CN113541986B - Fault prediction method and device for 5G slice and computing equipment - Google Patents
Fault prediction method and device for 5G slice and computing equipment Download PDFInfo
- Publication number
- CN113541986B CN113541986B CN202010294622.5A CN202010294622A CN113541986B CN 113541986 B CN113541986 B CN 113541986B CN 202010294622 A CN202010294622 A CN 202010294622A CN 113541986 B CN113541986 B CN 113541986B
- Authority
- CN
- China
- Prior art keywords
- slice
- graph
- matrix
- physical connection
- relation
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L41/00—Arrangements for maintenance, administration or management of data switching networks, e.g. of packet switching networks
- H04L41/14—Network analysis or design
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/044—Recurrent networks, e.g. Hopfield networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L41/00—Arrangements for maintenance, administration or management of data switching networks, e.g. of packet switching networks
- H04L41/06—Management of faults, events, alarms or notifications
- H04L41/0631—Management of faults, events, alarms or notifications using root cause analysis; using analysis of correlation between notifications, alarms or events based on decision criteria, e.g. hierarchy, tree or time analysis
Abstract
The invention discloses a method, a device and a computing device for predicting faults of 5G slices, wherein the method comprises the following steps: collecting network slice data to be predicted, and dividing a slice physical connection graph and a slice resource logical relation graph; converting the sliced physical connection diagram into a first adjacency matrix and a first characteristic matrix, and converting the sliced resource logic relationship diagram into a second adjacency matrix and a second characteristic matrix; and inputting the first adjacent matrix, the first characteristic matrix, the second adjacent matrix and the second characteristic matrix into a space-time multi-graph convolution network model, and predicting to obtain fault information of the historical network slice data. By the method, the potential faults of the 5G slices can be predicted by utilizing the space-time multi-graph convolution network model, and the space-time relation of the slice topology network is considered, so that the faults can be automatically and accurately found.
Description
Technical Field
The invention relates to the technical field of communication, in particular to a method and a device for predicting faults of a 5G slice and computing equipment.
Background
A Network Slice (Network Slice) is an end-to-end logical function and a physical or virtual resource set required by the end-to-end logical function, including an access Network, a transmission Network, a core Network, and the like, and the Network Slice can be regarded as a virtualized "private Network" in a 5G Network; the network slice is constructed based on the uniform infrastructure of the NFV, and low-cost and efficient operation is achieved. Network slice techniques may enable logical isolation of a communication network, allowing network elements and functionality to be configured and reused in each network slice to meet specific industry application needs. At the same time, the network slice may also have various failures.
However, in the prior art, the slice latent fault prediction is mainly realized by setting a threshold value for each KPI in a slice network through expert experience, the prediction accuracy is low, the spatio-temporal relationship of a slice topology network is not considered, and the slice network has many nodes, various dependency relationships exist among the nodes, the spatio-temporal correlation of slices is complex, the difficulty of automatic prediction is high, and no good slice fault prediction means exists at present.
Disclosure of Invention
In view of the above problems, embodiments of the present invention are proposed to provide a failure prediction method, apparatus and computing device for 5G slice that overcome or at least partially solve the above problems.
According to an aspect of the embodiments of the present invention, there is provided a failure prediction method for a 5G slice, including:
collecting network slice data to be predicted, and dividing a slice physical connection graph and a slice resource logical relation graph;
converting the sliced physical connection diagram into a first adjacency matrix and a first characteristic matrix, and converting the sliced resource logic relationship diagram into a second adjacency matrix and a second characteristic matrix;
and inputting the first adjacent matrix, the first characteristic matrix, the second adjacent matrix and the second characteristic matrix into a space-time multi-graph convolution network model, and predicting to obtain the fault information of the historical network slice data.
According to another aspect of the embodiments of the present invention, there is provided a failure prediction apparatus for a 5G slice, including:
the dividing module is suitable for acquiring historical network slice data and dividing a slice physical connection graph and a slice resource logical relation graph;
the conversion module is suitable for converting the slice physical connection diagram into a first adjacency matrix and a first characteristic matrix and converting the slice resource logic relation diagram into a second adjacency matrix and a second characteristic matrix;
and the prediction module is suitable for inputting the first adjacent matrix, the first characteristic matrix, the second adjacent matrix and the second characteristic matrix into a space-time multi-graph convolution network model to predict and obtain the fault information of the historical network slice data.
According to still another aspect of an embodiment of the present invention, there is provided a computing device including: the system comprises a processor, a memory, a communication interface and a communication bus, wherein the processor, the memory and the communication interface are communicated with each other through the communication bus;
the memory is used for storing at least one executable instruction, and the executable instruction enables the processor to execute the operation corresponding to the fault prediction method of the 5G slice.
According to a further aspect of the embodiments of the present invention, there is provided a computer storage medium having at least one executable instruction stored therein, where the executable instruction causes a processor to perform operations corresponding to the failure prediction method of the 5G slice.
According to the method, the device and the computing equipment for predicting the fault of the 5G slice, two types of topological relations are obtained by acquiring network slice data to be predicted and dividing a slice physical connection graph and a slice resource logical relation graph; converting both the two types of topological relations into corresponding adjacency matrixes and characteristic matrixes to obtain input data of the space-time multi-graph convolution network model; and inputting the first adjacency matrix and the first characteristic matrix as well as the second adjacency matrix and the second characteristic matrix into a time-space multi-graph convolution network model as two groups of input data respectively, extracting time-space characteristics in two types of topological relations respectively, and then automatically mining the relation between the time-space characteristics and potential faults through multi-graph aggregation so as to predict and obtain fault information. Therefore, according to the scheme of the invention, the spatio-temporal characteristics corresponding to the adjacency matrix and the characteristic matrix of the two types of topological structures can be extracted by utilizing the spatio-temporal multi-graph convolution network model, then the relation between the slice spatio-temporal characteristics and the potential faults is excavated, and the fault information is obtained by prediction, so that the fault prediction is automatically and accurately carried out, and the faults existing in the slice network are found in advance.
