B i (t) represents the bandwidth utilization of each node, C i(t) represents the computational load (specifically, CPU / GPU utilization), D i (t) represents the delay;

[0062] Furthermore, the network load fluctuation is calculated based on the fluctuation rate of bandwidth utilization, and the sampling frequency of the distributed network probes is dynamically adjusted.

[0063] Specifically, the formula for calculating network load fluctuation is as follows:

[0064]

[0065] in, The volatility of bandwidth utilization, derived from the bandwidth utilization of each node, can be expressed as: ∈ t This represents the degree of network load fluctuation. When it exceeds the current upper limit of network load fluctuation (determined based on the deployment environment of the 5G core network), the collection frequency of the distributed network probe is shortened to Δt′=Δt / (1+∈ t Otherwise, extend to Δt′=min(1.5Δt,500ms);

[0066] Furthermore, the adjacency matrix and node feature matrix are denoised and normalized to generate structured topological data.

[0067] Specifically, the denoising method uses wavelet transform (Daubechies wavelet, decomposition level 4) to filter out high-frequency noise in order to preserve the variation trend of the structured topology data;

[0068] Specifically, A ij (t) is normalized to the interval [0,1], and x is then normalized based on the time window [t-10Δt,t]. i Each dimension of (t) is standardized to zero mean and unit variance;

[0069] It should be noted that since denoising and normalization are routine data processing operations and are not related to the innovative direction of this invention, they will not be described in detail.

[0070] Specifically, the data after denoising and normalization is stored (e.g., in Redis) to support fast queries;

[0071] It should be noted that the data collection operation of the dynamic distributed network probe based on the degree of network load fluctuation can effectively reduce the amount of redundant data and enhance the accuracy of structured topology data representation.

[0072] S2. Based on the topology data, calculate the degree distribution entropy value of the network nodes, generate a virtual resource topology map, and predict the resource requirements of the network nodes.

[0073] Furthermore, based on the adjacency matrix in the structured topology data, the degree of each node is calculated as the sum of the connection strengths between each node and its neighboring nodes.

[0074] Specifically, assuming node j is a neighbor of node i, then the degree d of each node... i (t):

[0075]

[0076] Furthermore, based on the degree of the nodes, a smooth node degree distribution function is constructed, and the entropy value of the structured topology data is calculated.

[0077] Specifically, the node degree distribution is constructed and smoothed using kernel density estimation (emphasizing a Gaussian kernel function, with the kernel density estimation bandwidth expressed as h = 0.1·std(d), i.e., 0.1 times the standard deviation of the node degree, chosen empirically).

[0078]

[0079] Where P(d,t) represents the degree distribution function of the smooth node, which can reflect changes in structured topology data (such as base station handover or VNF migration) in uRLLC slicing scenarios; K is the Gaussian kernel function;

[0080] It should be noted that utilizing the smoothing properties of the Gaussian kernel function helps to filter short-term noise (such as transient connection breaks) in structured topological data in uRLLC scenarios, thereby improving the robustness of entropy calculation.

[0081] Furthermore, the entropy value of the current structured topology data is fused with its historical entropy value using a time-weighted method to generate a smooth effective entropy value;

[0082] Specifically, the entropy value of the current structured topology data is calculated as follows:

[0083]

[0084] It should be noted that, based on the Shannon entropy formula, the information contribution of each node's degree value d is quantified by logP(d,t) to show the change in network topology complexity, which provides support for the prediction of resource demand for uRLLC slices.

[0085] Specifically, time weighting is introduced to generate a smooth effective entropy value H. eff (t):

[0086] H eff (t)=(1-α)H(t)+αH(t-1)

[0087] Where α is the time weighting factor, usually taken as 0.7. By adjusting the time weighting factor, the smoothness is controlled, thereby reducing the impact of short-term fluctuations on the entropy value and improving the reliability of triggering the generation of the virtual resource topology map; H(t-1) refers to the smoothed entropy value of the previous time step, which is incorporated into the current calculation as a representative historical entropy value.

[0088] Furthermore, the rate of change of the effective entropy value is calculated and stored. When the effective entropy value exceeds a preset threshold, the generation operation of the virtual resource topology map is triggered.

