Qci matching method, apparatus, device, and storage medium

Abstract

The application discloses a QCI matching method and device, equipment and storage medium, and belongs to the technical field of mobile communication. When receiving an industry application request information sent by a user, industry application characteristic attributes are extracted from the industry application request information; a quality of service parameter class identifier QCI characteristic attribute allocated to a data channel by a user plane function network element is acquired; the industry application characteristic attributes and the QCI characteristic attribute are input into a QCI matching model, and a matching relationship between an industry application and each QCI output by the QCI matching model is acquired; and a corresponding QCI is matched for the industry application according to the matching relationship, so that the required QCI can be accurately allocated to the industry application.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of mobile communication technology, and in particular to a QCI matching method, device, equipment and storage medium. BACKGROUND

[0002] Currently, 3GPP allocates new QCI for Data channel: 71 (150ms delay, 10-6 packet loss rate), 72 (150ms delay, 10-6 packet loss rate), 73 (150ms delay, 10-6 packet loss rate), 74 (150ms delay, 10-6 packet loss rate), 76 (150ms delay, 10-6 packet loss rate). But the Data channel transmits diversified application content, and the media type and format of transmission are also various, and different industry applications need to be allocated with the required QCI. But currently, 3GPP does not clearly define the required QCI matching for different types of content in the Data channel, that is, it is unable to allocate the required QCI for different industry applications.

[0003] The above content is only used to assist in understanding the technical solutions of the present application, and does not represent the acknowledgement of the above content as prior art. SUMMARY

[0004] The main purpose of the present application is to provide a QCI matching method, device, equipment and storage medium, aiming at solving the technical problem that the prior art cannot allocate the required QCI for different industry applications.

[0005] To achieve the above purpose, the present application provides a QCI matching method, which comprises the following steps:

[0006] When receiving the industry application request information sent by the user, extracting the industry application characteristic attribute from the industry application request information;

[0007] Obtaining the quality of service parameter class identifier QCI characteristic attribute allocated by the user plane function network element to the data channel;

[0008] Inputting the industry application characteristic attribute and the QCI characteristic attribute into the QCI matching model, and obtaining the matching relationship between the industry application and each QCI output by the QCI matching model;

[0009] According to the matching relationship, matching the corresponding QCI for the industry application.

[0010] Optionally, before the industry application characteristic attribute and the QCI characteristic attribute are input into the QCI matching model, the method further comprises:

[0011] obtaining a first attribute type of an industry application feature attribute and a second attribute type of a QCI feature attribute;

[0012] preprocessing the industry application feature attribute according to a first data processing manner corresponding to the first attribute type to obtain a target industry application feature attribute;

[0013] preprocessing the QCI feature attribute according to a second data processing manner corresponding to the second attribute type to obtain a target QCI feature attribute;

[0014] Correspondingly, the inputting the industry application feature attribute and the QCI feature attribute into the QCI matching model comprises:

[0015] inputting the target industry application feature attribute and the target QCI feature attribute into the QCI matching model.

[0016] Optionally, before the inputting the industry application feature attribute and the QCI feature attribute into the QCI matching model, the method further comprises:

[0017] obtaining historical industry application feature attributes and effective QCI feature attributes from a server;

[0018] obtaining a historical matching degree score between the historical industry application feature attributes and the effective QCI feature attributes;

[0019] constructing a total data set according to the historical industry application feature attributes, the effective QCI feature attributes and the historical matching degree score;

[0020] selecting a target data set from the total data set according to a preset proportion;

[0021] training a preset neural network model according to the target data set to obtain a QCI matching model.

[0022] Optionally, the training the preset neural network model according to the target data set to obtain the QCI matching model comprises:

[0023] setting the preset neural network model according to a preset training parameter to obtain a reference neural network model;

[0024] inputting the historical industry application feature attributes and the effective QCI feature attributes in the target data set into the reference neural network model to obtain a predicted matching degree score;

[0025] obtaining a score error between the predicted matching degree score and the historical matching degree score in the target data set;

[0026] The reference neural network model is optimized according to the scoring error to obtain a QCI matching model.

[0027] Optionally, the obtaining of the matching relationship between the industry application and each QCI includes:

[0028] The industry application feature attribute and the QCI feature attribute are respectively extracted by a long short-term memory layer and a discard layer in the QCI matching model to obtain an industry application feature vector and a QCI feature vector;

[0029] The industry application feature vector and the QCI feature vector are spliced and extracted by a merging layer, a full connection layer and a discard layer in the QCI matching model to obtain the matching relationship between the industry application and each QCI.