The foregoing description is only an overview of the technical solutions of the embodiments of the present invention, and the embodiments of the present invention can be implemented according to the content of the description in order to make the technical means of the embodiments of the present invention more clearly understood, and the detailed description of the embodiments of the present invention is provided below in order to make the foregoing and other objects, features, and advantages of the embodiments of the present invention more clearly understandable.
Drawings
Various other advantages and benefits will become apparent to those of ordinary skill in the art upon reading the following detailed description of the preferred embodiments. The drawings are only for purposes of illustrating the preferred embodiments and are not to be construed as limiting the embodiments of the invention. Also, like reference numerals are used to refer to like parts throughout the drawings. In the drawings:
fig. 1 is a flowchart illustrating a failure prediction method for a 5G slice according to an embodiment of the present invention;
FIG. 2 is a flow chart illustrating a method for predicting failure of a 5G slice according to another embodiment of the present invention;
FIG. 3 is a schematic diagram illustrating the components of a spatio-temporal multi-map convolutional network model according to an embodiment of the present invention;
fig. 4 is a schematic structural diagram of a failure prediction apparatus for 5G slices according to an embodiment of the present invention;
fig. 5 is a schematic structural diagram of a computing device provided by an embodiment of the present invention.
Detailed Description
Exemplary embodiments of the present invention will be described in more detail below with reference to the accompanying drawings. While exemplary embodiments of the invention are shown in the drawings, it should be understood that the invention can be embodied in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the invention to those skilled in the art.
Before carrying out embodiments of the present invention, reference is first made to the following terms referred to herein:
1. slice management architecture: mainly comprises CSMF, NSMF and NSSMF.
The CSMF (Communication Service Management Function) completes order and processing of a user Service Communication Service requirement, is responsible for converting a Communication Service requirement of an operator/third-party client into a requirement for a network slice, sends the requirement for the network slice (such as a request for creating, terminating, modifying a network slice instance and the like) to the NSMF through an interface between the CSMF and the NSMF, and acquires Management data (such as performance, fault data and the like) of the network slice from the NSMF.
The NSMF (Network Slice Management Function) is responsible for receiving a Network Slice requirement sent by the CSMF (Communication Service Management Function), managing life cycle, performance, fault and the like of the Network Slice instance, arranging composition of the Network Slice instance, decomposing the requirement of the Network Slice instance into the requirement of each Network Slice subnet instance or Network Function, and sending a Network Slice subnet instance Management request to each NSSMF.
The NSSMF (Network Slice Subnet Management Function) receives a Network Slice Subnet deployment requirement issued by the NSMF, manages a Network Slice Subnet instance, arranges the composition of the Network Slice Subnet instance, maps an SLA requirement of the Network Slice Subnet into a QoS requirement of a Network service, and issues a deployment request of the Network service to an NFVO system of an ETSI NFV domain.
Lstm (long short-term memory) is a special type of recurrent neural network, and by controlling the time for which values in a buffer are stored, long-term information can be remembered, which is suitable for prediction of time series. Each neuron has four inputs and one output, and each neuron stores a memorized value. LSTM neurons were as follows:
Y t ＝σ(W'h t ) (7)
each LSTM neuron contains three gates: forget gate, input gate, output gate. Equation (1) represents a forgetting gate, new information is added in equations (2) and (3), equation (4) fuses the new and old information, and equations (5) and (6) output information about the next timestamp that the LSTM unit has learned so far. The long-short term memory neural network has a good effect on the learning of long-time sequences, each connecting line in the LSTM unit contains corresponding weight, xt represents an input vector, ht represents a hidden state, ct represents a neuron state at t, yt represents the output of a neuron, W is a trainable weight matrix, and b is a bias vector.
Fig. 1 shows a flowchart of a failure prediction method for a 5G slice according to an embodiment of the present invention. The method may be performed by any computing device having data processing capabilities. As shown in fig. 1, the method comprises the steps of:
step S110: collecting network slice data to be predicted, and dividing a slice physical connection graph and a slice resource logic relation graph.
Wherein the network slice data to be predicted can be collected from a Network Slice Management Function (NSMF).
Step S120: the sliced physical connection graph is converted into a first adjacency matrix and a first feature matrix, and the sliced resource logical relationship graph is converted into a second adjacency matrix and a second feature matrix.
Specifically, the sliced physical connection graph and the sliced resource logical relationship graph are converted into an adjacency matrix (a first adjacency matrix and a second adjacency matrix) and a feature matrix (a first feature matrix and a second feature matrix), wherein the first adjacency matrix and the second adjacency matrix are matrices formed by connection relationships of nodes in two types of topological relationships (physical connection and logical relationship, the same applies below), and the first feature matrix and the second feature matrix are matrices formed by slice features of nodes in the two types of topological relationships.
Step S130: and inputting the first adjacent matrix, the first characteristic matrix, the second adjacent matrix and the second characteristic matrix into a space-time multi-graph convolution network model, and predicting to obtain the fault information of the historical network slice data.
In the invention, a space-time multi-graph convolution network (ST-MGCN) model is utilized to learn the space-time relation of various topologies in 5G network slices, and meanwhile, the dependency relation of the space and time of the network slices is captured, the model combines a Graph Convolution Network (GCN) and a long-short term memory neural network (LSTM), wherein the GCN is used for learning a complex topological structure to capture the space dependency relation, and the LSTM is used for learning the dynamic change of each node KPI (Key Performance Indicator) of the slices to capture the time dependency relation. Because the slice network is not a two-dimensional grid but a graph, the prediction is carried out through a space-time multi-graph convolution network model, the complex topological structure of the slice network can be reflected, the space-time dependency relationship can be accurately captured, and the prediction can be automatically and accurately realized compared with a CNN (convolutional neural network) model and the like.
Specifically, a first adjacent matrix and a first characteristic matrix, and a second adjacent matrix and a second characteristic matrix are respectively used as two groups of input data to be input into a space-time multi-graph convolution network model, a space-time multi-graph network is used for carrying out graph convolution on the physical connection relation and the logical relation of the slice resources among slice network areas, respective space relation characteristics are extracted, then time characteristics are extracted from continuous T moments KPI of each node of the slices through a long-short term memory layer, multi-graph aggregation is carried out on potential variables of the slice relation graph obtained after the space-time characteristics are extracted, the relation between the slice space-time characteristics and potential faults is automatically mined, and finally predicted fault information is output through a full connection layer. The failure information includes any information that can reflect whether the slice network has a failure and/or the specific time, location, type, level, etc. of the failure.