[0089] Specifically, the rate of change of the effective entropy value is obtained by comparing the current effective entropy value with the effective entropy value of the previous time step:

[0090] ΔH(t)=H eff (t)-H eff (t-1)

[0091] Wherein, ΔH(t) represents the rate of change of the effective entropy value;

[0092] It should be noted that the rate of change of the effective entropy value is mainly used for adjusting the dynamic confidence interval, thereby connecting the entropy value with the uRLLC slice resource allocation;

[0093] Specifically, the preset threshold for effective entropy is obtained by statistically analyzing the time series distribution of entropy values ​​in the current structured topology data, calculating the mean and standard deviation of the effective entropy, and the network topology change coefficient (k = 1.5):

[0094]

[0095] Where, ΔH th (t) represents the preset threshold for the effective entropy value. This represents the mean of the effective entropy values. The standard deviation of the effective entropy value;

[0096] Specifically, the time series distribution of the entropy value of the current structured topology data refers to the entropy value of the current structured topology data when there are significant changes in the network topology (such as node disconnection / addition, sudden changes in traffic patterns);

[0097] It should be noted that because uRLLC slicing is sensitive to changes in network topology, a threshold for effective entropy needs to be preset to ensure that resource prediction is triggered in a timely manner when the complexity of network topology increases significantly, thereby meeting the requirements for low latency.

[0098] S3. Based on the virtual resource topology map and entropy change rate, construct a dynamic confidence interval and design a resource pre-allocation strategy;

[0099] Furthermore, based on structured topology data, a network topology graph is constructed, where nodes represent network nodes, edges represent connection relationships, and node features are performance indicators.

[0100] Specifically, a network topology diagram can be represented by the formula:

[0101] G(t)=(V,E,X)

[0102] Where V is the node set, E is the edge set representing the connection relationship between nodes, defined by the adjacency matrix A(t), and X is the node feature, obtained from the node feature matrix;

[0103] Furthermore, by applying graph neural networks, the network topology graph is processed through a message passing mechanism to generate a virtual resource topology graph, which is then represented as a resource demand matrix. The resource demand matrix contains the bandwidth and computing resource requirements of each node.

[0104] Specifically, the applied Graph Neural Network (GNN) consists of two layers of Graph Convolutional Network (GCN), with each layer having an output dimension of 64 (F′=64). First, the GNN is initialized by inputting node features, adjacency matrix, and effective entropy values ​​from the network topology graph. The pre-trained GNN model is then loaded using [t-10Δt,t] to optimize the weight matrix. Then, the GCN forward propagation is performed using a message passing mechanism to update the node embeddings. The first-layer GCN aggregates neighbor node information, and the second-layer GCN receives the aggregation results from the first-layer GCN and performs secondary aggregation to obtain the final node embeddings. Finally, the node embeddings are transformed into resource requirements through a fully connected layer and output by the output layer to generate a virtual resource topology graph.

[0105] Specifically, the updated node embedding is expressed by the formula:

[0106]

[0107] in, This represents the embedding vector of node i in the (l+1)th layer, with an initial value of N i Let A(t) represent the set of neighbors of node i. ω represents the weight matrix of the l-th layer; j (t) represents the attention weight, used to dynamically adjust the contribution of neighboring node j. Based on the node's degree and effective entropy value, it can enhance the resource prediction priority of nodes in high-entropy regions and can be further expressed as:

[0108]

[0109] Specifically, the fully connected layer embeds nodes. Transformed into resource requirements:

[0110]

[0111] Among them, R i (t) represents the resource demand vector of node i, including bandwidth and computing resources (dimension 2); W out Represented as the output layer weight matrix; b out Represented as a bias vector;

[0112] Specifically, the virtual resource topology graph generated by the GNN model is represented as a resource demand matrix. Resource pre-allocation is directly used for uRLLC slices to prioritize low-latency requirements;

[0113] Furthermore, based on the resource demand matrix, the predicted resource demand for each node and its fluctuation level are calculated;

[0114] Specifically, predicting resource demand Represented as:

[0115]

[0116] Specifically, predicting the degree of fluctuation in resource demand ∈ R (t) is similar to the aforementioned network load fluctuation, and can be expressed as:

[0117]

[0118] in, This represents the predicted volatility of resource demand, obtained from the resource demand vector of each node, and can be expressed as:

[0119] Furthermore, based on the predicted resource demand and its volatility, and combined with the rate of change of effective entropy, the confidence level is dynamically adjusted to construct the resource demand confidence interval for each node;

[0120] Specifically, the constructed confidence interval CI for resource demand. i (t) is represented as:

[0121]

[0122] It should be noted that in the uRLLC scenario, the resource demand confidence interval needs to ensure that the resource demand of high-priority nodes (such as VNFs) is met in order to avoid increased latency due to errors in predicting resource demand.