[0030] Optionally, the matching of the corresponding QCI for the industry application according to the matching relationship includes:

[0031] The matching degree score between the industry application and each QCI is determined according to the matching relationship.

[0032] A target matching degree score is screened out from the matching degree scores.

[0033] The QCI corresponding to the target matching degree score is matched for the industry application by the user plane function network element.

[0034] Optionally, after the matching of the corresponding QCI for the industry application according to the matching relationship, the method further includes:

[0035] The industry application feature attribute is detected in real time whether to be changed.

[0036] When the industry application feature attribute is detected to be changed, the changed industry application feature is obtained, and the changed industry application feature is input into the QCI matching model to obtain a new matching relationship.

[0037] The corresponding QCI is matched for the industry application according to the new matching relationship.

[0038] In addition, in order to achieve the above-mentioned purpose, the application further provides a QCI matching device, which includes:

[0039] The receiving module is configured to extract an industry application feature attribute from the industry application request information when receiving the industry application request information sent by the user.

[0040] The extraction module is configured to extract an industry application feature attribute from the industry application request information.

[0041] The acquisition module is configured to acquire a quality of service parameter class identifier (QCI) characteristic attribute allocated to a data channel by a user plane function network element.

[0042] The input module is configured to input the industry application characteristic attribute and the QCI characteristic attribute into a QCI matching model, and obtain a matching relationship between the industry application and each QCI output by the QCI matching model.

[0043] The matching module is configured to match a corresponding QCI for the industry application according to the matching relationship.

[0044] In addition, to achieve the above object, the application further provides a QCI matching device, which comprises a memory, a processor, and a QCI matching program stored in the memory and executable on the processor, and the QCI matching program is configured to implement the QCI matching method as described above.

[0045] In addition, to achieve the above object, the application further provides a storage medium, which stores a QCI matching program, and the QCI matching program is executed by a processor to implement the QCI matching method as described above.

[0046] When receiving the industry application request information sent by the user, the application extracts the industry application characteristic attribute from the industry application request information, acquires the quality of service parameter class identifier (QCI) characteristic attribute allocated to the data channel by the user plane function network element, inputs the industry application characteristic attribute and the QCI characteristic attribute into the QCI matching model, and obtains the matching relationship between the industry application and each QCI output by the QCI matching model. According to the matching relationship, the corresponding QCI is matched for the industry application. By inputting the industry application characteristic attribute QCI and the QCI characteristic attribute into the QCI matching model, the matching relationship between the industry application and each QCI is obtained, and the corresponding QCI is matched for the industry application according to the matching relationship, so that the required QCI can be accurately allocated to the industry application. BRIEF DESCRIPTION OF DRAWINGS

[0047] Figure 1 FIG. 1 is a structural schematic diagram of a QCI matching device of a hardware running environment related to an embodiment scheme of the application;

[0048] Figure 2 FIG. 2 is a flowchart of a first embodiment of the QCI matching method of the application;

[0049] Figure 3 FIG. 3 is a schematic diagram of a long short-term memory neuron in the QCI matching method of the application;

[0050] Figure 4 FIG. 4 is a flowchart of a second embodiment of the QCI matching method of the application;

[0051] Figure 5 This is a flowchart illustrating the third embodiment of the QCI matching method of the present invention;

[0052] Figure 6 This is a schematic diagram of the multi-branch long short-term memory neural network model of the QCI matching method of the present invention;

[0053] Figure 7 This is a structural block diagram of the first embodiment of the QCI matching device of the present invention.

[0054] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0055] It should be understood that the specific embodiments described herein are for illustrative purposes only and are not intended to limit the scope of the invention.

[0056] Reference Figure 1 , Figure 1 This is a schematic diagram of the QCI matching device structure of the hardware operating environment involved in the embodiments of the present invention.

[0057] like Figure 1 As shown, the QCI matching device may include: a processor 1001, such as a central processing unit (CPU), a communication bus 1002, a user interface 1003, a network interface 1004, and a memory 1005. The communication bus 1002 is used to enable communication between these components. The user interface 1003 may include a display screen or an input unit such as a keyboard; optionally, the user interface 1003 may also include a standard wired interface or a wireless interface. The network interface 1004 may optionally include a standard wired interface or a wireless interface (such as a Wireless-Fidelity (Wi-Fi) interface). The memory 1005 may be a high-speed random access memory (RAM) or a stable non-volatile memory (NVM), such as a disk drive. The memory 1005 may also optionally be a storage device independent of the aforementioned processor 1001.

[0058] Those skilled in the art will understand that Figure 1 The structure shown does not constitute a limitation on the QCI matching device and may include more or fewer components than shown, or combine certain components, or have different component arrangements.