According to the fault prediction method of the 5G slice provided by the embodiment, two types of topological relations are obtained by acquiring network slice data to be predicted and dividing a slice physical connection graph and a slice resource logical relation graph; converting both the two types of topological relations into corresponding adjacency matrixes and characteristic matrixes to obtain input data of the space-time multi-graph convolution network model; and inputting the first adjacency matrix and the first characteristic matrix as well as the second adjacency matrix and the second characteristic matrix into a time-space multi-graph convolution network model as two groups of input data respectively, extracting time-space characteristics in two types of topological relations respectively, and then automatically mining the relation between the time-space characteristics and potential faults through multi-graph aggregation so as to predict and obtain fault information. Therefore, according to the scheme of the embodiment, the space-time multi-graph convolution network model can be used for extracting the space-time characteristics corresponding to the adjacency matrix and the characteristic matrix of the two types of topological structures, then the relation between the slice space-time characteristics and the potential faults is excavated, the fault information is obtained through prediction, the fault prediction is automatically and accurately carried out, and the faults existing in the slice network are found in advance.
Fig. 2 shows a flowchart of a failure prediction method for a 5G slice according to another embodiment of the present invention. As shown in fig. 2, the method comprises the steps of:
step S210: and training to obtain a space-time multi-graph convolution network model.
Specifically, a historical network slice data sample is collected and divided into a slice physical connection diagram sample and a slice resource logical relation diagram sample; converting the sliced physical connection map sample into a first adjacency matrix sample and a first feature matrix sample, and converting the sliced resource logical relationship map sample into a second adjacency matrix sample and a second feature matrix sample; marking a fault information sample; taking the first adjacent matrix sample, the first characteristic matrix sample, the second adjacent matrix sample and the second characteristic matrix sample as training input data, and taking the fault information sample as training output data; and training by using the training input data and the training output data to obtain a space-time multi-graph convolution network model.
Further, the process of training to obtain the spatio-temporal multi-map convolutional network model is mainly divided into the following two steps:
step one, data preprocessing.
Firstly, processing to obtain an adjacency matrix sample, a characteristic matrix sample and a label matrix sample:
historical network slice data is collected from a Network Slice Management Function (NSMF), and is divided into a slice physical connection graph (physical connection relationship topology of each node) sample and a slice resource logical relationship graph (resource association relationship topology obtained by a resource CMDB (Configuration Management Database)) sample as a total data set, and the two relationship graph samples are converted into input of a model, namely, an adjacency matrix sample (a first adjacency matrix sample and a second adjacency matrix sample) and a feature matrix sample (a first feature matrix sample and a second feature matrix sample). The adjacency matrix sample is different connection relations of all nodes in the two types of topological relation graphs, the characteristic matrix sample is divided into a static slicing characteristic and a dynamic slicing characteristic, and the dynamic slicing characteristic is a KPI time sequence of T historical time steps of all nodes in a slicing network. Meanwhile, the predicted result is marked manually.
The slicing physical connection graph sample is represented as G1 (V, E1), the slicing resource logical relationship graph sample is represented as G2 (V, E2), V is a set of slicing nodes V = { V1, V2, V3, …, VN }, and E is a set of edges, and a connection value between nodes is determined according to correlations of different dimensions, in the physical connection and logical relationship, if there is a connection between a slicing node Vi and a slicing node Vj, a connection value eij is a first connection value, and if there is no connection between the slicing node Vi and the slicing node Vj, the connection value eij is a second connection value, for example, the first connection value is 1, and the second connection value is 0. Converting the two types of slice topological graphs into inputs of a model, namely an adjacency matrix sample and a characteristic matrix sample:
the adjacency matrix sample a is different connection relations of each node in the two types of topological relation graphs, and is respectively represented as A1 (first adjacency matrix sample) and A2 (second adjacency matrix sample). eij represents the connection relation between the networking node Vi and the networking node Vj, the nodes are connected to be 1, and otherwise, the nodes are 0. The shape is N x N (N is the number of nodes).
The feature matrix sample X is a feature representation of each node in a network slice in two types of topological relational graphs, which is respectively represented as X1 (a first feature matrix sample) and X2 (a second feature matrix sample), and is divided into a static slice feature and a dynamic slice feature, where the static slice feature includes, but is not limited to, one or more of the following: the slicing subnet to which the slicing node belongs, the type of the slicing node and the capacity of the slicing node, and the dynamic slicing characteristics include, but are not limited to, one or more of the following: network Concurrency (CAPS), request delay, request success rate, error code number and resource utilization rate. Furthermore, the ith node has T consecutive time attributes in the history at time T, which are expressed as { Xi } t-T+1 、…、Xi t-2 、Xi t-1 、Xi t And T represents the historical time sequence length of the node attribute.
The label matrix samples Y are the failure information labeled for the slice to be predicted.
Secondly, the data set is standardized:
this is done separately for each dimension, subtracting the mean from the data by attribute (by column) and dividing by its variance, i.e.: the processing formula is (X-mean)/std, wherein X is the value of an element in the matrix, mean is the mean value of the column where the element is located, and std is the variance of the column where the element is located. After the standardized post-processing, the convergence rate of the model and the precision of the model are improved.
Finally, dividing training data and test data:
the total data set is divided into training data and test data, and 80% of the whole data set is taken as training data, and the rest 20% is taken as test data. And training by using a training set, so that the closer the reconstructed data is to the original data, the better the reconstructed data is, and evaluating the verification model by using a test set.
And step two, building and training a model.