[0123] Specifically, z(t) represents the dynamic confidence level, which is adjusted based on the rate of change of the effective entropy value:

[0124] z(t)=z0+γ·tanh(κ·|ΔH(t)|)

[0125] Where z0 is the initial confidence level, with a value of 1.96 (corresponding to a 95% confidence level, based on a standard normal distribution); γ is used to adjust the magnitude of the change in confidence level, with a value of 0.5; and κ is used to control the sensitivity of the entropy rate of change, with a value of 2.

[0126] It should be noted that when the network topology complexity changes significantly (|ΔH(t)| increases), z(t) increases, widening the confidence interval for resource demand and thus enhancing the conservatism of slice resource allocation (i.e., allocating more resources to cover the boundary demand and reducing the risk of resource shortage due to errors in predicted resource demand). When the network topology is stable, z(t) is close to z0, resulting in a narrower confidence interval for resource demand and thus more accurate resource demand prediction. In this case, optimization algorithms (such as linear programming solvers) can more closely approximate the predicted value. This reduces resource waste;

[0127] Furthermore, a resource pre-allocation strategy is designed to prioritize meeting the boundary requirements of the resource demand confidence interval, while constraining the total resource demand to not exceed the available network resources.

[0128] Specifically, the confidence interval is input into the optimization algorithm to establish the optimization objective:

[0129]

[0130] in, This represents the actual resource allocation, obtained by an optimization algorithm. It is directly applied to the SDN / NFV architecture of 5G networks, and the resource configuration of Virtual Network Functions (VNFs) is adjusted through the Management and Orchestration (MANO) system to achieve resource allocation.

[0131] Specifically, establish the constraints for the optimization objective:

[0132] (Total resource requirements shall not exceed the available network resources)

[0133] (Resource allocation results are within the resource demand confidence interval)

[0134] Among them, R total This is expressed as the total resource requirement;

[0135] Specifically, the allocation results obtained after executing the resource pre-allocation strategy will be stored, and the current network resource status will be updated.

[0136] S4. Detect bursts of network traffic and quickly adjust resource allocation and reduce reconfiguration latency by incrementally updating the graph neural network.

[0137] Furthermore, based on the time-series changes in bandwidth utilization, a traffic surge index is calculated to determine whether there is a sudden traffic surge. When the traffic surge index exceeds the preset threshold, the affected subgraph is identified.

[0138] Specifically, the formula for calculating the flow change index is:

[0139]

[0140] Where S(t) represents the flow rate mutation index;

[0141] It should be noted that reconfiguration is only triggered when the traffic mutation index exceeds the preset threshold.

[0142] Furthermore, the preset threshold for the flow mutation index is obtained by calculating the mean and twice the standard deviation of the flow mutation index within the time window:

[0143] ΔS th (t)=μ S +2σ S

[0144] Where, ΔS th (t) represents the preset threshold for the flow mutation index, μ S The mean of the flow change index, σ S Expressed as the standard deviation of the flow rate mutation index;

[0145] Furthermore, by incrementally updating the graph neural network, the nodes of the affected subgraph are recalculated to generate a new resource demand matrix;

[0146] Specifically, incrementally updating node embeddings in a graph neural network:

[0147]

[0148] Where, h′ i (t) represents the updated node embedding. Represents the affected neighbor set;

[0149] Specifically, based on the confidence interval of resource demand, optimize resource reconfiguration to meet the high-priority node requirements of uRLLC slices while keeping the reconfiguration latency below 1 millisecond, and feed back the resource reconfiguration results to the current network resource status;

[0150] It should be noted that incremental GNN updates overcome the limitation of high reconfiguration latency in dynamic scenarios by traditional static or semi-static resource allocation methods.

[0151] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code. The solutions in the embodiments of this application can be implemented using various computer languages, such as the object-oriented programming language Java and the interpreted scripting language JavaScript.

[0152] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0153] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0154] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0155] Although preferred embodiments of this application have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of this application.

[0156] Obviously, those skilled in the art can make various modifications and variations to this application without departing from the spirit and scope of this application. Therefore, if such modifications and variations fall within the scope of the claims of this application and their equivalents, this application also intends to include such modifications and variations.

