Method for predicting video playback quality, and electronic device and storage medium

Abstract

Provided are a method for predicting a video playback quality, and an electronic device and a storage medium. The method for predicting a video playback quality comprises: acquiring a video service message of a target video application within the current time period; according to the video service message of the target video application within the current time period, determining traffic characteristic data of the target video application within the current time period; and inputting the traffic characteristic data of the target video application within the current time period into a video playback quality prediction model based on a convolutional neural network, so as to obtain a video playback quality of the target video application within a future time period.

Description

Video playback quality prediction method, electronic device, and storage medium

[0001] This disclosure claims priority to Chinese patent application No. 202310794515.2, filed on June 29, 2023, the entire contents of which are incorporated herein by reference. Technical Field

[0002] The present disclosure relates to the field of computer technology, and in particular to a method for predicting video playback quality, an electronic device, and a storage medium. Background Art

[0003] The rapid development of communications technology has driven the widespread adoption and popularity of online video services. Online video services are gaining increasing prominence in internet applications, and the number of video users is also increasing. To improve user retention and growth, network operators are focusing on predicting video playback quality and ensuring timely intervention. Furthermore, with the internet industry's growing focus on user privacy and security, and the increasing trend toward network traffic encryption, predicting video playback quality is becoming increasingly difficult.

[0004] Summary of the Invention

[0005] In one aspect, embodiments of the present disclosure provide a method for predicting video playback quality. The method comprises: obtaining video service packets of a target video application in a current time period; determining traffic characteristic data of the target video application in the current time period based on the video service packets of the target video application in the current time period; and inputting the traffic characteristic data of the target video application in the current time period into a video playback quality prediction model based on a convolutional neural network to obtain the video playback quality of the target video application in a future time period.

[0006] On the other hand, an embodiment of the present disclosure provides a device for predicting video playback quality. The device for predicting video playback quality includes: a communication module and a processing module. The communication module is used to obtain video service messages of the target video application in the current time period; the processing module is used to determine the traffic characteristic data of the target video application in the current time period based on the video service messages of the target video application in the current time period; the processing module is also used to input the traffic characteristic data of the target video application in the current time period into a video playback quality prediction model based on a convolutional neural network to obtain the video playback quality of the target video application in the future time period.

[0007] In another aspect, embodiments of the present disclosure provide an electronic device comprising: a memory and a processor; the memory and the processor are coupled; the memory is configured to store computer program instructions executable by the processor; and the processor implements the method for predicting video playback quality according to any of the above embodiments when executing the computer program instructions.

[0008] In another aspect, embodiments of the present disclosure provide a computer-readable storage medium having computer program instructions stored thereon, which, when executed on an electronic device, implements the method for predicting video playback quality of any of the above embodiments.

[0009] In another aspect, embodiments of the present disclosure provide a computer program product, which includes computer program instructions, and when executed, implements the method for predicting video playback quality according to any of the above embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] FIG1 is a schematic structural diagram of a system for predicting video playback quality according to some embodiments.

[0011] FIG2 is a flowchart of a method for predicting video playback quality according to some embodiments.

[0012] FIG3 is a flowchart of a method for training a video playback quality prediction model according to some embodiments.

[0013] FIG4 is a schematic structural diagram of a sample according to some embodiments.

[0014] FIG5 is a schematic diagram of the structure of raw data according to some embodiments.

[0015] FIG6 is a schematic structural diagram of a device for predicting video playback quality according to some embodiments.

[0016] FIG7 is a schematic structural diagram of an electronic device according to some embodiments. DETAILED DESCRIPTION

[0017] The following will be combined with the accompanying drawings in the embodiments of the present disclosure to clearly and completely describe the technical solutions in the embodiments of the present disclosure. Obviously, the embodiments described are only part of the embodiments of the present disclosure, not all of the embodiments. Based on the embodiments of the present disclosure, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present disclosure.

[0018] In the description of the present disclosure, unless otherwise specified, " / " means "or", for example, A / B can mean A or B. "And / or" in this article is merely a description of the association relationship of associated objects, indicating that there can be three relationships. For example, A and / or B can mean: only A, only B, and A and B. In addition, "at least one" means one or more, and "a plurality" means two or more. Expressions such as "first" and "second" do not limit the quantity and execution order, and the words "first" and "second" do not necessarily limit them to be different.