And constructing a space-time multi-graph convolutional neural network (ST-MGCN) consisting of a graph convolutional layer, a long-short term memory layer, a multi-graph convergence layer and a full connection layer. And performing graph convolution on two correlation relations (a slice physical connection graph and a slice resource logic relation graph) between slice network areas by using a space-time multi-graph network respectively, extracting respective spatial relation characteristics, then extracting time characteristics from continuous T moments KPI (Key Performance indicator) of each node of a slice through a long-short term memory layer, performing multi-graph aggregation on potential variables of the slice relation graph obtained after the space-time characteristics are extracted, and finally outputting predicted fault information through a full connection layer.
FIG. 3 is a schematic diagram showing the components of a spatio-temporal multi-map convolutional network model according to an embodiment of the present invention. As shown in fig. 3, the spatio-temporal multi-map convolutional network model is constructed as follows:
1) Graph Conv (GCN): inputting the adjacent matrix and characteristic matrix corresponding to slice physical connection diagram G1 (V, E1) and slice resource logical relation diagram G2 (V, E2) into respective graph convolution layer, extracting spatial characteristics of the above two relation diagrams, encoding the slice physical connection diagram by L-layer graph convolution layer to represent as first slice relation diagram H 1 L The slice resource logic relation graph is represented as a second slice relation graph H after being encoded by the L-layer graph convolutional layer 2 L . The number of convolution kernels is 32 and the activation function is set to "relu". Each graph convolution layer can be written as such a non-linear function:
H l+1 ＝RELU(D' -1/2 A'D' -1/2 H l W l )
where H (0) = X is input data, H (L) = Z is output data, and L is the number of layers of the map stack layer. Wherein W l Is a parameter matrix of the l-th map convolution layer, A being an adjacency matrix, D' -1/2 A'D' -1/2 Is a symmetric normalization to the adjacency matrix a, a ' = a + I, D ' is the node-degree diagonal matrix of a '.
2) Long short term memory Layer (LSTM): a slice relation graph H after coding the graph convolutional layer 1 L And H 2 L Time characteristic extraction is carried out, the number of the neurons is set to be 64, the activation function is set to be relu, and a first chip relation graph latent variable H 'corresponding to the chip physical connection graph and the chip resource logic relation graph respectively is output' 1 L And a second slice relationship graph latent variable H' 2 L ：
H' 1 L ＝LSTM(H 1 L )
H' 2 L ＝LSTM(H 2 L )
3) Multi-graph convergence layer: the two relation graphs are subjected to space-time characteristic extraction of the graph volume layer and the long and short term memory layer to obtain a slice relation graph latent variable H' 1 L And H' 2 L Performing multi-graph aggregation, wherein aggregation is performed by taking nodes as units, and the result obtained by aggregation is H' 1，2 L ：
H' 1，2 L ＝RELU(Agg(H' 1 L ，H' 2 L ))
Agg () is an Aggregation function (Aggregation function) and the method of Aggregation may be summation, maximum, average, etc.
4) Fully connected layer (Dense): the number of neurons is set to 4, that is, the failure region of the slice (for example, the ith node in the slice), the failure time (for example, t seconds in the future), the failure level (which can be classified into high, medium and low levels) and the failure type (which can be classified into M types) predicted by the corresponding output, and the activation function is set to "relu".
The model will train 1000 rounds (epochs = 1000), set the batch size to 32 (batch _ size = 32), select the Mean Squared Error MSE (Mean Squared Error) as the loss function, i.e. the objective function (loss = 'Mean Squared Error'):
the gradient descent optimization algorithm selects an adam optimizer for improving the learning speed of the traditional gradient descent (optimizer = 'adam'). The neural network can find the optimal weight value which enables the target function to be minimum through gradient descent, and the neural network can learn the weight value automatically through training. And deriving the weight of the model after the model converges.
Step S220: collecting network slice data to be predicted, and dividing a slice physical connection graph and a slice resource logic relation graph.
After the space-time multi-graph convolution network model is obtained through training, if the slice fault prediction requirement exists, network slice data to be predicted are collected.
Step S230: the sliced physical connection graph is converted into a first adjacency matrix and a first feature matrix, and the sliced resource logical relationship graph is converted into a second adjacency matrix and a second feature matrix.
Wherein the first adjacency matrix represents the physical connection relation between nodes in the slice physical connection graph; the second adjacency matrix represents the logical connection relation among all nodes in the slice resource logical relation graph; and the first feature matrix represents static slicing features and/or dynamic slicing features of each node in the sliced physical connection graph; the second feature matrix represents static and/or dynamic slicing features of nodes in the sliced resource logical relationship graph.
Wherein the static slice features include one or more of: the slicing sub-network to which the slicing node belongs, the type of the slicing node and the capacity of the slicing node; and/or, the dynamic slice characteristics include one or more of: network concurrency, request time delay, request success rate, error code number and resource utilization rate.
After the conversion, input data of the space-time multi-graph convolution network model are obtained, and prediction can be carried out subsequently.
Step S240: and inputting the first adjacent matrix, the first characteristic matrix, the second adjacent matrix and the second characteristic matrix into a space-time multi-graph convolution network model, and predicting to obtain the fault information of the historical network slice data.
Wherein the fault information includes one or more of: slice fault region information, slice fault time information, slice fault level information, and slice fault type information.
Specifically, multi-graph convolution refers to graph convolution of a physical connection relation and a resource logic relation respectively, feature fusion is performed, the purpose that a graph of the physical connection relation and the resource logic relation at one moment is combined into one graph is achieved, and time dimension prediction is to fuse information of T historical time steps into one graph; the prediction of the spatial dimension is to combine the graphs of different correlations at a time into one graph.