[0059] like Figure 1As shown, the memory 1005 as a storage medium can include an operating system, a network communication module, a user interface module and a QCI matching program.

[0060] In Figure 1 In the QCI matching device shown, the network interface 1004 is mainly used for data communication with a network server; the user interface 1003 is mainly used for data interaction with a user; the processor 1001 and the memory 1005 in the QCI matching device of the application can be arranged in the QCI matching device, and the QCI matching device calls the QCI matching program stored in the memory 1005 through the processor 1001, and executes the QCI matching method provided by the embodiments of the application.

[0061] The embodiments of the application provide a QCI matching method, which refers to Figure 2 , Figure 2 The flowchart of a first embodiment of the QCI matching method of the application is shown.

[0062] In this embodiment, the QCI matching method includes the following steps:

[0063] Step S10: When receiving the industry application request information sent by the user, extracting the industry application characteristic attribute from the industry application request information.

[0064] It should be noted that the execution subject of the embodiment can be a QCI matching device, and the QCI matching device can be a personal computer, a server or a vehicle-mounted terminal, etc. electronic device, and can also be other devices or servers that can realize the same or similar functions, and the embodiments are not limited thereto. In the embodiments and the following embodiments, the QCI matching method of the application is described by taking the QCI matching device as an example.

[0065] It should be noted that QCI (QoS Class Identifier) is a scale value, QoS (Quality of Service), Class Identifier represents a class identifier, which is used to measure the specific packet forwarding behavior (such as packet loss rate, packet delay budget) provided to SDF (Service Data Flow), which is applied to both GBR and Non-GBR bearers, and is used to specify the control bearer level packet forwarding mode (such as scheduling weight, admission threshold, queue management threshold, link layer protocol configuration, etc.) defined in the access node, which is pre-configured by the operator into the access network node. The industry application in the embodiment takes the 5G new call industry application as an example for illustration. Industry applications such as autonomous driving, intelligent transportation, and smart factories, etc. When the industry application needs to be carried out through 5G, the user will send the corresponding industry application request information to the QCI matching device. When the industry application request information is received, the media type, required delay, required bandwidth, and application scene, etc. multiple industry application characteristic attributes are extracted from the industry application request information. The industry application characteristic attributes can be extracted according to actual needs, and the present embodiment does not limit this.

[0066] Step S20: Obtain the QCI characteristic attribute of the service quality parameter allocated to the data channel by the user plane function network element.

[0067] In the present embodiment, the user plane function network element (User plane function, UPF) has the functions of user plane QoS processing and user plane policy rule implementation, and can allocate corresponding QCI characteristic attributes to the data channel. In the present embodiment, the QCI characteristic attributes include delay, packet loss rate, and priority, etc. The QCI characteristic attributes can be set according to actual needs, and the present embodiment does not limit this.

[0068] Step S30: Input the industry application characteristic attribute and the QCI characteristic attribute into the QCI matching model, and obtain the matching relationship between the industry application and each QCI output by the QCI matching model.

[0069] It should be noted that in the 5G era, operators can upgrade user call experience based on VoNR / VoLTE high-definition video call service to improve traffic revenue with the advantages that OTT applications do not have, such as APP installation, number-based, and deterministic network experience guarantee. First, cultivate user habits, seize users, and layout the future. GSMA proposes Enriched Calling in RCS, which introduces enhanced capabilities in calls and defines three scenarios: pre-call, in-call, and post-call. Before calling, the caller can send call topics, importance, pictures, location, and other information to the callee; during the call, the caller and the callee can send each other arbitrary files, send IM messages, share locations, share maps, share graffiti, and real-time video; when the callee does not answer, the caller can send a text message or a voice message to the callee. Operators can upgrade VoNR / VoLTE high-definition video call services based on the three scenarios of pre-call, in-call, and post-call. LSTM neurons are shown in Figure 4

[0070] 3GPP also formulates IMS Data Channel standard, which is based on VoLTE / VoNR high-definition audio and video calls, combined with WebRTC technology, provides data channel IMS Data Channel through extension, synchronizes voice and video calls with extended data channel, so as to realize screen sharing, superimposed AR, and even auditory, visual, tactile, and dynamic synchronous full-immersive experience in high-definition video calls. IMS Data Channel provides high real-time single-stream or multi-stream data interaction channel based on UDP, which can simultaneously perform desktop sharing, whiteboard sharing, and file sending during the call process.

