Network data processing method and device, equipment and storage medium

Abstract

The embodiment of the invention provides a network data processing method and device, equipment and a storage medium. In the method, respective reference data corresponding to a group of network scenes are obtained, and the reference data reflect the performance of the corresponding network scene. Next, for a first scene of the set of network scenes, at least one positive sample data associated with the first scene and at least one negative sample data associated with other scenes of the set of network scenes are selected from the reference data. A contrast model is then generated based on the at least one positive sample data and the at least one negative sample data. The contrast model describes an association relationship between the network scenario and contrast features of the network scenario. The contrast model causes a similarity between the contrast feature and another contrast feature of another network scene to indicate whether the network scene and the another network scene belong to the same network scene.

Description

Technical Field

[0001] The exemplary embodiments disclosed herein generally relate to the field of computer technology, and particularly to methods, apparatus, devices, and computer-readable storage media for network data processing. Background Technology

[0002] Currently, the internet provides users with various services, such as communication services. Networks are typically dynamic and subject to change. Sometimes, these changes can impact the services provided. Different network scenarios may exhibit different trends of change and have varying impacts on the services offered. The goal is to optimize the impact of these changing trends on services across different network types. Summary of the Invention

[0003] In a first aspect of this disclosure, a network data processing method is provided. The method includes: acquiring reference data corresponding to a set of network scenarios, the reference data reflecting the performance of the respective network scenarios; for a first scenario in the set of network scenarios, selecting from the reference data at least one positive sample data associated with the first scenario and at least one negative sample data associated with other scenarios in the set of network scenarios; and generating a comparison model based on the at least one positive sample data and the at least one negative sample data, the comparison model describing the association between comparison features of network scenarios, the comparison model such that the similarity between a comparison feature and another comparison feature of another network scenario indicates whether the network scenario and the other network scenario belong to the same network scenario.

[0004] In a second aspect of this disclosure, a network data processing apparatus is provided. The apparatus includes: an acquisition module configured to acquire reference data corresponding to each of a set of network scenarios, the reference data reflecting the performance of the respective network scenarios; a data selection module configured to, for a first scenario in the set of network scenarios, select from the reference data at least one positive sample data associated with the first scenario and at least one negative sample data associated with other scenarios in the set of network scenarios; and a generation module configured to generate a comparison model based on the at least one positive sample data and the at least one negative sample data, the comparison model describing the association between comparison features of network scenarios, and the comparison model such that the similarity between a comparison feature and another comparison feature of another network scenario indicates whether the network scenario and the other network scenario belong to the same network scenario.

[0005] In a third aspect of this disclosure, an electronic device is provided. The device includes at least one processing unit; and at least one memory coupled to the at least one processing unit and storing instructions for execution by the at least one processing unit. When executed by the at least one processing unit, the instructions cause the device to perform the method of the first aspect.

[0006] In a fourth aspect of this disclosure, a computer-readable storage medium is provided. The medium stores a computer program that, when executed by a processor, implements the method of the first aspect.

[0007] It should be understood that the description in this section is not intended to limit the key or essential features of the embodiments of this disclosure, nor is it intended to restrict the scope of this disclosure. Other features of this disclosure will become readily apparent from the following description. Attached Figure Description

[0008] The above and other features, advantages, and aspects of the embodiments of this disclosure will become more apparent from the accompanying drawings and the following detailed description. In the drawings, the same or similar reference numerals denote the same or similar elements, wherein:

[0009] Figure 1 A schematic diagram of an example environment in which embodiments of the present disclosure can be implemented is shown;

[0010] Figure 2 A block diagram illustrating a process for training a contrastive model according to some embodiments of the present disclosure is shown;

[0011] Figure 3 A block diagram illustrating a process for training a classification model according to some embodiments of the present disclosure is shown;

[0012] Figure 4 A block diagram illustrating a process for processing network data according to some embodiments of the present disclosure is shown;

[0013] Figure 5 A flowchart of a method for processing network data according to some embodiments of the present disclosure is shown;

[0014] Figure 6 A block diagram of an apparatus for network data processing according to some embodiments of the present disclosure is shown; and

[0015] Figure 7 A block diagram of an apparatus capable of implementing several embodiments of the present disclosure is shown. Detailed Implementation

[0016] Embodiments of this disclosure will now be described in more detail with reference to the accompanying drawings. While some embodiments of this disclosure are shown in the drawings, it should be understood that this disclosure can be implemented in various forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided to provide a more thorough and complete understanding of this disclosure. It should be understood that the accompanying drawings and embodiments of this disclosure are for illustrative purposes only and are not intended to limit the scope of protection of this disclosure.

[0017] In the description of embodiments of this disclosure, the term "comprising" and similar terms should be understood as open-ended inclusion, i.e., "including but not limited to". The term "based on" should be understood as "at least partially based on". The term "one embodiment" or "the embodiment" should be understood as "at least one embodiment". The term "some embodiments" should be understood as "at least some embodiments". Other explicit and implicit definitions may also be included below.

[0018] In this document, unless explicitly stated otherwise, performing a step in response to A does not mean that the step is performed immediately after A, but may include one or more intermediate steps.

[0019] It is understood that the data involved in this technical solution (including but not limited to the data itself, the acquisition or use of the data) shall comply with the requirements of relevant laws, regulations and related provisions.

[0020] It is understood that before using the technical solutions disclosed in the various embodiments of this disclosure, users should be informed of the types, scope of use, and usage scenarios of the personal information involved in this disclosure through appropriate means in accordance with relevant laws and regulations, and user authorization should be obtained.

[0021] For example, in response to receiving a user's active request, a prompt message is sent to the user to clearly inform the user that the requested operation will require the acquisition and use of the user's personal information, thereby enabling the user to choose whether to provide personal information to the software or hardware such as electronic devices, applications, servers or storage media that perform the operation of the technical solution disclosed herein, based on the prompt message.

[0022] As an optional but non-restrictive implementation, in response to a user's active request, a prompt message can be sent to the user, such as a pop-up window, where the prompt message can be presented in text format. Furthermore, the pop-up window can also include a selection control allowing the user to choose "agree" or "disagree" to provide personal information to the electronic device.

[0023] It is understood that the above notification and user authorization process are merely illustrative and do not constitute a limitation on the implementation of this disclosure. Other methods that comply with relevant laws and regulations may also be applied to the implementation of this disclosure.

[0024] As briefly described above, dynamic changes in networks can impact network services, such as real-time communication. This dynamic change is a major challenge in real-time communication. To optimize network services, network classification is necessary. Some traditional solutions attempt to classify networks based on objective network metrics and human experience, thereby enabling targeted optimization based on network type. However, these traditional solutions typically rely on manually defined network classification rules. Due to the diverse dynamic changes in networks, manually defined rules are unlikely to cover all network types. Manually defined rules establish a mapping between objective network metrics and network types, failing to uncover the underlying relationships within these metrics. Furthermore, this traditional approach struggles to differentiate between scenarios at the boundaries of adjacent network types.

[0025] To address the above and other potential problems, embodiments of this disclosure propose a scheme for network data processing. In this scheme, reference data corresponding to a set of network scenarios is obtained, the reference data reflecting the performance of the respective network scenarios. Next, for a first scenario in the set of network scenarios, at least one positive sample data associated with the first scenario and at least one negative sample data associated with other scenarios in the set of network scenarios are selected from the reference data. Then, a comparison model is generated based on the at least one positive sample data and the at least one negative sample data. The comparison model describes the association between comparison features of network scenarios. The comparison model makes the similarity between a comparison feature and another comparison feature of another network scenario indicate whether the network scenario and the other network scenario belong to the same network scenario.

[0026] The disclosed scheme utilizes a contrastive model to extract contrastive features from data of a specific network scenario. Through contrastive learning of this model, the contrastive feature representations of network scenarios of the same type can be made as similar as possible, while the contrastive feature representations of different network scenario types can be made as similar as possible. The resulting contrastive features can more easily distinguish the features of different network types, thereby enabling more accurate network classification based on these features.

[0027] Some exemplary embodiments of this disclosure will now be described with reference to the accompanying drawings.

[0028] Figure 1 A block diagram of an example environment 100 in which the implementation of this disclosure can be implemented is shown. Figure 1 In environment 100, it is desired to train and use a model (i.e., network data processing model 130) configured to classify networks based on network data. For example, networks can be classified according to different network scenarios based on their communication data. Figure 1 As shown, environment 100 includes model training system 150 and model application system 152. Figure 1The upper part illustrates the model training phase, and the lower part illustrates the model application phase. Before training, the parameter values ​​of the network data processing model 130 can have initial values, or they can have pre-trained parameter values ​​obtained through a pre-training process. During the training process, the parameter values ​​of the network data processing model 130 can be updated and adjusted. After training is completed, a network data processing model 130' can be obtained. At this point, the parameter values ​​of the network data processing model 130' have been updated, and based on the updated parameter values, the network data processing model 130 can be used to achieve the prediction task in the model application phase.

[0029] During the model training phase, the network data processing model 130 can be trained using a reference dataset 110 comprising multiple reference data 112 and a model training system 150. In this paper, "reference data" is also referred to as "training data," and "reference dataset" is also referred to as "training dataset." Here, each reference data 112 can reflect the performance of a corresponding network scenario. That is, different reference data 112 can correspond to different network categories (also called network scenario categories).

[0030] Specifically, the training process can be performed iteratively using a large amount of reference data. After training is complete, the network data processing model 130 can include knowledge about the network scene categories identified in the network data. During the model application phase, the model application system 152 can be used to invoke the network data processing model 130' (at this time, the prediction model 130' has the trained parameter values). For example, it can receive input data 140 (including network data to be processed or network data to be classified) and output the network scene category 144 corresponding to the input data 140.

[0031] exist Figure 1 In this context, the model training system 150 and the model application system 152 can include any computing system with computing capabilities, such as various computing devices / systems, terminal devices, servers, etc. Terminal devices can involve any type of mobile terminal, fixed terminal, or portable terminal, including mobile phones, desktop computers, laptop computers, notebook computers, netbook computers, tablet computers, media computers, multimedia tablets, or any combination thereof, including accessories and peripherals of these devices or any combination thereof. Servers include, but are not limited to, mainframes, edge computing nodes, computing devices in cloud environments, etc.

[0032] It should be understood that the structure and function of environment 100 are described for illustrative purposes only and do not imply any limitation on the scope of this disclosure. Figure 1The components and arrangements shown in environment 100 are merely examples, and a computing system suitable for implementing the exemplary implementations described in this disclosure may include one or more different components, other components, and / or different arrangements. For example, although shown as separate, model training system 150 and model application system 152 may be integrated in the same system or device. Implementations of this disclosure are not limited in this respect. Exemplary implementations of model training and model application will continue to be described below with reference to the accompanying drawings.

[0033] According to the scheme disclosed herein, in order to better determine the network scene category corresponding to network data, a contrastive model can be used to determine the contrastive features of the network data. In this way, the contrastive features of the same network scene and different network scenes can be learned more effectively, ensuring that the contrastive features of the same network scene are as close as possible in the feature space, and that the contrastive features of different network scenes are as far apart as possible in the feature space. Therefore, it is possible to better distinguish between various network scenes.

[0034] Figure 2 A block diagram 200 illustrating a process for training a contrastive model according to some embodiments of the present disclosure is shown. This process can be implemented at a model training system 150. (The last sentence appears to be incomplete and possibly refers to a separate topic.) Figure 1 describe Figure 3 The comparison model training process.

[0035] exist Figure 2 In this embodiment, the network data processing model 130 includes a contrastive model 220. The contrastive model 220 may employ currently known and / or future-developed contrastive learning network structures, such as any structure employing stacked Transformers or other suitable contrastive learning network structures. Figure 2 The model training system 150 acquires reference data corresponding to each of a set of network scenarios. The reference data reflects the performance of the corresponding network scenario.

[0036] In some embodiments, a set of network scenarios may include packet loss scenarios, jitter scenarios, latency scenarios, buffer throttling scenarios, out-of-order scenarios, etc. Buffer throttling scenarios may include large buffer throttling scenarios and / or small buffer throttling scenarios. In some embodiments, a network scenario may be a combination of any of the above network scenarios or other suitable network scenarios.

[0037] In some embodiments, for each network scenario in a set of network scenarios, a network loss meter can be used to construct or simulate the corresponding network scenario, thereby collecting communication data of the corresponding network scenario as reference data. The network loss meter can be a software network loss meter, a hardware network loss meter, or a combination thereof. The network loss meter can use any currently known and / or future-developed method to simulate network data or network communication data under different network scenarios. The embodiments of this disclosure are not limited thereto. Alternatively, in some embodiments, the reference data can be collected from networks operating in different scenarios.

[0038] As an example, communication data from the network scenario constructed by the network loss meter and / or communication data from the actual network scenario can be collected periodically at predetermined intervals (e.g., 100 milliseconds or other appropriate intervals) as reference data in the reference dataset. Alternatively, reference data can be collected at predetermined time points at non-fixed intervals. This reference data may include characteristic data (also known as indicator data) of its corresponding network scenario. Examples of characteristic data include, but are not limited to: packet loss rate, recovered packet loss rate, transmit bit rate (e.g., actual transmit bit rate), receive bit rate (e.g., actual receive bit rate), jitter data, out-of-order magnitude, network round-trip time (RTT), delay gradient, packet arrival interval (e.g., minimum, average, or maximum packet arrival interval), etc. A collected reference data d i It can be represented as (1):

[0039]

[0040] Among them, ATT i Indicates reference data d i The set of attribute values, S i Indicates reference data d i The corresponding network scenario category, where i represents the reference data d. i Index in the reference dataset. The entire reference dataset can be represented as D: {d0, d1, ..., d...} N}, where N is an integer greater than or equal to 1.

[0041] For a given scenario within a set of network scenarios (e.g., the first scenario), the model training system 150 selects at least one positive sample data 212 associated with the first scenario and at least one negative sample data 214 associated with other scenarios within the set of network scenarios from a reference dataset 110. That is, data belonging to the same network type can be labeled as positive samples, while data not belonging to the same network type can be labeled as negative samples. Assuming the first scenario is a packet loss scenario, the at least one positive sample data 212 can include communication data from the packet loss scenario. The at least one negative sample data 214 can include communication data from other network scenarios (e.g., jitter scenarios, latency scenarios, buffer-limited scenarios, out-of-order scenarios, etc.).

[0042] In some embodiments, positive and negative samples can be selected in batches. Assuming each batch is k (where k is an integer greater than or equal to 1), k / 2 samples can be sampled each time, and for each sample, one more sample can be taken from its corresponding category.

[0043] The model training system 150 generates a contrast model 220 based on at least one positive sample data 212 and at least one negative sample data 214. The contrast model 220 describes the association between contrast features of network scenes. The contrast model 220 makes the similarity between a contrast feature of one network scene and another contrast feature of another network scene indicate whether the network scene and the other network scene belong to the same network scene.

[0044] In some embodiments, a loss function for the contrast model 220 can be generated based on positive sample data 212 and negative sample data 214, and the contrast model 220 can be trained based on the loss function. Specifically, using the contrast model 220, contrast features of the first scene, at least one positive contrast feature of at least one positive sample data 212, and at least one negative contrast feature of at least one negative sample data 214 can be determined respectively. Based on the contrast features, at least one positive contrast feature, and at least one negative contrast feature, the loss function for the contrast model 220 can be generated.

[0045] As an example, the loss function of contrast model 220 can be determined using the following equation (2):

[0046]

[0047] Where l represents the loss, q represents the output of the reference data after being encoded by f, and k + k represents the output after positive sample data is encoded by f. i q represents the output after f encoding, which does not distinguish between positive and negative samples. + Let |q| represent the set of positive samples. +| represents the number of positive samples, and f represents the function of the comparison model 220.

[0048] It should be understood that the loss function shown in Equation (2) is merely exemplary and not restrictive. The loss function in Equation (2) can be adjusted or modified. Other suitable loss functions for contrastive learning can be used to train the contrastive model 220. The embodiments of this disclosure are not limited in this respect.

[0049] In some embodiments, a trained comparison model can be obtained by minimizing the loss function through training. In one example embodiment, a threshold for the loss function can be set. If the loss function value of the comparison model 220 is less than the threshold, training of the comparison model 220 can be terminated. In other example embodiments, the number of training iterations can be set. Training of the comparison model 220 can be terminated after a predetermined number of iterations have been completed.

[0050] In some embodiments, the feature extraction granularity of the comparison model can be adjusted by setting different loss function thresholds or different training iterations. For example, setting a larger loss function or fewer training iterations allows the comparison model 220 to generate comparison features with a larger granularity. Conversely, setting a smaller loss function or more training iterations allows the comparison model 220 to generate comparison features with a smaller granularity. Specifically, for two network scenarios located at the boundary of adjacent network categories, a smaller loss function or more training iterations can enable the comparison model 220 to distinguish the comparison features of these two similar network scenarios.

[0051] In this paper, the trained contrastive model 220 is also referred to as contrastive model 220'. Using contrastive model 220', contrastive features (also called contrastive feature representations or feature representations) of network data can be generated. For example, network data d can be obtained by using the trained contrastive model 220'. i The contrastive feature representation, i.e., f(Att) i Comparative feature representations of data belonging to the same network scene type show a smaller distance in the feature space. A smaller distance indicates greater similarity. Conversely, a larger distance in the feature space indicates less similarity. If the comparative feature representations of data from two network scenes show a large distance in the feature space, then these two network scenes belong to different network scene categories.

[0052] This technical solution utilizes a contrastive model to determine the contrastive features of data from different network scenarios. In this way, it can better learn the contrastive features of the same network scenario category and different network scenario categories, ensuring that the contrastive features of the same network scenario are as close as possible in the feature space, and that the contrastive features of different network scenarios are as far apart as possible in the feature space. Therefore, it can better distinguish between various network scenario categories. Using the exemplary implementation of this disclosure, more discriminative and consistent contrastive features can be provided, thereby ensuring accurate identification of each network scenario category.