Further, the method is consistent with a space-time multi-graph convolution network model established during training, the space-time multi-graph convolution network model comprises a graph convolution layer, a long-short term memory layer, a multi-graph convergence layer and a full connection layer, and correspondingly, the process of predicting by using the space-time multi-graph convolution network model is as follows:
taking the first adjacency matrix and the first characteristic matrix as well as the second adjacency matrix and the second characteristic matrix as two groups of inputs to be respectively input to the graph convolution layer, extracting the spatial characteristics of the slice physical connection graph and the slice resource logical relationship graph, and outputting to obtain a first slice relationship graph and a second slice relationship graph which respectively correspond to the slice physical connection graph and the slice resource logical relationship graph; and taking the first slice relation graph and the second slice relation graph as two groups of inputs to be respectively input to the long-term and short-term memory layer, extracting the time characteristics of the slice physical connection graph and the slice resource logical relation graph, and outputting to obtain the first slice relation graph latent variable and the second slice relation graph latent variable which respectively correspond to the slice physical connection graph and the slice resource logical relation graph. The process is that the first adjacent matrix and the first characteristic matrix are input into a graph volume layer for extracting the space characteristic in the slice physical connection relation graph, after the space characteristic is extracted, the first slice relation graph coded by the graph volume layer is obtained, then the first slice relation graph is input into a long-short term memory layer for extracting the time characteristic in the slice physical connection relation graph, after the time characteristic is extracted, the first slice relation graph latent variable coded by the long-short term memory layer is obtained, similarly, the second adjacent matrix and the second characteristic matrix are sequentially input into the graph volume layer for extracting the space characteristic of the slice resource logic relation graph and the long-short term memory layer for extracting the time characteristic, the second slice relation graph latent variable can be obtained, and the space-time characteristic extraction can be respectively carried out on the slice physical connection graph and the slice resource logic relation graph by the mode, and the space-time characteristic extraction can be easily carried out on the slice physical connection graph and the slice resource logic relation graph.
And then inputting the latent variables of the first and second slice relational graphs into a multi-graph convergence layer for multi-graph convergence processing, and finally outputting the predicted fault information of the historical network slice data through a full connection layer, wherein the multi-graph convergence processing is to fuse the latent variables of the two slice relational graphs by nodes and automatically mine the relation between the space-time characteristics of the slices and the latent faults.
Step S250: and feeding back the fault information to the network slice management function module.
After the fault information is obtained, the fault information is fed back to the NSMF, so that the slice management framework can be facilitated to perform corresponding processing according to a prediction result, for example, troubleshooting or fault elimination and the like.
According to the fault prediction method of the 5G slice, historical network slice data are collected from a network slice management function NSMF, the historical network slice data are divided into a slice physical connection graph and a slice resource logic relation graph as a total data set, the two relation graphs are respectively converted into input of a model, namely an adjacency matrix and a characteristic matrix, a space-time multi-graph convolution neural network (ST-MGCN) consisting of a graph convolution layer, a long-short term memory layer, a multi-graph convergence layer and a full connection layer is built, graph convolution is respectively carried out on the slice physical connection relation and the slice resource logic relation among the slice networks by using the space-time multi-graph network, respective space relation characteristics are extracted, time characteristics are extracted from continuous T moments KPI of nodes of the slice through the long-short term memory layer, multi-graph convergence is carried out on potential variables of the slice relation graph obtained after the space-time characteristics are extracted, the area relation between the slice space-time characteristics and potential faults is automatically excavated, and finally the predicted fault area, fault time, fault level and fault type of the slice are output through the full connection layer, and the potential fault prediction result is returned to the NSMF, so that the potential fault capability of network slice discovery is improved.
Fig. 4 shows a schematic structural diagram of a failure prediction apparatus for a 5G slice according to an embodiment of the present invention.
As shown in fig. 4, the apparatus includes:
a dividing module 410, adapted to collect historical network slice data, and divide a slice physical connection diagram and a slice resource logical relationship diagram;
a conversion module 420 adapted to convert the sliced physical connection map into a first adjacency matrix and a first feature matrix, and convert the sliced resource logical relationship map into a second adjacency matrix and a second feature matrix;
and the prediction module 430 is adapted to input the first adjacency matrix, the first feature matrix, the second adjacency matrix and the second feature matrix into a spatio-temporal multi-graph convolution network model, and predict the fault information of the historical network slice data.
In an alternative mode, the first adjacency matrix represents a physical connection relation between nodes in the slice physical connection graph; the second adjacency matrix represents the logical connection relation among all nodes in the slice resource logical relation graph; and the number of the first and second groups,
the first feature matrix represents static slicing features and/or dynamic slicing features of each node in a sliced physical connection diagram; the second feature matrix represents static slicing features and/or dynamic slicing features of each node in the sliced resource logical relationship diagram.
In an alternative approach, the static slice features include one or more of: the type of the slicing node and the capacity of the slicing node; and/or the presence of a gas in the gas,
the dynamic slicing features include one or more of: network concurrency, request delay, request success rate, error code number and resource utilization rate.
In an optional mode, the space-time multi-graph convolutional network model comprises graph convolutional layers, a long short-term memory layer, a multi-graph convergence layer and a full connection layer;
the prediction module is further adapted to:
taking the first adjacency matrix and the first characteristic matrix as well as the second adjacency matrix and the second characteristic matrix as two groups of inputs to be respectively input to the graph convolution layer, extracting the spatial characteristics of the slice physical connection graph and the slice resource logical relationship graph, and outputting to obtain a first slice relationship graph and a second slice relationship graph which respectively correspond to the slice physical connection graph and the slice resource logical relationship graph;
the first slice relation graph and the second slice relation graph are used as two groups of input to be respectively input to the long-short term memory layer, the time characteristics of the slice physical connection graph and the slice resource logical relation graph are extracted, and the first slice relation graph latent variable and the second slice relation graph latent variable which respectively correspond to the slice physical connection graph and the slice resource logical relation graph are output;
and inputting the latent variables of the first slice relation diagram and the latent variables of the second slice relation diagram into a multi-graph convergence layer for multi-graph convergence processing, and finally outputting the predicted fault information of the historical network slice data through a full connection layer.
In an optional manner, the fault information includes one or more of: slice fault region information, slice fault time information, slice fault level information, and slice fault type information.
In an optional manner, the apparatus further comprises: and the feedback module is suitable for feeding the fault information back to the network slice management function module.
In an optional manner, the apparatus further comprises: the training module is suitable for acquiring historical network slice data samples and dividing the historical network slice data samples into slice physical connection graph samples and slice resource logical relation graph samples;
converting the sliced physical connection diagram sample into a first adjacency matrix sample and a first feature matrix sample, and converting the sliced resource logical relationship diagram sample into a second adjacency matrix sample and a second feature matrix sample; marking a fault information sample;
taking the first adjacency matrix sample, the first characteristic matrix sample, the second adjacency matrix sample and the second characteristic matrix sample as training input data, and taking the fault information sample as training output data;
and training by using the training input data and the training output data to obtain a space-time multi-graph convolution network model.