[0071] By introducing real-time interactive channel (IMS Data Channel), industry-specific interactive applications are added in the process of high-definition real-time multimedia calls for vertical industries. End-to-end QoS guarantee is not only voice and video two specific QoS strategies and execution, but also customized QoS policy control for different data business types, supporting media QoS attribute negotiation according to multiple scenarios. IMS Data Channel does not concern the content and format of the channel, and only needs to agree on the communication format between the two parties. It can use the Webpage+JavaScript script general way to deliver diversified application content through IMS Data Channel.

[0072] ​It needs to be emphasized that the diversified application contents delivered in the data channel, the delivered media types and formats are also various, and the current 3GPP cannot allocate the required QCI for different industry applications, in the embodiment, the matching degree between the 5G industry applications and the supported QCI of the existing data channel is automatically learned by using the multi-branch long short-term memory neural network, so as to allocate appropriate QCI for the 5G industry applications.

[0073] It needs to be explained that the long short-term memory (LSTM) in the embodiment is a special type of recurrent neural network. The so-called recurrent neural network (RNN) is a kind of neural network with memory, the output of each hidden layer in the RNN is stored in the cache, when the next time the hidden layer has data input, the data in the cache can also be regarded as part of the input, at each time point, the output of the neuron is put into the cache, and at the next time point, the value in the cache is overwritten. Compared with RNN, LSTM can learn long-term dependence information, by controlling the time of saving the value in the cache, long-term information can be remembered, which is suitable for time series prediction. Each neuron has four inputs and one output, and each neuron has a Cell to store memory values. As shown in the formula (1), each LSTM neuron contains three gates: forget gate, input gate and output gate. The LSTM neuron contains 6 formulas, and the LSTM neuron Figure 3 In the formula 1, the forget gate is represented, in the formulas 2 and 3, new information is added, the formula 4 fuses new information and old information, the formulas 5 and 6 output the information about the next time stamp that the LSTM unit has learned. The long short-term memory neural network has good effect on long time series learning, each connection line in the LSTM unit contains a corresponding weight, Xt represents the input vector, ht represents the hidden state, Ct represents the neuron state at t time, W is a trainable weight matrix, and b is a bias vector. By creating a QCI matching model, then the industry application feature attributes and the QCI feature attributes are used as the input of the model, after the model processing, the input result can be obtained, and the input result is the matching relationship between the industry application and each QCI.

[0074] ​Further, in order to improve the convergence speed and accuracy of the model, so that the QCI matching model can allocate more accurate QCI for industry applications, the embodiment further includes, before step S30: obtaining a first attribute type of the industry application characteristic attribute and a second attribute type of the QCI characteristic attribute; pre-processing the industry application characteristic attribute according to a first data processing mode corresponding to the first attribute type to obtain a target industry application characteristic attribute; and pre-processing the QCI characteristic attribute according to a second data processing mode corresponding to the second attribute type to obtain a target QCI characteristic attribute.

[0075] It should be noted that, in the embodiment, the industry application characteristic attribute and the QCI characteristic attribute need to be pre-processed before being input into the QCI matching model. Specifically, the characteristic attributes are processed according to different data processing modes corresponding to the attribute types. In the embodiment, the first attribute type of the industry application characteristic attribute and the second attribute type of the QCI characteristic attribute can be obtained. The data processing mode corresponding to the first attribute type is the first data processing mode, and the data processing mode corresponding to the second attribute type is the second data processing mode. In the embodiment, the industry application characteristic attribute includes n attributes such as media type, required delay, required bandwidth, and application scenario, which can be represented as {x1, x2, x3,..., xn}; the QCI characteristic attribute of the data channel includes m attributes such as delay, packet loss rate, and priority, which can be represented as {a1, a2, a3,..., am}. After converting the above attributes into a machine recognizable form, the corresponding attribute types can be determined. The attribute types include non-numeric types and numeric types. The data processing mode corresponding to the non-numeric type attribute type is one-hot encoding, and the data processing mode corresponding to the numeric type attribute type is normalization processing, which is normalized to a mean of 0 and a variance of 1. Of course, other data processing modes can also be used in the embodiment, which can be set according to actual conditions, and the embodiment does not limit this. n m

[0076] Step S40: matching a corresponding QCI for the industry application according to the matching relationship.

[0077] In specific implementation, after determining the matching relationship, the most matched QCI among the QCIs can be determined based on the matching relationship, and the most matched QCI is allocated to the industry application.

[0078] ​​Further, in order to make the QCI matching more flexible and intelligent, the step S40 further includes: detecting whether the industry application characteristic attribute is changed in real time; when it is detected that the industry application characteristic attribute is changed, obtaining the changed industry application characteristic, and inputting the changed industry application characteristic into the QCI matching model to obtain a new matching relationship; and matching the industry application with a corresponding QCI according to the new matching relationship.