[0053] In some embodiments, the network data processing model 130 may further include a classification model for determining the category of a network scene based on the comparative features of the network data determined by the comparison model 220. Figure 3 A block diagram 300 illustrates a process for training a classification model 330 according to some embodiments of the present disclosure. The training process of the classification model 330 can be implemented in... Figure 1 The model training system has 150 locations. It will be combined with... Figure 1 describe Figure 3 The classification model training process.

[0054] In some embodiments, a set of reference contrastive features 322 for a set of network scenes can be determined using a trained contrastive model 220' based on reference data 112 for each of the network scenes. Next, a classification model 330 for network scene classification can be generated based on the set of reference contrastive features 322 for the network scenes. The l-th data used to generate the classification model 330 can be represented as... {f(Att l ),S l}, where l is an integer greater than or equal to 1, f(Att) l S represents the contrastive feature representation of the l-th reference data derived from the trained contrastive model 220'. l Let represent the network scene category of the l-th reference data. The entire dataset used to generate classification model 330 can be represented as Where N is an integer greater than or equal to 1.

[0055] In the context of this disclosure, classification model 330 can be implemented based on various models currently known and / or to be developed in the future. As an example, classification model 330 can be created using any modeling method such as Support Vector Machine (SVM), Deep Neural Network (DNN), etc. The parameter values ​​of classification model 330 can be initialized. The process of generating classification model 330 based on a set of reference contrastive features 322 is also referred to as the training process of classification model 330. That is, the parameter values ​​of classification model 330 are determined through this training process.

[0056] Specifically, using classification model 330, a set of predicted network scenes corresponding to a set of reference contrast features 322 can be determined. Based on the comparison between the network scene category 332 of each set of network scenes and the set of predicted network scenes, the value of the loss function of classification model 330 can be determined. Based on the value of the loss function, classification model 330 can be trained.

[0057] In some embodiments, an appropriate loss function may be selected based on the type or modeling method of the chosen classification model 330. As an example, the loss function may be mean squared error, cross-entropy, etc. It should be understood that any suitable loss function can be used to train the classification model 330. Embodiments of this disclosure are not limited in this respect.

[0058] Through training, the loss function of the classification model 330 can be minimized, thereby obtaining the contrastive feature representation (e.g., the feature vector f(Att)). l The relationship between the network scene type and the network scene type is discussed. The trained classification model 330 is also referred to as classification model 330' in this paper. By using the comparison model 220' and classification model 330', the network scene type corresponding to different network data can be determined.

[0059] Figure 4 A block diagram 400 illustrating a process for processing network data according to some embodiments of the present disclosure is shown. Figure 4 Network data processing can be implemented in Figure 1 The model application system 152 is described above. The network data processing model 130' used in the model application system 152 includes the trained classification model 220' and the trained classification model 330' described above.

[0060] Specifically, for the network data to be processed or classified, such as input data 140, the target contrast feature 422 of the network data can be determined using the trained contrast model 220'. Further, based on the target contrast feature 422, the network scene corresponding to the network data to be processed, such as input data 140, can be determined using the trained classification model 330', for example, network scene category 144.

[0061] In this way, by utilizing a trained contrastive model, the comparative characteristics of network data under different network scenarios can be determined. Then, based on these comparative characteristics, a classification model can be used to determine the network category corresponding to the network data. This contrastive learning-based network scenario classification method is applicable to network services such as real-time communication. This method can classify network types that cannot be covered by manually defined network classification rules. Furthermore, this contrastive learning approach can effectively learn the potential relationships within network data, such as network indicator data, thereby improving network classification accuracy. This disclosure can also effectively classify scenarios where adjacent network type boundaries are difficult to classify manually.

[0062] Figure 5 A flowchart of a method 500 for network data processing according to some embodiments of the present disclosure is shown. Method 500 may, for example, be... Figure 1 The network data processing device 110 or other appropriate device shall perform the operation.

[0063] In box 510, network data processing device 110 acquires reference data corresponding to each of a set of network scenarios. The reference data reflects the performance of the corresponding network scenario.

[0064] In box 520, network data processing device 110 selects, for a first scenario in a set of network scenarios, at least one positive sample data associated with the first scenario and at least one negative sample data associated with other scenarios in the set of network scenarios from reference data.

[0065] In box 530, the network data processing device 110 generates a contrast model based on at least one positive sample data and at least one negative sample data. The contrast model describes the association between contrast features of network scenarios. The contrast model makes the similarity between a contrast feature and another contrast feature of another network scenario indicate whether the network scenario and the other network scenario belong to the same network scenario.

[0066] In some exemplary embodiments, generating a contrast model includes: using the contrast model to determine contrast features of a first scene, at least one positive contrast feature of at least one positive sample data, and at least one negative contrast feature of at least one negative sample data; generating a loss function of the contrast model based on the contrast features, at least one positive contrast feature, and at least one negative contrast feature; and training the contrast model based on the loss function.

[0067] In some exemplary embodiments, method 500 further includes: determining feature data associated with a first scenario, the feature data including at least one of the following: packet loss rate, transmit bit rate, receive bit rate, jitter data, out-of-order magnitude, network round-trip delay, delay gradient, and packet arrival interval; and updating a loss function based on the feature data.

[0068] In some exemplary embodiments, method 500 further includes: determining a set of reference contrast features for a set of network scenes based on reference data for each of the network scenes, using a trained contrast model; and generating a classification model for network scene classification based on the set of reference contrast features for the network scenes.

[0069] In some exemplary embodiments, generating a classification model includes: using the classification model to determine a set of predicted network scenes corresponding to a set of reference contrast features; generating a loss function for the classification model based on a comparison between the set of network scenes and the set of predicted network scenes; and training the classification model based on the loss function of the classification model.

[0070] In some exemplary embodiments, method 500 further includes: determining target contrast features of the network data to be processed using a trained contrast model; and determining the network scene corresponding to the network data to be processed using a trained classification model based on the target contrast features.

[0071] In some exemplary embodiments, a set of network scenarios includes at least one of the following: packet loss scenario, jitter scenario, latency scenario, buffer limiting scenario, and out-of-order scenario.

[0072] In some exemplary embodiments, obtaining reference data corresponding to each of a set of network scenarios includes: constructing a corresponding network scenario using a network loss meter for each network scenario in the set of network scenarios; and collecting communication data of the corresponding network scenario as reference data.

[0073] In some exemplary embodiments, Figure 6 A block diagram of an apparatus 600 for network data processing according to some embodiments of the present disclosure is shown. The apparatus 600 may be implemented as or included in... Figure 1 The network data processing device 110. The various modules / components in the device 600 can be implemented by hardware, software, firmware, or any combination thereof.

[0074] As shown in the figure, device 600 includes an acquisition module 610 configured to acquire reference data corresponding to each of a set of network scenarios. The reference data reflects the performance of the corresponding network scenario. Device 600 also includes a data selection module 620 configured to, for a first scenario in the set of network scenarios, select from the reference data at least one positive sample data associated with the first scenario and at least one negative sample data associated with other scenarios in the set of network scenarios. Device 600 also includes a generation module 630 configured to generate a comparison model based on at least one positive sample data and at least one negative sample data. The comparison model describes the association between comparison features of network scenarios. The comparison model makes the similarity between a comparison feature and another comparison feature of another network scenario indicate whether the network scenario and the other network scenario belong to the same network scenario.

[0075] In some exemplary embodiments, the generation module 630 may be configured to: use a contrast model to determine contrast features of a first scene, at least one positive contrast feature of at least one positive sample data, and at least one negative contrast feature of at least one negative sample data, respectively; generate a loss function of the contrast model based on the contrast features, at least one positive contrast feature, and at least one negative contrast feature; and train the contrast model based on the loss function.

[0076] In some exemplary embodiments, the apparatus 600 further includes: a first determining module configured to determine feature data associated with a first scenario, the feature data including at least one of the following: packet loss rate, transmit code rate, receive code rate, jitter data, out-of-order magnitude, network round-trip delay, delay gradient, and packet arrival interval; and an updating module configured to update a loss function based on the feature data.

[0077] In some exemplary embodiments, the apparatus 600 further includes: a second determining module configured to determine a set of reference contrast features of a set of network scenes based on reference data of each of the network scenes and using a trained contrast model; and a second generating module configured to generate a classification model for network scene classification based on the set of reference contrast features of the network scenes.

[0078] In some exemplary embodiments, the second generation module is configured to: use a classification model to determine a set of predicted network scenes corresponding to a set of reference contrast features; generate a loss function for the classification model based on a comparison between the set of network scenes and the set of predicted network scenes; and train the classification model based on the loss function of the classification model.

[0079] In some exemplary embodiments, the apparatus 600 further includes: a third determining module configured to determine target contrast features of the network data to be processed using a trained contrast model; and a fourth determining module configured to determine the network scene corresponding to the network data to be processed using a trained classification model based on the target contrast features.

[0080] In some exemplary embodiments, a set of network scenarios includes at least one of the following: packet loss scenario, jitter scenario, latency scenario, buffer limiting scenario, and out-of-order scenario.

[0081] In some exemplary embodiments, the acquisition module 610 may be configured to: construct a corresponding network scenario using a network loss meter for each network scenario in a set of network scenarios; and collect communication data of the corresponding network scenario as reference data.

[0082] Figure 7 A block diagram of an electronic device 700 in which one or more embodiments of the present disclosure may be implemented is shown. It should be understood that... Figure 7 The electronic device 700 shown is merely exemplary and should not be construed as limiting the functionality and scope of the embodiments described herein. Figure 7 The electronic device 700 shown can be used to achieve Figure 1 Network data processing equipment 110 or Figure 6 Device 600 for network simulation.

[0083] like Figure 7 As shown, electronic device 700 is in the form of a general-purpose computing device. Components of electronic device 700 may include, but are not limited to, one or more processors or processing units 710, memory 720, storage device 730, one or more communication units 740, one or more input devices 750, and one or more output devices 760. Processing unit 710 may be a physical or virtual processor and is capable of performing various processes according to programs stored in memory 720. In a multiprocessor system, multiple processing units execute computer-executable instructions in parallel to improve the parallel processing capability of electronic device 700.

[0084] Electronic device 700 typically includes multiple computer storage media. Such media can be any available media accessible to electronic device 700, including but not limited to volatile and non-volatile media, removable and non-removable media. Memory 720 can be volatile memory (e.g., registers, cache, random access memory (RAM)), non-volatile memory (e.g., read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory), or some combination thereof. Storage device 730 can be removable or non-removable media and can include machine-readable media, such as flash drives, disks, or any other media capable of storing information and / or data and accessible within electronic device 700.

[0085] Electronic device 700 may further include additional removable / non-removable, volatile / non-volatile storage media. Although not explicitly stated... Figure 7 As shown, disk drives for reading from or writing to removable, non-volatile disks (e.g., "floppy disks") and optical disk drives for reading from or writing to removable, non-volatile optical disks can be provided. In these cases, each drive can be connected to a bus (not shown) via one or more data media interfaces. Memory 720 may include computer program product 725 having one or more program modules configured to perform various methods or actions of various embodiments of this disclosure.

[0086] The communication unit 740 enables communication with other electronic devices via a communication medium. Additionally, the functionality of the components of the electronic device 700 can be implemented using a single computing cluster or multiple computing machines capable of communicating via communication connections. Therefore, the electronic device 700 can operate in a networked environment using logical connections to one or more other servers, network personal computers (PCs), or another network node.

[0087] Input device 750 can be one or more input devices, such as a mouse, keyboard, trackball, etc. Output device 760 can be one or more output devices, such as a monitor, speaker, printer, etc. Electronic device 700 can also communicate with one or more external devices (not shown) via communication unit 740 as needed. These external devices include storage devices, display devices, etc., and can communicate with one or more devices that enable user interaction with electronic device 700, or with any device that enables electronic device 700 to communicate with one or more other electronic devices (e.g., network card, modem, etc.). Such communication can be performed via input / output (I / O) interface (not shown).

[0088] According to an exemplary implementation of this disclosure, a computer-readable storage medium is provided that stores computer-executable instructions thereon, wherein the computer-executable instructions are executed by a processor to implement the methods described above. According to an exemplary implementation of this disclosure, a computer program product is also provided, which is tangibly stored on a non-transitory computer-readable medium and includes computer-executable instructions, which are executed by a processor to implement the methods described above.

[0089] Various aspects of this disclosure are described herein with reference to flowchart illustrations and / or block diagrams of methods, apparatuses, devices, and storage media implemented according to this disclosure. It should be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer-readable program instructions.

[0090] These computer-readable program instructions can be provided to a processing unit of a general-purpose computer, a special-purpose computer, or other programmable data processing apparatus to produce a machine such that, when executed by the processing unit of the computer or other programmable data processing apparatus, they create means for implementing the functions / actions specified in one or more blocks of the flowchart and / or block diagram. These computer-readable program instructions can also be stored in a computer-readable storage medium that causes a computer, programmable data processing apparatus, and / or other device to operate in a particular manner. Thus, the computer-readable medium storing the instructions comprises an article of manufacture that includes instructions for implementing aspects of the functions / actions specified in one or more blocks of the flowchart and / or block diagram.

[0091] Computer-readable program instructions can be loaded onto a computer, other programmable data processing apparatus, or other device to cause a series of operational steps to be performed on the computer, other programmable data processing apparatus, or other device to produce a computer-implemented process, thereby causing the instructions that execute on the computer, other programmable data processing apparatus, or other device to perform the functions / actions specified in one or more boxes of a flowchart and / or block diagram.

[0092] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this disclosure. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of an instruction, which contains one or more executable instructions for implementing the specified logical function. In some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutive blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, may be implemented using a dedicated hardware-based system that performs the specified function or action, or using a combination of dedicated hardware and computer instructions.

[0093] Various implementations of this disclosure have been described above. These descriptions are exemplary and not exhaustive, nor are they limited to the disclosed implementations. Many modifications and variations will be apparent to those skilled in the art without departing from the scope and spirit of the described implementations. The terminology used herein is chosen to best explain the principles, practical applications, or improvements to technology in the market, or to enable others skilled in the art to understand the various implementations disclosed herein.

Claims

1. A method for network data processing, comprising: Obtain reference data corresponding to a set of network scenarios, wherein the reference data reflects the performance of the corresponding network scenarios; For a first scenario in the set of network scenarios, at least one positive sample data associated with the first scenario and at least one negative sample data associated with other scenarios in the set of network scenarios are selected from the reference data; as well as A comparison model is generated based on the at least one positive sample data and the at least one negative sample data. The comparison model describes the association between the network scene and the comparison features of the network scene. The comparison model makes the similarity between the comparison feature and another comparison feature of another network scene indicate whether the network scene and the other network scene belong to the same network scene.

2. The method according to claim 1, wherein generating the comparison model comprises: Using the comparison model, the comparison features of the first scene, at least one positive comparison feature of the at least one positive sample data, and at least one negative comparison feature of the at least one negative sample data are determined respectively. Based on the contrast features, the at least one positive contrast feature, and the at least one negative contrast feature, a loss function for the contrast model is generated; and The comparison model is trained based on the loss function.

3. The method according to claim 2, further comprising: The feature data associated with the first scenario is determined, and the feature data includes at least one of the following: packet loss rate, transmit bit rate, receive bit rate, jitter data, out-of-order magnitude, network round-trip delay, delay gradient, and packet arrival interval; and The loss function is updated based on the feature data.

4. The method according to claim 1, further comprising: Based on the reference data of each of the network scenarios, a set of reference contrast features of the network scenarios are determined using the trained contrast model. as well as Based on the set of reference comparison features of the set of network scenarios, a classification model for network scenario classification is generated.

5. The method of claim 4, wherein generating the classification model comprises: Using the classification model, a set of prediction network scenarios corresponding to the set of reference contrast features are determined respectively; Based on the comparison between the set of network scenarios and the set of predicted network scenarios, a loss function for the classification model is generated; and The classification model is trained based on the loss function of the classification model.

6. The method according to claim 4, further comprising: For the network data to be processed, the target contrast features of the network data to be processed are determined using the trained contrast model. as well as Based on the target contrast features, the network scene corresponding to the network data to be processed is determined using the trained classification model.

7. The method according to claim 1, wherein the set of network scenarios includes at least one of the following: packet loss scenario, jitter scenario, latency scenario, buffer limiting scenario, and out-of-order scenario.

8. The method according to claim 1, wherein obtaining reference data corresponding to each of a set of network scenarios includes: For each of the aforementioned network scenarios, a corresponding network scenario is constructed using a network loss meter; as well as The communication data of the corresponding network scenario is collected as the reference data.

9. A network data processing apparatus, comprising: The acquisition module is configured to acquire reference data corresponding to a set of network scenarios, wherein the reference data reflects the performance of the corresponding network scenario. The data selection module is configured to select, for a first scenario in the set of network scenarios, at least one positive sample data associated with the first scenario and at least one negative sample data associated with other scenarios in the set of network scenarios from the reference data. as well as The generation module is configured to generate a comparison model based on the at least one positive sample data and the at least one negative sample data. The comparison model describes the association between the network scene and the comparison features of the network scene. The comparison model makes the similarity between the comparison features and another comparison feature of another network scene indicate whether the network scene and the other network scene belong to the same network scene.

10. An electronic device, comprising: At least one processing unit; as well as At least one memory, coupled to the at least one processing unit and storing instructions for execution by the at least one processing unit, which, when executed by the at least one processing unit, cause the electronic device to perform the method according to any one of claims 1 to 8.

11. A computer-readable storage medium having a computer program stored thereon, the computer program being executable by a processor to implement the method according to any one of claims 1 to 8.

Images