An embodiment of the present invention provides a non-volatile computer storage medium, where the computer storage medium stores at least one executable instruction, and the computer executable instruction may execute the method for predicting a failure of a 5G slice in any of the method embodiments described above.
Fig. 5 is a schematic structural diagram of a computing device according to an embodiment of the present invention, and the specific embodiment of the present invention does not limit the specific implementation of the computing device.
As shown in fig. 5, the computing device may include: a processor (processor) 502, a Communications Interface 504, a memory 506, and a communication bus 508.
Wherein: the processor 502, communication interface 504, and memory 506 communicate with each other via a communication bus 508. A communication interface 504 for communicating with network elements of other devices, such as clients or other servers. The processor 502, configured to execute the program 510, may specifically perform relevant steps in the embodiment of the failure prediction method for a 5G slice of a computing device described above.
In particular, program 510 may include program code that includes computer operating instructions.
The processor 502 may be a central processing unit CPU, or an Application Specific Integrated Circuit ASIC (Application Specific Integrated Circuit), or one or more Integrated circuits configured to implement an embodiment of the present invention. The computing device includes one or more processors, which may be the same type of processor, such as one or more CPUs; or may be different types of processors such as one or more CPUs and one or more ASICs.
And a memory 506 for storing a program 510. The memory 506 may comprise high-speed RAM memory, and may also include non-volatile memory (non-volatile memory), such as at least one disk memory.
The program 510 may specifically be used to cause the processor 502 to perform the following operations:
collecting network slice data to be predicted, and dividing a slice physical connection graph and a slice resource logical relation graph;
converting the sliced physical connection diagram into a first adjacency matrix and a first characteristic matrix, and converting the sliced resource logic relationship diagram into a second adjacency matrix and a second characteristic matrix;
and inputting the first adjacent matrix, the first characteristic matrix, the second adjacent matrix and the second characteristic matrix into a space-time multi-graph convolution network model, and predicting to obtain fault information of the historical network slice data. In an alternative mode, the first adjacency matrix represents a physical connection relation between nodes in the slice physical connection graph; the second adjacency matrix represents the logical connection relation among all nodes in the slice resource logical relation graph; and the number of the first and second groups,
the first feature matrix represents static slice features and/or dynamic slice features of each node in a slice physical connection diagram; the second feature matrix represents static and/or dynamic slicing features of each node in the sliced resource logical relationship graph.
In an alternative approach, the static slice features include one or more of: the type of the slicing node and the capacity of the slicing node; and/or the presence of a gas in the gas,
the dynamic slicing features include one or more of: network concurrency, request delay, request success rate, error code number and resource utilization rate.
In an optional mode, the space-time multi-graph convolutional network model comprises graph convolutional layers, a long short-term memory layer, a multi-graph convergence layer and a full connection layer;
the program 510 further causes the processor 502 to:
taking the first adjacency matrix and the first characteristic matrix as well as the second adjacency matrix and the second characteristic matrix as two groups of inputs to be respectively input to the graph convolution layer, extracting the spatial characteristics of the slice physical connection graph and the slice resource logical relationship graph, and outputting to obtain a first slice relationship graph and a second slice relationship graph which respectively correspond to the slice physical connection graph and the slice resource logical relationship graph;
the first slice relation graph and the second slice relation graph are used as two groups of input to be respectively input to a long-term and short-term memory layer, the time characteristics of the slice physical connection graph and the slice resource logic relation graph are extracted, and the first slice relation graph latent variable and the second slice relation graph latent variable which respectively correspond to the slice physical connection graph and the slice resource logic relation graph are output;
and inputting the latent variables of the first slice relational graph and the latent variables of the second slice relational graph into a multi-graph convergence layer for multi-graph convergence processing, and finally outputting the predicted fault information of the historical network slice data through a full connection layer.
In an optional manner, the fault information includes one or more of: slice fault region information, slice fault time information, slice fault level information, and slice fault type information.
In an alternative, the program 510 further causes the processor 502 to:
and feeding back the fault information to a network slice management function module.
In an alternative, the program 510 further causes the processor 502 to:
collecting historical network slice data samples, and dividing the historical network slice data samples into slice physical connection graph samples and slice resource logic relation graph samples;
converting the sliced physical connection map sample into a first adjacency matrix sample and a first feature matrix sample, and converting the sliced resource logical relationship map sample into a second adjacency matrix sample and a second feature matrix sample; marking a fault information sample;
taking the first adjacency matrix sample, the first feature matrix sample, the second adjacency matrix sample and the second feature matrix sample as training input data, and taking the fault information sample as training output data;
and training by using the training input data and the training output data to obtain a space-time multi-graph convolution network model.
The algorithms or displays presented herein are not inherently related to any particular computer, virtual system, or other apparatus. Various general purpose systems may also be used with the teachings herein. The required structure for constructing such a system will be apparent from the description above. In addition, embodiments of the present invention are not directed to any particular programming language. It is appreciated that a variety of programming languages may be used to implement the teachings of embodiments of the present invention as described herein, and any descriptions of specific languages are provided above to disclose the best modes of embodiments of the invention.
In the description provided herein, numerous specific details are set forth. It is understood, however, that embodiments of the invention may be practiced without these specific details. In some instances, well-known methods, structures and techniques have not been shown in detail in order not to obscure an understanding of this description.
Similarly, it should be appreciated that in the foregoing description of exemplary embodiments of the invention, various features of the embodiments of the invention are sometimes grouped together in a single embodiment, figure, or description thereof for the purpose of streamlining the disclosure and aiding in the understanding of one or more of the various inventive aspects. However, the disclosed method should not be interpreted as reflecting an intention that: that is, the claimed embodiments of the invention require more features than are expressly recited in each claim. Rather, as the following claims reflect, inventive aspects lie in less than all features of a single foregoing disclosed embodiment. Thus, the claims following the detailed description are hereby expressly incorporated into this detailed description, with each claim standing on its own as a separate embodiment of this invention.