Claims

1. A method for intelligent allocation of low-latency slice resources for 5G networks, characterized in that, include: The system collects the connection status and performance indicators of network nodes in real time, and dynamically adjusts the collection frequency according to network load fluctuations to generate structured topology data. Based on the topology data, calculate the degree distribution entropy value of the network nodes, generate a virtual resource topology map, and predict the resource requirements of the network nodes. The generation of the virtual resource topology map includes: Based on the structured topology data, a network topology graph is constructed, where nodes represent network nodes, edges represent connection relationships, and node features are performance indicators. By applying a graph neural network and processing the network topology graph through a message passing mechanism, a virtual resource topology graph is generated, and the virtual resource topology graph is represented as a resource demand matrix, which includes the bandwidth and computing resource requirements of each node. Based on the virtual resource topology map and the rate of change of the entropy value, a dynamic confidence interval is constructed, and a resource pre-allocation strategy is designed. The system detects sudden traffic spikes in the network and rapidly adjusts resource allocation and reduces reconfiguration latency by incrementally updating the graph neural network.

2. The intelligent allocation method for low-latency slice resources in 5G networks as described in claim 1, characterized in that, The system collects real-time network node connection status and performance metrics, dynamically adjusts the collection frequency based on network load fluctuations, and generates structured topology data, including: By collecting connection status data of network nodes through distributed network probes, an adjacency matrix of connection strength between nodes is formed; Collect performance metrics for each node, including bandwidth utilization, computational load, and latency, to form a node feature matrix; The network load fluctuation is calculated based on the fluctuation rate of the bandwidth utilization, and the sampling frequency of the distributed network probe is dynamically adjusted. The adjacency matrix and node feature matrix are denoised and normalized to generate structured topology data.

3. The intelligent allocation method for low-latency slice resources in 5G networks as described in claim 2, characterized in that, Based on the topology data, calculate the degree distribution entropy value of the network nodes, including: Based on the adjacency matrix, calculate the degree of each node, which is the sum of the connection strengths between each node and its neighboring nodes; Based on the degree of the nodes, a degree distribution function for smoothing the nodes is constructed, and the entropy value of the structured topology data is calculated. By fusing the entropy values ​​of the current structured topology data with their historical entropy values ​​using a time-weighted method, a smooth and effective entropy value is generated. Calculate the rate of change of the effective entropy value and store the effective entropy value and the rate of change. When the effective entropy value exceeds a preset threshold, trigger the generation operation of the virtual resource topology map.

4. The intelligent allocation method for low-latency slice resources for 5G networks as described in claim 3, characterized in that, When the effective entropy value exceeds a preset threshold, the virtual resource topology graph generation operation is triggered, including: The preset threshold for the effective entropy value is obtained by calculating the mean and standard deviation of the effective entropy value and the network topology change coefficient.

5. The intelligent allocation method for low-latency slice resources for 5G networks as described in claim 1, characterized in that, Construct dynamic confidence intervals and design resource pre-allocation strategies, including: Based on the resource demand matrix, calculate the predicted resource demand and its fluctuation level for each node; Based on the predicted resource demand and volatility, and combined with the rate of change of effective entropy, the confidence level is dynamically adjusted to construct the resource demand confidence interval for each node. Design a resource pre-allocation strategy that prioritizes meeting the boundary requirements of the resource demand confidence interval, while constraining the total resource demand to not exceed the available network resources. The allocation results obtained after executing the resource pre-allocation strategy will be stored, and the current network resource status will be updated.

6. The intelligent allocation method for low-latency slice resources for 5G networks as described in claim 5, characterized in that, Detecting bursts of network traffic and rapidly adjusting resource allocation and reducing reconfiguration latency by incrementally updating the graph neural network, including: Based on the time-series changes in bandwidth utilization, a traffic mutation index is calculated to determine whether there is a sudden traffic surge. When the traffic mutation index exceeds a preset threshold, the affected sub-graph is identified. By incrementally updating the graph neural network, the nodes of the affected subgraph are recalculated to generate a new resource demand matrix; Based on the aforementioned resource demand confidence interval, resource reconfiguration is optimized to meet the high-priority node requirements of uRLLC slices while maintaining a reconfiguration latency of less than 1 millisecond. The results of resource reconfiguration are fed back to the current network resource status.

7. The intelligent allocation method for low-latency slice resources in 5G networks as described in claim 6, characterized in that, When the traffic flow mutation index exceeds a preset threshold, including: The preset threshold for the flow mutation index is obtained by calculating the mean and twice the standard deviation of the flow mutation index within a time window.

Images