[0019] It should be noted that, in this disclosure, expressions such as "exemplarily" or "for example" are used to indicate examples, illustrations, or explanations. Any embodiment or design described in this disclosure as "exemplarily" or "for example" should not be interpreted as being more preferred or more advantageous than other embodiments or designs. Specifically, an embodiment or design using "exemplarily" should not be interpreted as being more preferred or more advantageous than other embodiments or designs. Specifically, the use of expressions such as "exemplarily" or "for example" is intended to present the relevant concepts in a detailed manner.

[0020] On the one hand, the rapid development of communications technology has driven the widespread adoption and popularity of online video services. Online video services are becoming increasingly prominent in internet applications, and the number of video users is also increasing. To improve user retention and growth, network operators are focusing on predicting and promptly intervening in video playback quality. On the other hand, with the internet industry's growing focus on user privacy and security, and the increasing trend toward network traffic encryption, predicting video playback quality is becoming increasingly difficult.

[0021] Currently, some technical methods for predicting video playback quality include prediction methods based on bitrate value extraction and prediction methods based on traditional machine learning methods.

[0022] The prediction method based on bitrate value extraction is as follows: first, by analyzing the application layer payload of the online video service message, try to extract the video bitrate information it carries, then use the bitrate and playback time to infer the video buffer size required for smooth viewing, and compare it with the current amount of data downloaded. If the amount of downloaded data does not have a certain margin relative to the video buffer size, it is predicted that a freeze will occur. The premise of using this method is that the traffic is transmitted in plain text and the application developer writes the correct bitrate information into the message payload. At present, there is a clear trend towards encryption of Internet traffic, and the presence or absence of bitrate characteristics and the method of extracting the bitrate are strongly related to video services. Therefore, this method is not universal.

[0023] Traditional machine learning-based prediction methods analyze statistical information such as packet length, timestamp, and uplink and downlink traffic during a video playback period, transforming it into a set of more advanced, service-specific features (such as average traffic and latency distribution). Machine learning algorithms, such as random forests, are then used to predict the video playback quality at the next moment. This approach can, to a certain extent, address the inability to decode encrypted video traffic payloads. However, it relies heavily on feature engineering and specific services, and the resulting trained models are not very reusable, resulting in low overall efficiency.

[0024] In view of this, the embodiment of the present disclosure provides a method for predicting video playback quality, which includes obtaining video service messages of a target video application in a current time period; determining the traffic characteristic data of the target video application in the current time period based on the video service messages of the target video application in the current time period; inputting the traffic characteristic data of the target video application in the current time period into a video playback quality prediction model based on a convolutional neural network, and obtaining the video playback quality of the target video application in the future time period. On the one hand, the video playback quality prediction model is established by utilizing the idea of ​​a relatively mature convolutional neural network in the image field, thereby improving the prediction efficiency and accuracy. On the other hand, regardless of whether the video service message is encrypted, the video playback quality can be predicted by extracting the traffic characteristic data, which has strong versatility.

[0025] The disclosed embodiments provide a method for predicting video playback quality, which can be applied to a device for predicting video playback quality. For example, the device for predicting video playback quality can be a system architecture for predicting video playback quality. In some embodiments, the device for predicting video playback quality can also implement the functions of the device for predicting video playback quality through a virtual machine (VM) deployed on a physical machine. The device for predicting video playback quality can also be deployed on a terminal device.

[0026] In some embodiments, the terminal device can be a device with wireless transceiver function, and the terminal device can be deployed on land (including indoor or outdoor devices, handheld devices, wearable devices or vehicle-mounted devices). The terminal device can also be deployed on the water surface (such as ships, etc.). The terminal device can also be deployed in the air (for example, on airplanes, balloons and satellites, etc.). The terminal device can be a mobile phone, a tablet computer, a computer with wireless transceiver function, a virtual reality (VR) terminal, an augmented reality (AR) terminal, a wireless terminal in industrial control, a wireless terminal in self-driving, a wireless terminal in remote medical, a wireless terminal in smart grid, a wireless terminal in transportation safety, a wireless terminal in smart city, a wireless terminal in smart home, etc. The embodiments of the present disclosure do not limit the application scenarios. The terminal may sometimes also be referred to as a user, user equipment (UE), access terminal, UE unit, UE station, mobile station, mobile station, remote station, remote terminal, mobile device, UE terminal, wireless communication equipment, UE agent or UE device, etc., but the embodiments of the present disclosure are not limited to this.