Those skilled in the art will appreciate that the modules in the device in an embodiment may be adaptively changed and disposed in one or more devices different from the embodiment. The modules or units or components in the embodiments may be combined into one module or unit or component, and furthermore, may be divided into a plurality of sub-modules or sub-units or sub-components. All of the features disclosed in this specification (including any accompanying claims, abstract and drawings), and all of the processes or elements of any method or apparatus so disclosed, may be combined in any combination, except combinations where at least some of such features and/or processes or elements are mutually exclusive. Each feature disclosed in this specification (including any accompanying claims, abstract and drawings) may be replaced by alternative features serving the same, equivalent or similar purpose, unless expressly stated otherwise.
Furthermore, those skilled in the art will appreciate that while some embodiments herein include some features included in other embodiments, rather than other features, combinations of features of different embodiments are meant to be within the scope of the invention and form different embodiments. For example, in the following claims, any of the claimed embodiments may be used in any combination.
The various component embodiments of the invention may be implemented in hardware, or in software modules running on one or more processors, or in a combination thereof. Those skilled in the art will appreciate that a microprocessor or Digital Signal Processor (DSP) may be used in practice to implement some or all of the functionality of some or all of the components according to embodiments of the present invention. Embodiments of the invention may also be implemented as apparatus or device programs (e.g., computer programs and computer program products) for performing a portion or all of the methods described herein. Such programs implementing embodiments of the present invention may be stored on computer-readable media or may be in the form of one or more signals. Such a signal may be downloaded from an internet website or provided on a carrier signal or in any other form.
It should be noted that the above-mentioned embodiments illustrate rather than limit the invention, and that those skilled in the art will be able to design alternative embodiments without departing from the scope of the appended claims. In the claims, any reference signs placed between parentheses shall not be construed as limiting the claim. The word "comprising" does not exclude the presence of elements or steps not listed in a claim. The word "a" or "an" preceding an element does not exclude the presence of a plurality of such elements. Embodiments of the invention may be implemented by means of hardware comprising several distinct elements, and by means of a suitably programmed computer. In the unit claims enumerating several means, several of these means may be embodied by one and the same item of hardware. The usage of the words first, second and third, etcetera do not indicate any ordering. These words may be interpreted as names. The steps in the above embodiments should not be construed as limiting the order of execution unless specified otherwise.
Claims (9)
1. A failure prediction method for a 5G slice comprises the following steps:
collecting historical network slice data to be predicted, and dividing a slice physical connection graph and a slice resource logical relation graph;
converting the sliced physical connection diagram into a first adjacency matrix and a first characteristic matrix, and converting the sliced resource logic relationship diagram into a second adjacency matrix and a second characteristic matrix;
inputting the first adjacent matrix, the first characteristic matrix, the second adjacent matrix and the second characteristic matrix into a space-time multi-graph convolution network model, and predicting to obtain fault information of the historical network slice data; the space-time multi-graph convolution network model comprises graph convolution layers, a long-term and short-term memory layer, a multi-graph convergence layer and a full connection layer; specifically, a first adjacency matrix and a first characteristic matrix, and a second adjacency matrix and a second characteristic matrix are used as two groups of input and are respectively input into the graph convolution layer, the spatial characteristics of the slice physical connection graph and the slice resource logical relationship graph are extracted, and a first slice relationship graph and a second slice relationship graph respectively corresponding to the slice physical connection graph and the slice resource logical relationship graph are output; the first slice relation graph and the second slice relation graph are used as two groups of input to be respectively input to a long-term and short-term memory layer, the time characteristics of the slice physical connection graph and the slice resource logic relation graph are extracted, and the first slice relation graph latent variable and the second slice relation graph latent variable which respectively correspond to the slice physical connection graph and the slice resource logic relation graph are output; and inputting the latent variables of the first slice relational graph and the latent variables of the second slice relational graph into a multi-graph convergence layer for multi-graph convergence processing, and finally outputting the predicted fault information of the historical network slice data through a full connection layer.
2. The method of claim 1, wherein the first adjacency matrix represents a physical connection relationship between nodes in a sliced physical connection graph; the second adjacency matrix represents the logical connection relation among all nodes in the slice resource logical relation graph; and the number of the first and second groups,
the first feature matrix represents static slice features and/or dynamic slice features of each node in a slice physical connection diagram; the second feature matrix represents static and/or dynamic slicing features of each node in the sliced resource logical relationship graph.
3. The method of claim 2, wherein the static slice features include one or more of: the type of the slicing node and the capacity of the slicing node; and/or the presence of a gas in the gas,
the dynamic slicing features include one or more of: network concurrency, request delay, request success rate, error code number and resource utilization rate.
4. The method of any of claims 1-3, wherein the fault information includes one or more of: slice fault region information, slice fault time information, slice fault level information, and slice fault type information.
5. The method of any of claims 1-3, wherein after the predicting failure information for the historical network slice data, the method further comprises:
and feeding back the fault information to a network slice management function module.
6. The method according to any one of claims 1-3, wherein the spatio-temporal multi-graph convolutional network model is trained by:
collecting historical network slice data samples, and dividing the historical network slice data samples into slice physical connection graph samples and slice resource logic relation graph samples;
converting the sliced physical connection map sample into a first adjacency matrix sample and a first feature matrix sample, and converting the sliced resource logical relationship map sample into a second adjacency matrix sample and a second feature matrix sample; marking a fault information sample;
taking the first adjacency matrix sample, the first characteristic matrix sample, the second adjacency matrix sample and the second characteristic matrix sample as training input data, and taking the fault information sample as training output data;
and training by using the training input data and the training output data to obtain a space-time multi-graph convolution network model.