[0079] It should be noted that the industry application characteristic attribute is changed in real time, for example, the user needs to use different industry applications or requires different delay with the development of the same industry application, and the industry application characteristic attribute can be detected in the embodiment, and the changed industry application characteristic attribute is re-input into the QCI matching model, so as to allocate a new QCI to the industry application. It should be emphasized that the industry application characteristic attribute is changed, and the QCI characteristic attribute is not changed.

[0080] The embodiment extracts the industry application characteristic attribute from the industry application request information when receiving the industry application request information sent by the user, obtains the QCI characteristic attribute allocated to the data channel by the user plane function network element, inputs the industry application characteristic attribute and the QCI characteristic attribute into the QCI matching model, and obtains the matching relationship between the industry application and each QCI output by the QCI matching model. According to the matching relationship, the corresponding QCI is matched for the industry application. By inputting the industry application characteristic attribute QCI and the QCI characteristic attribute into the QCI matching model, the matching relationship between the industry application and each QCI is obtained, and the corresponding QCI is matched for the industry application according to the matching relationship, so that the required QCI can be accurately allocated to the industry application.

[0081] Reference Figure 4 , Figure 4 is a flowchart of a second embodiment of a QCI matching method.

[0082] Based on the above first embodiment, the QCI matching method of the embodiment further includes, before the step S30:

[0083] Step S030: obtaining historical industry application characteristic attributes and effective QCI characteristic attributes from a server.

[0084] It is easy to understand that before the industry application characteristic attribute and the QCI characteristic attribute are input into the QCI matching model, the QCI matching model needs to be constructed. In the embodiment, the model can be created and trained according to the historical industry application characteristic attributes and the effective QCI characteristic attributes. The effective QCI characteristic attribute represents the available QCI in the data channel.

[0085] Step S130: obtaining a historical matching degree score between the historical industry application feature attribute and the effective QCI feature attribute.

[0086] In a specific implementation, the historical industry application feature attribute also has a matching relationship with the effective QCI, and the historical matching degree score between the historical industry application feature attribute and the effective QCI feature attribute can be determined according to the matching relationship. The historical matching degree score is obtained by manually scoring based on expert experience.

[0087] Step S230: constructing a total data set according to the historical industry application feature attribute, the effective QCI feature attribute, and the historical matching degree score.

[0088] Step S330: selecting a target data set from the total data set according to a preset proportion.

[0089] It should be noted that the training and construction of the model need a corresponding data set. In this embodiment, the total data set can be constructed according to the historical industry application feature attribute, the effective QCI feature attribute, and the historical matching degree score. It should be emphasized that after the training of the model, testing and verification are needed, and the testing and verification of the model need to rely on a corresponding test set. In this embodiment, the training set and the test set can be selected from the total data set, the preset proportion can be set to 80%, and other proportions can also be set. The proportion can be adjusted according to the actual situation, and this embodiment does not limit this. Taking the preset proportion of 80% as an example, assuming that the preset proportion is 80%, 80% of the total data set can be obtained as the model training set, that is, the target data set, and 20% of the total data set can be obtained as the model test set.

[0090] Step S430: training the preset neural network model according to the target data set to obtain a QCI matching model.

[0091] It should be noted that, as described above, this embodiment is used to assign QCI to the industry application of 5G new calls by using a long short-term memory neural network model. In this embodiment, the long short-term memory neural network can be used as the preset neural network model, and then the target data set described above can be used as the training data to train the long short-term memory neural network, so that the QCI matching model can be obtained.

[0092] Further, in order to make the trained model more accurate and reduce errors, the step S403 specifically comprises: setting a preset neural network model according to preset training parameters to obtain a reference neural network model; inputting the historical industry application feature attributes and the effective QCI feature attributes in the target data set into the reference neural network model to obtain a predicted matching degree score; obtaining a score error between the predicted matching degree score and the historical matching degree score in the target data set; and optimizing the reference neural network model according to the score error to obtain a QCI matching model.

[0093] It should be noted that the preset training parameters in the embodiment include training epochs, batch size and loss function, i.e., objective function, etc. For example, the model is trained for 1000 epochs (epochs = 1000), the batch size is set to 10 (batch_size = 10), and the mean absolute error MSE (Mean Squared Error) is selected as the loss function, i.e., the objective function (loss ='mse'). where y i and The matching degree score is identified. After inputting the historical industry application feature attributes and the effective QCI feature attributes into the reference neural network model, a predicted matching degree score output by the model can be obtained, and finally the score error between the predicted matching degree score and the historical matching degree score can be obtained according to the above loss function.

[0094] In a specific implementation, the model optimization adopts a gradient descent optimization algorithm, and an adam optimizer is selected to improve the learning speed of the traditional gradient descent (optimizer = 'adam'). The neural network can find the optimal weight value that minimizes the objective function through gradient descent. With the increase of the number of training epochs, the training error gradually decreases, and the model gradually converges.

[0095] The embodiment obtains the historical industry application feature attributes and the effective QCI feature attributes from the server, obtains the historical matching degree score between the historical industry application feature attributes and the effective QCI feature attributes, constructs a total data set according to the historical industry application feature attributes, the effective QCI feature attributes and the historical matching degree score, selects a target data set from the total data set according to a preset proportion, and trains a preset neural network model according to the target data set to obtain a QCI matching model. The QCI matching model is more accurate by selecting a target data set from the historical industry application feature attributes, the effective QCI feature attributes and the historical matching degree score to construct a total data set for training.

[0096] Reference Figure 5 ,Figure 5 A flowchart of a third embodiment of the QCI matching method of the application is shown in the figure.

[0097] Based on the first embodiment, the third embodiment of the QCI matching method of the application is proposed.

[0098] In this embodiment, the step S30 comprises:

[0099] Step S301: Feature extraction is performed on the industry application feature attributes and the QCI feature attributes respectively through the long short-term memory layer and the dropout layer in the QCI matching model to obtain industry application feature vectors and QCI feature vectors.

[0100] Step S302: The industry application feature vectors and the QCI feature vectors are spliced and extracted through the merging layer, the fully connected layer and the dropout layer in the QCI matching model to obtain the matching relationship between the industry application and each QCI.

[0101] It should be noted that in this embodiment, a multi-branch long short-term memory neural network model is built using a deep learning framework, and the model is as shown in the figure. Figure 6 Each branch includes an input layer, two long short-term memory layers and two dropout layers. The two parallel input layers input the encoded 5G new call industry application attributes and the available QCI of the Data channel respectively. The long short-term memory layers are respectively set to 64 and 32 LSTM neurons, and the activation function is set to'relu'. The dropout layer is introduced after each LSTM layer to effectively avoid overfitting. The dropout layer refers to discarding neurons with a probability p and allowing other neurons to remain with a probability q=1-p. In this scheme, the dropout probability is set to 0.2, i.e., 20% of the neurons are randomly ignored and disabled. The merging layer (concatenate) splices the spatial vectors of the two types of data in the column dimension, and inputs them to the fully connected layer and the dropout layer to extract the merged feature vectors and automatically extract the matching relationship between the industry application and each QCI.

[0102] Further, in order to more accurately match the corresponding QCI for the industry application, the step S40 specifically comprises: determining the matching degree score between the industry application and each QCI according to the matching relationship; screening out a target matching degree score from the matching degree scores; and matching the QCI corresponding to the target matching degree score for the industry application through the user plane function network element.

[0103] It should be noted that according to the determined matching relationship, the matching degree between the industry application and each QCI can be determined, and different matching degrees correspond to different matching degree scores, for example, the higher the matching degree, the higher the corresponding matching degree score, and in the embodiment, the target matching degree score is the highest matching degree score among all matching degree scores, for example, the score ranking between matching degree scores y1, y2 and y3 is y1, y2 and y3, and then y1 can be determined as the target matching degree score.

[0104] In the embodiment, the industry application feature attributes and the QCI feature attributes are extracted by the long short memory layer and the discard layer in the QCI matching model respectively to obtain industry application feature vectors and QCI feature vectors, and the industry application feature vectors and the QCI feature vectors are spliced and extracted by the merging layer, the full connection layer and the discard layer in the QCI matching model to obtain the matching relationship between the industry application and each QCI, and the matching relationship between the industry application and each QCI is more accurately obtained through the long short term memory layer, the discard layer, the merging layer and the full connection layer, and finally the corresponding score of the matching degree relationship is accurately assigned to the corresponding QCI for the industry application.

[0105] In addition, the embodiment of the application further provides a storage medium, wherein the storage medium stores a QCI matching program, and the QCI matching program is executed by a processor to realize the steps of the QCI matching method as described above.

[0106] Reference Figure 7 , Figure 7 The structure block diagram of the first embodiment of the QCI matching device of the application is shown in the figure.

[0107] As Figure 7 shown, the QCI matching device provided by the embodiment of the application comprises:

[0108] The receiving module 10 is configured to extract the industry application feature attributes from the industry application request information sent by the user when receiving the industry application request information.

[0109] The extraction module 20 is configured to extract the industry application feature attributes from the industry application request information.

[0110] The acquisition module 30 is configured to acquire the quality of service parameter class identifier QCI feature attributes assigned to the data channel by the user plane function network element.

[0111] The input module 40 is configured to input the industry application feature attributes and the QCI feature attributes into the QCI matching model, and obtain the matching relationship between the industry application and each QCI output by the QCI matching model.

[0112] The matching module 50 is configured to match the QCI corresponding to the industry application according to the matching relationship.

[0113] According to the embodiment, the industry application feature attribute is extracted from the industry application request information when the industry application request information is received, the QCI feature attribute is obtained from the service quality parameter category identifier QCI assigned by the user plane function network element to the data channel, the industry application feature attribute and the QCI feature attribute are input into the QCI matching model, and the matching relationship between the industry application and each QCI output by the QCI matching model is obtained. According to the matching relationship, the QCI corresponding to the industry application is matched by inputting the industry application feature attribute QCI and the QCI feature attribute into the QCI matching model, and the matching relationship between the industry application and each QCI is matched. According to the matching relationship, the QCI corresponding to the industry application is matched, and the required QCI can be accurately assigned to the industry application.

[0114] It should be understood that the above is only an example, and does not constitute any limitation on the technical solutions of the present application. In specific applications, those skilled in the art can set up according to the needs, and the present application does not limit this.

[0115] It should be noted that the above-described workflow is only illustrative and does not limit the scope of protection of the present application. In actual application, those skilled in the art can select part or all of them to achieve the purpose of the embodiment according to the actual needs, which is not limited here.

[0116] In addition, technical details not described in detail in the embodiment can be referred to the QCI matching method provided by any embodiment of the present application, which will not be repeated here.

[0117] In addition, it should be noted that in this paper, the term "includes", "contains" or any other variant thereof is intended to cover non-exclusive inclusion, so that the process, method, article or system including a series of elements not only includes those elements, but also includes other elements not explicitly listed, or includes elements inherent to such process, method, article or system. Without more limitations, the element defined by the statement "includes a" does not exclude the presence of other identical elements in the process, method, article or system including the element.

[0118] The above embodiment number of the present application is only for description, not representing the advantages and disadvantages of the embodiments.

[0119] Those skilled in the art can clearly understand the above-mentioned embodiment method can be realized by means of software and the necessary general hardware platform, of course, can also be through hardware, but in many cases the former is a better embodiment. Based on such understanding, the technical solutions of the present application essentially or say the part of the prior art contribution can be embodied in the form of software products, the computer software product is stored in a storage medium (such as read only memory (Read Only Memory, ROM) / RAM, disk, optical disk), including a number of instructions to make a terminal device (may be a mobile phone, computer, server, or network equipment, etc.) executes the method described in various embodiments of the present application.

[0120] The above is only the preferred embodiment of the present application, not therefore limit the patent scope of the present application, any equivalent structure or equivalent flow transformation made by using the content of the present application specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present application.

Claims

1. A QCI matching method, characterized in that, The QCI matching method includes: Upon receiving an industry application request from a user, extract industry application feature attributes from the industry application request. Obtain the Quality of Service (QoS) parameters, category identifiers (QCI) and characteristic attributes assigned to the data channel by user plane functional network elements; The industry application feature attributes and the QCI feature attributes are input into the QCI matching model, and the matching relationship between the industry application and each QCI is obtained from the output of the QCI matching model. The QCI matching model includes a long short-term memory layer, a discard layer, a merging layer and a fully connected layer. The QCI matching model refers to a model that automatically learns the matching degree between 5G industry applications and QCIs supported by existing data channels based on a multi-branch long short-term memory neural network, and assigns appropriate QCIs to 5G industry applications. Based on the matching relationship, match the corresponding QCI for industry applications; Before inputting the industry application feature attributes and the QCI feature attributes into the QCI matching model, the method further includes: The first attribute type of the industry application characteristic attribute and the second attribute type of the QCI characteristic attribute are obtained. The attribute type includes non-numerical and numerical types. The data processing method corresponding to the non-numerical attribute type is one-hot encoding, and the data processing method corresponding to the numerical attribute type is normalization processing, which is normalized to a mean of 0 and a variance of 1. The industry application feature attributes are preprocessed according to the first data processing method corresponding to the first attribute type to obtain the target industry application feature attributes; The QCI feature attributes are preprocessed according to the second data processing method corresponding to the second attribute type to obtain the target QCI feature attributes; Accordingly, inputting the industry application feature attributes and the QCI feature attributes into the QCI matching model includes: The target industry application feature attributes and the target QCI feature attributes are input into the QCI matching model.

2. The QCI matching method as described in claim 1, characterized in that, Before inputting the industry application feature attributes and the QCI feature attributes into the QCI matching model, the following steps are also included: Retrieve historical industry application characteristic attributes and valid QCI characteristic attributes from the server; Obtain the historical matching score between the historical industry application feature attributes and the effective QCI feature attributes; A total dataset is constructed based on the historical industry application feature attributes, the effective QCI feature attributes, and the historical matching score. The target dataset is selected from the total dataset according to a preset ratio; The preset neural network model is trained according to the target dataset to obtain the QCI matching model.

3. The QCI matching method as described in claim 2, characterized in that, The step of training a preset neural network model according to the target dataset to obtain a QCI matching model includes: The preset neural network model is set according to the preset training parameters to obtain the reference neural network model; The historical industry application feature attributes and effective QCI feature attributes in the target dataset are input into the reference neural network model to obtain the predicted matching score; Obtain the score error between the predicted matching score and the historical matching scores in the target dataset; The reference neural network model is optimized based on the scoring error to obtain a QCI matching model.

4. The QCI matching method as described in claim 1, characterized in that, The process of obtaining the matching relationship between the industry application output by the QCI matching model and each QCI includes: The industry application feature attributes and the QCI feature attributes are extracted by the long short memory layer and the discard layer in the QCI matching model, respectively, to obtain the industry application feature vector and the QCI feature vector. The industry application feature vector and the QCI feature vector are concatenated and extracted by the merging layer, fully connected layer and discarding layer in the QCI matching model to obtain the matching relationship between the industry application and each QCI.

5. The QCI matching method as described in claim 1, characterized in that, The step of matching the corresponding QCI for industry applications based on the matching relationship includes: Based on the matching relationship, determine the matching degree score between the industry application and each QCI; Target matching scores are filtered from the matching scores; The user plane function network element is used to match the industry application with the QCI corresponding to the target matching score.

6. The QCI matching method as described in any one of claims 1 to 5, characterized in that, After matching the corresponding QCI for the industry application according to the matching relationship, the method further includes: Real-time detection of whether the industry application characteristic attributes change; When a change in the industry application feature attribute is detected, the changed industry application feature is obtained and input into the QCI matching model to obtain a new matching relationship; Based on the new matching relationship, the corresponding QCI is matched for industry applications.

7. A QCI matching device, characterized in that, The QCI matching device includes: The receiving module is used to extract industry application feature attributes from the industry application request information sent by the user when it receives the industry application request information. The extraction module is used to extract industry application feature attributes from the industry application request information; The acquisition module is used to acquire the Quality of Service (QoS) parameter category identifier (QCI) characteristic attributes assigned to the data channel by the user plane function network element. The input module is used to input the industry application feature attributes and the QCI feature attributes into the QCI matching model, and obtain the matching relationship between the industry application and each QCI output by the QCI matching model. The QCI matching model includes a long short-term memory layer, a discard layer, a merging layer and a fully connected layer. The QCI matching model refers to a model that automatically learns the matching degree between 5G industry applications and QCIs supported by existing data channels based on a multi-branch long short-term memory neural network, and assigns appropriate QCIs to 5G industry applications. The matching module is used to match the corresponding QCI for industry applications based on the matching relationship; Before inputting the industry application feature attributes and the QCI feature attributes into the QCI matching model, the method further includes: The first attribute type of the industry application characteristic attribute and the second attribute type of the QCI characteristic attribute are obtained. The attribute type includes non-numerical and numerical types. The data processing method corresponding to the non-numerical attribute type is one-hot encoding, and the data processing method corresponding to the numerical attribute type is normalization processing, which is normalized to a mean of 0 and a variance of 1. The industry application feature attributes are preprocessed according to the first data processing method corresponding to the first attribute type to obtain the target industry application feature attributes; The QCI feature attributes are preprocessed according to the second data processing method corresponding to the second attribute type to obtain the target QCI feature attributes; Accordingly, inputting the industry application feature attributes and the QCI feature attributes into the QCI matching model includes: The target industry application feature attributes and the target QCI feature attributes are input into the QCI matching model.

8. A QCI matching device, characterized in that, The QCI matching device includes: a memory, a processor, and a QCI matching program stored in the memory and executable on the processor, the QCI matching program being configured to implement the QCI matching method as described in any one of claims 1 to 6.

9. A storage medium, characterized in that, The storage medium stores a QCI matching program, which, when executed by a processor, implements the QCI matching method as described in any one of claims 1 to 6.

Images