[0027] Exemplarily, FIG1 is a schematic diagram of a system architecture for predicting video playback quality according to some embodiments, and the system architecture for predicting video playback quality includes a data acquisition module 11 , a data processing module 12 , a model training module 13 and an online operation module 14 .

[0028] The data collection module 11 is used to collect video service messages generated by the target video application to support the subsequent training process. When the terminal where the target video application is installed is in a preset scenario, the video service messages of the target video application in the historical time period are obtained.

[0029] The data processing module 12 is used to process the video service messages of the target video application obtained by the data acquisition module 11 within the historical time period, and determine the traffic characteristic data of the target video application in the historical time period; the traffic characteristic data of the target video application in the historical time period is used to form a sample corresponding to the preset scenario, and the video playback quality of the target video application in the next time period after the historical time period is used as a label to obtain a training set.

[0030] The model training module 13 is used to train the initial model based on the training set output by the data processing module 12 to obtain a video playback quality prediction model. The video playback quality prediction model can predict the video playback quality of the target video application in the future time period based on the traffic characteristic data of the target video application in the current time period.

[0031] The online operation module 14 is configured to load the video playback quality prediction model output by the model training module 13 onto the terminal device. The module then determines the traffic characteristic data of the target video application in the current time period based on the video service packets generated by the target video application in the current time period. Finally, the traffic characteristic data of the target video application in the current time period is input into the video playback quality prediction model based on a convolutional neural network to determine the video playback quality of the target video application in the future time period.

[0032] The following describes in detail the method for predicting video playback quality provided by the embodiments of the present disclosure in conjunction with the accompanying drawings.

[0033] The present disclosure provides a method for predicting video playback quality. As shown in FIG2 , the method includes the following steps.

[0034] S101: Obtain video service messages of a target video application in a current time period.

[0035] In some embodiments, the duration of the current time period is a preset duration, for example, the duration of the current time period is 15 seconds.

[0036] In some embodiments, the preset duration is determined based on a service requirement of the target video application, which may include at least one of the following: a delay requirement, a bandwidth requirement, and a jitter requirement.

[0037] Latency refers to the transmission delay from the data transmitter to the data receiver. Video applications typically have a maximum latency tolerance (i.e., the maximum permissible delay in data transmission). Based on this latency requirement, a preset duration can be selected to keep data transmission latency within the required range and ensure that the video application's performance meets the requirements.

[0038] Bandwidth requirements refer to the data transmission rate supported by the system or network. The bandwidth requirements of a video application may be affected by factors such as video resolution, video encoding method, and frame rate. Based on bandwidth requirements, you can estimate the bandwidth consumption of the video application within a specific time period and select a preset duration to ensure that the network provides sufficient bandwidth to meet the video application's requirements.

[0039] Jitter requirements refer to the latency fluctuations in data transmission. Video applications have a certain tolerance for latency fluctuations. Based on the jitter requirements, you can select a preset duration to keep data transmission jitter within the required range and ensure the video application playback experience meets the requirements.

[0040] It is understandable that, depending on the business requirements of the target video application, a suitable preset duration is selected to ensure that the performance of the target video application meets the business requirements.

[0041] In some embodiments, the video service message includes an encrypted video service message and / or an unencrypted video service message. Thus, identifying and analyzing these messages can help monitor and improve the security of video transmission and enhance user trust and experience.

[0042] In some embodiments, packet capture software is used to capture network packets passing through a terminal device; based on the characteristics and protocol of a target video application, deep packet inspection (DPI) and analysis are performed on the captured network packets to obtain video service packets of the target video application in the current time period.

[0043] Packet capture software is a tool used to capture and analyze network data packets. It monitors network interfaces, captures packets transmitted through them, and provides analysis and display capabilities to help users gain a deeper understanding of network communications, network protocols, and data flows. Common packet capture software includes Wireshark, tcpdump, Microsoft Message Analyzer, and Fiddler.

[0044] DPI is a network traffic analysis technology that analyzes and parses network data packets to identify packets corresponding to different applications (such as video service packets).

[0045] S102: Determine traffic characteristic data of the target video application in the current time period based on video service messages of the target video application in the current time period.

[0046] Traffic characteristics include uplink traffic, downlink traffic, uplink packet count, and downlink packet count. Uplink traffic represents traffic sent from a terminal device to the network, while downlink traffic represents traffic sent from the network to a terminal device. The uplink packet count and downlink packet count indicate the number of data packets in the corresponding traffic.

[0047] As an implementation method, the video service messages of the target video application in the current time period are equally divided and feature calculated on the timeline from several dimensions included in the traffic feature data to obtain the uplink traffic, downlink traffic, number of uplink packets and number of downlink packets of the target video application at different time granularities in the current time period.

[0048] In some embodiments, the time granularity of the segmentation is a preset value, which is determined based on the service requirements of the target video application. The service requirements include at least one of the following: latency requirement, bandwidth requirement, and jitter requirement. For example, the time granularity of the segmentation is 0.1 second.

[0049] For example, assuming the current time period is divided into five time granularities, each in minutes, the following example shows how the target video is applied to the upstream traffic, downstream traffic, upstream packet count, and downstream packet count at each time granularity within the current time period.

[0050] Assume that the current time period is 2023-06-28 09:00:00 to 2023-06-28 10:00:00;

[0051] Uplink traffic (bytes): [1000, 2000, 1500, 1200, 1800];

[0052] Downlink traffic (bytes): [3000, 4000, 3500, 3200, 3800];

[0053] Number of uplink packets: [10, 15, 12, 8, 13];

[0054] Number of downlink packets: [20, 25, 22, 18, 23].

[0055] The current time period is divided equally into the following: time granularity 1 is 09:00-09:12; time granularity 2 is 09:12-09:24; time granularity 3 is 09:24-09:36; time granularity 4 is 09:36-09:48; and time granularity 5 is 09:48-10:00.

[0056] After feature calculation, the video service message in time granularity 1 has the following characteristics: upstream traffic: 1000 bytes, downstream traffic: 3000 bytes, upstream packet number: 10 packets (data packets), downstream packet number: 20 packets,

[0057] In time granularity 2, the upstream traffic is 2000 bytes, the downstream traffic is 4000 bytes, the number of upstream packets is 15 packets, and the number of downstream packets is 25 packets.

[0058] In time granularity 3, the upstream traffic is 1500 bytes, the downstream traffic is 3500 bytes, the number of upstream packets is 12 packets, and the number of downstream packets is 22 packets.

[0059] In time granularity 4, the upstream traffic is 1200 bytes, the downstream traffic is 3200 bytes, the number of upstream packets is 8 packets, and the number of downstream packets is 18 packets.

[0060] In time granularity 5, the upstream traffic is 1800 bytes, the downstream traffic is 3800 bytes, the number of upstream packets is 13 packets, and the number of downstream packets is 23 packets.

[0061] S103: Input the traffic feature data of the target video application in the current time period into a video playback quality prediction model based on a convolutional neural network to obtain the video playback quality of the target video application in the future time period.

[0062] A convolutional neural network (CNN) is a deep learning model widely used in fields such as image recognition, computer vision, and speech recognition. A CNN is a specialized artificial neural network with basic components such as convolutional layers, pooling layers, and fully connected layers. The core concept of a CNN is to extract features from input traffic data through convolution operations and then abstract and classify high-level features through a stacked neural network structure.

[0063] Based on this, the target video application's traffic characteristics for the current time period are determined based on the video service packets within the current time period. This traffic characteristic data is then fed into a convolutional neural network-based video playback quality prediction model to predict video playback quality for future time periods. This model leverages the well-established convolutional neural network approach in the image processing field to improve prediction efficiency and accuracy. Furthermore, the model can be used to predict video playback quality by extracting traffic characteristic data, regardless of whether the video service packets are encrypted, demonstrating its high versatility.

[0064] In some embodiments, the present disclosure provides a method for training a video playback quality prediction model. As shown in FIG3 , the training method includes the following steps.

[0065] S201. Obtain a training set.

[0066] The training set includes multiple labeled samples. These samples contain traffic feature data for the target video application within a historical time period. The labels indicate the video playback quality of the target video application in the next time period after the historical time period. Traffic feature data includes uplink traffic, downlink traffic, uplink packet count, and downlink packet count.

[0067] In some embodiments, the duration of the historical time period is a preset duration.

[0068] In some embodiments, the preset duration is determined based on a service requirement of the target video application, which may include at least one of the following: a delay requirement, a bandwidth requirement, and a jitter requirement.

[0069] In some embodiments, the labels of samples in the training set include excellent, good, and poor.

[0070] In some embodiments, the video playback quality of the target video application in the next time period after the historical time period is evaluated based on the video quality evaluation index to obtain an evaluation score for the video playback quality of the target video application in the next time period after the historical time period; and a label for the sample is determined based on the evaluation score. The video quality evaluation index includes at least one of video resolution, frame rate, compression ratio, and loading speed.

[0071] In some embodiments, when the evaluation score is greater than a first preset score, the label of the sample is determined to be excellent; when the evaluation score is less than the first preset score and greater than a second preset score, the label of the sample is determined to be good; when the evaluation score is less than the second preset score, the label of the sample is determined to be poor.

[0072] For example, FIG4 provides a structural diagram of a sample. As shown in FIG4 , the sample is labeled as excellent. The sample includes the uplink traffic, downlink traffic, number of uplink packets, and number of downlink packets of the target video application at different time granularities within the historical time period. For example, the uplink traffic of the target video application at time granularity n in the historical time period is V1n; the downlink traffic of the target video application at time granularity n in the historical time period is V2n; the number of uplink packets of the target video application at time granularity n in the historical time period is V3n; and the number of downlink packets of the target video application at time granularity n in the historical time period is V4n.

[0073] In some embodiments, when the terminal on which the target video application is installed is in a preset scenario, video service messages of the target video application within a historical time period are obtained; based on the video service messages of the target video application within the historical time period, traffic characteristic data of the target video application within the historical time period is determined; and the traffic characteristic data of the target video application within the historical time period is used to form a sample corresponding to the preset scenario.

[0074] In some embodiments, the preset scenarios include at least one of the following: the terminal is in a high-speed moving area; the terminal is in a non-high-speed moving area; the terminal accesses a base station in a high-load state; the terminal accesses a base station in a non-high-load state; the terminal is in a weak signal coverage area; the terminal is in a strong signal coverage area.

[0075] In some embodiments, the high-speed moving area is an area where the moving speed is greater than a first preset speed; and the non-high-speed moving area is an area where the moving speed is less than a second preset speed.

[0076] In some embodiments, a base station in a high-load state is a base station whose load is greater than a first preset threshold; a base station in a non-high-load state is a base station whose load is less than a second preset threshold.

[0077] In some embodiments, the weak signal coverage area is an area where the signal strength is less than a third preset threshold; the strong signal coverage area is an area where the signal strength is greater than a fourth preset threshold.

[0078] In some embodiments, there are at least two samples in the training set corresponding to different preset scenarios.

[0079] For example, when a terminal with a target video application installed is in a preset scenario, tcpdump software is used to capture network packets passing through the target device to obtain a packet capture file for the preset scenario. Then, using manual or automated methods, the video playback quality at each time point in the historical time period is recorded to obtain a labeled file. Every 5 seconds is recorded as a time point. The packet capture file and the labeled file are used as one set of raw data. The terminal is then placed in other preset scenarios and the above steps are repeated to obtain multiple sets of raw data. Figure 5 provides a schematic diagram of the structure of the raw data.

[0080] Then, for each time point in each piece of original data, the video playback quality corresponding to the time point is used as the label of each sample. The network message of the preset time before the time point is obtained, and DPI detection and analysis are performed on it to obtain the video service message of the preset time before the time point; the video service message of the preset time before the time point is divided into equal parts on the timeline and feature calculation is performed according to the time granularity of 0.1 seconds to obtain the uplink traffic, downlink traffic, number of uplink packets and number of downlink packets of the video service message of the preset time before the time point at different time granularities; the uplink traffic, downlink traffic, number of uplink packets and number of downlink packets of the video service message of the preset time before the time point at different time granularities are used as the sample corresponding to the time point.

[0081] S202: Based on the training set, the initial model is trained to obtain a video playback quality prediction model.

[0082] In some embodiments, the model parameters of the initial model are determined based on the service type of the video played by the target video application and the sample structure included in the training set; and the initial model is determined based on the model parameters of the initial model. The model parameters of the initial model include the depth of the network and the size of the convolution kernel.

[0083] Exemplarily, a training set is obtained. Each of the multiple labeled samples in the training set is used as input to an untrained convolutional neural network. The video playback quality corresponding to the input sample is used as the output of the convolutional neural network. The convolutional neural network is trained using a machine learning method (such as a self-supervised learning method) to obtain a video playback quality prediction model.

[0084] During the training process, samples can be input into the convolutional neural network one by one to obtain the video playback quality output by the convolutional neural network. Then, the loss value can be determined based on the output video playback quality and the label value of the sample label corresponding to the input sample. The loss value can be used to characterize the difference between the video playback quality of the output target video application in the next time period after the historical time period and the video playback quality of the actual target video application in the next time period after the historical time period. The larger the loss value, the greater the difference. The loss value can be determined based on Euclidean distance, etc. Afterwards, the parameters of the convolutional neural network can be updated using the loss value. Thus, each time a sample is input, the parameters of the convolutional neural network can be updated based on the loss value.

[0085] In some embodiments, the video playback quality prediction model can determine whether training is completed in a variety of ways. As an example, when the error between the video playback quality of the target video application output by the video playback quality prediction model in the next time period after the historical time period and the video playback quality of the actual target video application in the next time period after the historical time period is less than a preset value (for example, 0.01), it can be determined that the training is complete. As another example, if the number of training times of the convolutional neural network is equal to the preset number of times, it can be determined that the training is complete. Here, when it is determined that the training is complete, the trained convolutional neural network can be determined as the video playback quality prediction model.

[0086] Based on this, network traffic is represented as a one-dimensional image. Using convolutional neural networks, a well-established approach in the image field, the initial model is trained to generate a video playback quality prediction model. Leveraging the self-learning capabilities of deep learning networks, the model automatically finds appropriate high-level features during training, eliminating the complex feature engineering steps required by traditional machine learning methods. Furthermore, the excellent transfer learning properties of convolutional neural networks can be leveraged to assist in learning playback quality models for other video applications, using an already learned model for a particular video application as a model. This accelerates the generation and deployment of new models and meets the timeliness requirements of new services.

[0087] The above mainly introduces the scheme of the embodiment of the present disclosure from the perspective of method. A device for predicting video playback quality is also shown below, which is used to execute the method for predicting video playback quality in any of the above embodiments and possible implementations thereof. It can be understood that the device for predicting video playback quality includes hardware structures and / or software modules corresponding to the execution of various functions in order to implement the method for predicting video playback quality; those skilled in the art should easily realize that, in combination with the algorithm steps of the various examples described in the embodiments of the present disclosure, the present disclosure can be implemented in the form of hardware or a combination of hardware and computer software. Whether a function is executed in the form of hardware or computer software driving hardware depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of the present disclosure.

[0088] The embodiment of the present disclosure can divide the functional modules of the device for predicting video playback quality according to the above-mentioned method embodiment. For example, each functional module can be divided corresponding to each function, or two or more functions can be integrated into one functional module. The above-mentioned integrated module can be implemented in the form of hardware or software. It should be noted that the division of modules in the embodiment of the present disclosure is schematic and is only a logical function division. There may be other division methods in actual implementation. The following is an example of dividing each functional module corresponding to each function.

[0089] FIG6 shows a device for predicting video playback quality according to some embodiments. The device 30 for predicting video playback quality includes: a communication module 31 , a processing module 32 , and a model training module 33 .

[0090] The communication module 31 is configured to obtain video service messages of a target video application in a current time period.

[0091] The processing module 32 determines traffic characteristic data of the target video application in the current time period according to the video service messages of the target video application in the current time period.

[0092] The processing module 32 is further configured to input the traffic characteristic data of the target video application in the current time period into a video playback quality prediction model based on a convolutional neural network to obtain the video playback quality of the target video application in the future time period.

[0093] In some embodiments, the traffic characteristic data includes uplink traffic, downlink traffic, the number of uplink packets, and the number of downlink packets.

[0094] In some embodiments, the model training module 33 is used to obtain a training set, which includes multiple samples carrying labels, the samples include traffic feature data of the target video application in a historical time period, and the labels are used to indicate the video playback quality of the target video application in the next time period after the historical time period.

[0095] The model training module 33 is further used to train the initial model based on the training set to obtain a video playback quality prediction model.

[0096] In some embodiments, the communication module 31 is configured to obtain video service packets of the target video application during a historical time period when the terminal on which the target video application is installed is in a preset scenario. The processing module 32 is configured to determine traffic characteristic data of the target video application during the historical time period based on the video service packets of the target video application during the historical time period; and to construct a sample corresponding to the preset scenario using the traffic characteristic data of the target video application during the historical time period.

[0097] In some embodiments, the preset scenario includes at least one of the following:

[0098] The terminal is in a high-speed moving area;

[0099] The terminal is in a non-high-speed mobility area;

[0100] The terminal accesses a base station in a high-load state;

[0101] The terminal accesses a base station that is not in a high-load state;

[0102] The terminal is in a weak signal coverage area;

[0103] The terminal is in an area with strong signal coverage.

[0104] In some embodiments, there are at least two samples in the training set corresponding to different preset scenarios.

[0105] In some embodiments, the duration of the previous time period and the duration of the historical time period are both preset durations, which are determined based on the business requirements of the target video application. The business requirements include at least one of the following: delay requirements, bandwidth requirements, and jitter requirements.

[0106] In some embodiments, the video service packet includes an encrypted video service packet and / or an unencrypted video service packet.

[0107] In the case of implementing the functions of the above-mentioned integrated modules in hardware, the embodiments of the present disclosure also provide a structure of an electronic device for executing the method for predicting video playback quality provided by the embodiments of the present disclosure. As shown in Figure 7, the electronic device 400 includes: a communication interface 403, a processor 402, and a bus 404. In some embodiments, the electronic device 400 may also include a memory 401.

[0108] Processor 402 may implement or execute the various exemplary logic blocks, modules, and circuits described in conjunction with the embodiments of the present disclosure. Processor 402 may be a central processing unit, a general-purpose processor, a digital signal processor, an application-specific integrated circuit, a field-programmable gate array, or other programmable logic device, a transistor logic device, a hardware component, or any combination thereof. Processor 402 may implement or execute the various exemplary logic blocks, modules, and circuits described in conjunction with the embodiments of the present disclosure. Processor 402 may also be a combination that implements computing functions, such as a combination of one or more microprocessors, a combination of a DSP and a microprocessor, and the like.

[0109] The communication interface 403 is used to connect to other devices via a communication network, such as Ethernet, wireless access network, wireless local area network (WLAN), etc.

[0110] The memory 401 may be a read-only memory (ROM) or other type of static storage device that can store static information and instructions, a random access memory (RAM) or other type of dynamic storage device that can store information and instructions, or an electrically erasable programmable read-only memory (EEPROM), a disk storage medium or other magnetic storage device, or any other medium that can be used to carry or store desired program code in the form of instructions or data structures and can be accessed by a computer, but is not limited thereto.

[0111] As an implementation, memory 401 may exist independently of processor 402. Memory 401 may be connected to processor 402 via bus 404 to store instructions or program codes. When processor 402 calls and executes the instructions or program codes stored in memory 401, the method for predicting video playback quality provided in the embodiment of the present disclosure can be implemented.

[0112] In another implementation, the memory 401 may also be integrated with the processor 402 .

[0113] Bus 404 can be an Extended Industry Standard Architecture (EISA) bus, etc. Bus 404 can be divided into an address bus, a data bus, a control bus, etc. For ease of illustration, FIG7 shows only one thick line, but this does not mean that there is only one bus or only one type of bus.

[0114] Some embodiments of the present disclosure provide a computer-readable storage medium (e.g., a non-transitory computer-readable storage medium), which stores computer program instructions. When the computer program instructions are executed on a computer, the computer executes the method for predicting video playback quality as described in any of the above embodiments.

[0115] In one embodiment, the computer may be the aforementioned video playback quality prediction device, and the present disclosure does not limit the form of the computer.

[0116] In some examples, the computer-readable storage media described above may include, but are not limited to, magnetic storage devices (e.g., hard disks, floppy disks, or magnetic tapes), optical disks (e.g., compact disks (CDs), digital versatile disks (DVDs), etc.), smart cards, and flash memory devices (e.g., erasable programmable read-only memories (EPROMs), cards, sticks, or key drives, etc.). The various computer-readable storage media described in this disclosure may represent one or more devices and / or other machine-readable storage media for storing information. The term "machine-readable storage medium" may include, but is not limited to, wireless channels and various other media capable of storing, containing, and / or carrying instructions and / or data.

[0117] An embodiment of the present disclosure provides a computer program product comprising instructions. When the computer program product is run on a computer, the computer is enabled to execute the method for predicting video playback quality described in any one of the above embodiments.

[0118] The above is only a specific embodiment of the present disclosure, but the scope of protection of the present disclosure is not limited thereto. Any changes or replacements within the technical scope disclosed in the present disclosure should be included in the scope of protection of the present disclosure. Therefore, the scope of protection of the present disclosure should be based on the scope of protection of the claims.

Claims

1. A method for predicting video playback quality, comprising: Obtain the video service packets of the target video application in the current time period; Determining traffic characteristic data of the target video application in the current time period according to the video service messages of the target video application in the current time period; The traffic feature data of the target video application in the current time period is input into a video playback quality prediction model based on a convolutional neural network to obtain the video playback quality of the target video application in the future time period.

2. The method according to claim 1, wherein: The traffic characteristic data includes uplink traffic, downlink traffic, the number of uplink packets and the number of downlink packets.

3. The method according to claim 1, wherein: The video playback quality prediction model is trained in the following way: Acquire a training set, wherein the training set includes a plurality of samples carrying labels, wherein the samples include traffic feature data of the target video application in a historical time period, and the labels are used to indicate the video playback quality of the target video application in a next time period after the historical time period; Based on the training set, the initial model is trained to obtain the video playback quality prediction model.

4. The method according to claim 3, further comprising: When the terminal on which the target video application is installed is in a preset scenario, obtaining a video service message of the target video application in the historical time period; Determining traffic characteristic data of the target video application in the historical time period according to video service messages of the target video application in the historical time period; The traffic feature data of the target video application in the historical time period constitutes a sample corresponding to the preset scenario.

5. The method according to claim 4, wherein: The preset scenario includes at least one of the following: The terminal is in a high-speed moving area; The terminal is in a non-high-speed mobile area; The terminal accesses a base station in a high-load state; The terminal accesses a base station which is in a non-high load state; The terminal is in a weak signal coverage area; The terminal is in a strong signal coverage area.

6. The method according to claim 4 or 5, wherein: There are at least two samples in the training set corresponding to different preset scenarios.

7. The method according to claim 3, wherein: The duration of the current time period and the duration of the historical time period are both preset durations, and the preset duration is determined according to the business requirements of the target video application, and the business requirements include at least one of the following: delay requirements, bandwidth requirements, and jitter requirements.

8. The method according to claim 1, wherein: The video service message includes an encrypted video service message and / or an unencrypted video service message.

9. An electronic device, comprising: Memory and processor; Memory and processor coupling; The memory is used to store instructions executable by the processor; When the processor executes the instructions, the method according to any one of claims 1 to 8 is performed.

10. A computer-readable storage medium, wherein: The computer-readable storage medium stores computer instructions, and when the computer instructions are executed on an electronic device, the electronic device executes the method according to any one of claims 1 to 8.

Images