7. A failure prediction apparatus for a 5G slice, comprising:
the dividing module is suitable for acquiring historical network slice data and dividing a slice physical connection graph and a slice resource logical relation graph;
the conversion module is suitable for converting the slice physical connection diagram into a first adjacency matrix and a first characteristic matrix and converting the slice resource logic relation diagram into a second adjacency matrix and a second characteristic matrix;
the prediction module is suitable for inputting the first adjacent matrix, the first characteristic matrix, the second adjacent matrix and the second characteristic matrix into a space-time multi-graph convolution network model and predicting to obtain fault information of the historical network slice data; the space-time multi-graph convolution network model comprises graph convolution layers, a long-term and short-term memory layer, a multi-graph convergence layer and a full connection layer; specifically, a first adjacent matrix and a first characteristic matrix, and a second adjacent matrix and a second characteristic matrix are used as two groups of inputs to be respectively input to the graph convolution layer, the spatial characteristics of the slice physical connection graph and the slice resource logical relationship graph are extracted, and a first slice relationship graph and a second slice relationship graph respectively corresponding to the slice physical connection graph and the slice resource logical relationship graph are output; the first slice relation graph and the second slice relation graph are used as two groups of input to be respectively input to a long-term and short-term memory layer, the time characteristics of the slice physical connection graph and the slice resource logic relation graph are extracted, and the first slice relation graph latent variable and the second slice relation graph latent variable which respectively correspond to the slice physical connection graph and the slice resource logic relation graph are output; and inputting the latent variables of the first slice relational graph and the latent variables of the second slice relational graph into a multi-graph convergence layer for multi-graph convergence processing, and finally outputting the predicted fault information of the historical network slice data through a full connection layer.
8. A computing device, comprising: the system comprises a processor, a memory, a communication interface and a communication bus, wherein the processor, the memory and the communication interface complete mutual communication through the communication bus;
the memory is used for storing at least one executable instruction, and the executable instruction causes the processor to execute the operation corresponding to the fault prediction method of the 5G slice in any one of claims 1-6.
9. A computer storage medium having stored therein at least one executable instruction for causing a processor to perform operations corresponding to the method of 5G slice failure prediction according to any one of claims 1-6.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010294622.5A CN113541986B (en) | 2020-04-15 | 2020-04-15 | Fault prediction method and device for 5G slice and computing equipment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010294622.5A CN113541986B (en) | 2020-04-15 | 2020-04-15 | Fault prediction method and device for 5G slice and computing equipment |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113541986A CN113541986A (en) | 2021-10-22 |
| CN113541986B true CN113541986B (en) | 2023-04-18 |
Family
ID=78088227
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010294622.5A Active CN113541986B (en) | 2020-04-15 | 2020-04-15 | Fault prediction method and device for 5G slice and computing equipment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113541986B (en) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114143833A (en) * | 2021-10-29 | 2022-03-04 | 北京励图锐新科技有限公司 | Data slice transmission method |
| CN114422332B (en) * | 2022-03-16 | 2022-07-12 | 深圳艾灵网络有限公司 | Network slice control method, device, processing equipment and storage medium |
Family Cites Families (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20180317134A1 (en) * | 2017-04-28 | 2018-11-01 | Huawei Technologies Co., Ltd. | Nssmf nsmf interaction connecting virtual 5g networks and subnets |
| CN109936464B (en) * | 2017-12-15 | 2022-04-01 | 中国移动通信集团浙江有限公司 | Communication capacity open load prediction method and device based on multilayer LSTMRNN |
| CN110138595A (en) * | 2019-04-12 | 2019-08-16 | 中国科学院深圳先进技术研究院 | Time link prediction technique, device, equipment and the medium of dynamic weighting network |
| CN110188176B (en) * | 2019-04-30 | 2022-12-23 | 深圳大学 | Deep learning neural network, and training and predicting method, system, device and medium |
| CN110264709B (en) * | 2019-05-06 | 2021-02-12 | 北京交通大学 | Method for predicting traffic flow of road based on graph convolution network |
| CN110932962B (en) * | 2019-11-20 | 2021-11-30 | 重庆邮电大学 | Network structure de-anonymization system and method based on matrix decomposition |
-
2020
- 2020-04-15 CN CN202010294622.5A patent/CN113541986B/en active Active
Also Published As
| Publication number | Publication date |
|---|---|
| CN113541986A (en) | 2021-10-22 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110263280B (en) | Multi-view-based dynamic link prediction depth model and application | |
| CN111368888B (en) | Service function chain fault diagnosis method based on deep dynamic Bayesian network | |
| CN111612153A (en) | Method and device for training model | |
| CN113541986B (en) | Fault prediction method and device for 5G slice and computing equipment | |
| CN112217674B (en) | Alarm root cause identification method based on causal network mining and graph attention network | |
| CN111431819A (en) | Network traffic classification method and device based on serialized protocol flow characteristics | |
| CN115358487A (en) | Federal learning aggregation optimization system and method for power data sharing | |
| CN111767472A (en) | Method and system for detecting abnormal account of social network | |
| CN105471647A (en) | Power communication network fault positioning method | |
| CN113869521A (en) | Method, device, computing equipment and storage medium for constructing prediction model | |
| CN116489038A (en) | Network traffic prediction method, device, equipment and medium | |
| CN112202584B (en) | Alarm correlation method, device, computing equipment and computer storage medium | |
| CN111027591A (en) | Node fault prediction method for large-scale cluster system | |
| CN113825165B (en) | 5G slice network congestion early warning method and device based on time diagram network | |
| CN112396166A (en) | Graph convolution neural network training method and device based on mixed granularity aggregator | |
| CN111935005A (en) | Data transmission method, device, processing equipment and medium | |
| CN113765683A (en) | Cutover alarm shielding method and device for network slices | |
| CN115001937B (en) | Smart city Internet of things-oriented fault prediction method and device | |
| CN113822454B (en) | Method and device for predicting slice complaint processing resources | |
| CN113055218A (en) | Redundancy evaluation method and device for NFV network and computing equipment | |
| CN113411841B (en) | 5G slice cutting and joining method and device and computing equipment | |
| CN115913979A (en) | Global network topology processing method and device | |
| CN113822453B (en) | Multi-user complaint commonality determining method and device for 5G slices | |
| CN113949633A (en) | 5G network slice disaster recovery pool resource management method and device based on machine learning | |
| CN113806498A (en) | Online question-answering method and device for network slice instance running state |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |