An Internet-based cloud classroom live transmission method and system

Abstract

The present invention provides an Internet-based cloud classroom live transmission method and system, which relates to the field of live broadcast technology. The method includes: obtaining user parameters and live event parameters, and allocating first broadband resources for the live event according to the content type, platform network condition and number of participants of the live event; based on a transmission strategy set, transmitting first live data from a first device set to a second device set, wherein the transmission strategy set includes a plurality of different transmission strategies, the first device set includes one or more first devices, and the second device set includes one or more second devices; dynamically adjusting each transmission strategy according to the user's network condition and data transmission quality to improve the transmission efficiency, stability and user experience of the cloud classroom live broadcast, and at the same time being able to dynamically adjust according to the changes in the real-time network condition and user requirements to ensure high-quality live broadcast services can be provided in various complex environments.

Description

Technical Field

[0001] This application relates to the technical field of webcasting, and in particular to an Internet-based cloud classroom live transmission method and system. Background Art

[0002] With the booming development of the online education industry, cloud classroom live broadcast has become one of the important forms of educational services. However, when facing complex and changeable network environments, large-scale user participation, and diverse user needs, the current cloud classroom live transmission methods have exposed a series of problems:

[0003] Existing live transmission methods often cannot effectively adapt to the network conditions of different users. In an environment with unstable or low bandwidth, live videos are prone to stuttering, latency, or even interruption, seriously affecting the user experience; the resource allocation of live activities is usually based on static rules or simple predictions, lacking real-time and dynamic features. During peak live periods, uneven resource allocation may cause some users to be unable to watch smoothly, while some resources are not fully utilized.

[0004] Traditional methods adopt fixed transmission strategies and cannot be flexibly adjusted according to changes in the type of live content, user participation, and network conditions. For example, during periods with dense video content or frequent user interactions, the bandwidth is not increased or the transmission protocol is not optimized accordingly, resulting in low data transmission efficiency. Existing live systems often cannot dynamically adjust the media quality (such as video resolution, frame rate, audio bit rate, etc.) according to the actual network conditions, device performance, and viewing needs of users. This leads to the situation that users may not be able to enjoy higher video quality in a high-bandwidth environment, while in a low-bandwidth environment, the learning effect may be reduced due to blurred video quality. In the case of multiple communication links available, existing methods often lack intelligent link selection and load balancing mechanisms. This may lead to an unreasonable data transmission path, increasing the transmission latency and packet loss rate, and at the same time, the potential of network resources is not fully utilized. Traditional caching strategies are often based on fixed cache sizes and policies and cannot be dynamically adjusted according to the user's geographical location, network conditions, and real-time traffic changes. This may result in outdated, redundant, or insufficient cached data, affecting the efficiency of data transmission and the user experience. Summary of the Invention

[0005] In view of the above problems, the present invention proposes an Internet-based cloud classroom live transmission method and system; by introducing an intelligent broadband resource allocation algorithm, a flexible set of transmission strategies, a dynamic media quality adjustment mechanism, intelligent link selection and load balancing technologies, and a multi-level caching mechanism, it aims to improve the transmission efficiency, stability, and user experience of cloud classroom live broadcasts, and at the same time can be dynamically adjusted according to changes in real-time network conditions and user needs to ensure high-quality live broadcast services in various complex environments.

[0006] The object of the present application is achieved by the following technical solutions:

[0007] In a first aspect, the present application provides an Internet-based cloud classroom live transmission method, and the method includes:

[0008] Obtain user parameters and live event parameters, where the user parameters include the user's network condition and data transmission quality, and the live event parameters include the content type of the live event, the platform network condition, the number of participants, and the first live data;

[0009] Allocate the first broadband resource of the live event according to the content type of the live event, the platform network condition, and the number of participants;

[0010] Based on a set of transmission policies, transmit the first live data from a first set of devices to a second set of devices, where the set of transmission policies includes a plurality of different transmission policies, the first set of devices includes one or more first devices, and the second set of devices includes one or more second devices;

[0011] Dynamically adjust each transmission policy according to the user's network condition and data transmission quality, where the transmission policies include intelligently selecting a data transmission link, dynamically adjusting media quality parameters, and dynamically adjusting the buffer size and the time interval between two adjacent adjustments.

[0012] Preferably, the step of allocating the first broadband resource of the live event according to the content type of the live event, the platform network condition, and the number of participants; includes:

[0013] Establish a first broadband resource prediction model through historical data; the historical data includes the content type of the live event, the peak number of people, the average number of online people, the platform network condition, the type and total amount of live events in the same time period, and the total number of participants;

[0014] Predict the first broadband resource of the live event corresponding to the t + 1 moment through the broadband resource prediction model according to the real-time data at the t moment.

[0015] Preferably, the set of transmission policies includes:

[0016] Perform real-time scoring on multiple types of first live data; obtain a transmission priority according to the real-time scoring; set a transmission queue according to the transmission priority; the types of the first live data include file data, video data, audio data, and / or writing data;

[0017] In response to a change in the file data, the first device only transmits the changed part of the file data to the second electronic device;

[0018] Set the transmission frequency of the first live data type of the corresponding type according to the refresh frequency of video data, audio data or interactive data;

[0019] Conduct a comprehensive assessment of the network environment and intelligently select a data transmission link; Dynamically adjust the media quality parameters according to the network conditions, user information and environmental information;

[0020] Establish a multi-level caching mechanism, and dynamically adjust the buffer size and the time interval between two adjacent adjustments according to the user's geographical location, network conditions and node scores.

[0021] Preferably, the conducting a comprehensive assessment of the network environment and intelligently selecting a data transmission link includes:

[0022] According to the user's geographical location, obtain multiple communication links between the sender and the receiver through the network topology structure;

[0023] Predict the available links at time t + 1 through constraint conditions based on the first link score, geographical distance and link load at time t.

[0024] Preferably, the predicting the available links at time t + 1 through constraint conditions based on the first link score, geographical distance and link load at time t includes:

[0025] Collect the real-time performance metrics of the available links, where the performance metrics include packet loss rate, packet loss residual, change rate of packet loss rate residual, real-time bandwidth, delay time and link reliability;

[0026] Output the first link score at time t through a machine learning model;

[0027] Use time series analysis to predict the link performance metrics at time t + 1, and combine the first link score at time t to predict the first score at time t + 1;

[0028] Calculate the second link score at time t + 1 through the prediction of the geographical distance and load of the link;

[0029] Define constraint conditions, where the constraint conditions include the maximum acceptable packet loss rate, minimum bandwidth requirement, maximum delay and load balancing;

[0030] Predict the available links at time t + 1 according to the second link score and constraint conditions of the link at time t + 1;

[0031] Select multiple available links according to the redundancy requirements of the service; And select the available link with the highest second score as the main link; Allocate the data in the transmission queue to different available links according to the data volume in the transmission queue and the link load.

[0032] Preferably, the method further includes:

[0033] Establish a closed-loop feedback mechanism to continuously monitor the link performance and adjust the parameters of the scoring model and prediction algorithm according to the actual performance;

[0034] According to historical data and newly emerging network conditions, adaptively adjust the constraint conditions and scoring weights; introduce an anomaly detection algorithm to identify and isolate abnormal behaviors and / or faulty links in the network.

[0035] Preferably, allocate the data in the transmission queue to different available links based on the data volume in the transmission queue and the link load; including:

[0036] Divide the data with the same priority into different data blocks; determine the boundaries of the data blocks through a hash function; ensure that data blocks with the same content have the same hash value;

[0037] Allocate the data block with the highest priority to the main link until a preset load threshold is reached;

[0038] Allocate the remaining data blocks in sequence according to the second score of the remaining selected links and the load of the corresponding links.

[0039] Preferably, dynamically adjust the media quality parameters according to the network conditions, user information, and environmental information, including:

[0040] Collect the first score of the user-side transmission link, user information, live broadcast time, environmental information, and media quality parameters; the user information includes user device information, device battery life and power, geographical location, and preferences; the media quality parameters include audio quality, media clarity, and interaction response speed;

[0041] Train a deep learning model through historical training data to learn the relationship between media quality and user satisfaction under different conditions;

[0042] According to the real-time network conditions, user information, live broadcast time, and environmental information, use the deep learning model to predict the optimal media quality parameters.

[0043] Preferably, establish a multi-level caching mechanism and dynamically adjust the buffer size according to the user's geographical location, network conditions, and node scores, including:

[0044] Construct a distributed caching network; the distributed caching network includes multiple nodes; and obtain the node scores;

[0045] Adjust the buffer size and the time interval between two adjacent buffer size adjustments according to the node scores.

[0046] This application provides an Internet-based live cloud classroom transmission system, and the system includes:

[0047] A first acquisition module, configured to acquire user parameters and live event parameters, where the user parameters include the user's network condition and data transmission quality, and the live event parameters include the content type of the live event, the platform network condition, the number of participants, and first live data;

[0048] A first allocation module, which allocates first broadband resources for the live event according to the content type of the live event, the platform network condition, and the number of participants;

[0049] A transmission module, configured to transmit first live data from a first device set to a second device set based on a transmission policy set, where the transmission policy set includes a plurality of different transmission policies, the first device set includes one or more first devices, and the second device set includes one or more second devices;

[0050] An adjustment module, configured to dynamically adjust each transmission policy according to the user's network condition and data transmission quality, where the transmission policies include intelligent selection of data transmission links, dynamic adjustment of media quality parameters, dynamic adjustment of buffer sizes, and the time interval between two adjacent adjustments.

[0051] The beneficial effects of the present invention include: The first broadband resource prediction model predicts the broadband resources required for a live event by analyzing historical data, ensuring that the resource allocation is neither excessive nor insufficient, and improving resource utilization efficiency. The real-time scoring and priority setting ensure the timely transmission of different types of data, especially the accurate transmission of file data changes, reducing unnecessary data transmission volume and improving transmission efficiency. The intelligent link selection and media quality parameter adjustment can be dynamically adjusted according to the network condition and user needs, ensuring the stability and high quality of the transmission, while reducing latency and stuttering. The combination of the multi-level caching mechanism and the dynamic adjustment of buffer sizes ensures that even when the network condition is poor, users can obtain a smooth viewing experience and reduce the buffer waiting time. The deep learning model predicts media quality parameters and can provide personalized content quality settings according to information such as user device conditions, geographical locations, and preferences, enhancing user satisfaction and participation. The closed-loop feedback mechanism and the anomaly detection algorithm can timely detect and isolate abnormal behaviors or faulty links in the network, reducing the possibility of live interruption and improving the stability and reliability of the system. The dynamic adjustment mechanism enables the system to adapt to the changing network environment and user needs, improving the flexibility and scalability of the system and being able to handle sudden large traffic demands. Through precise resource allocation and efficient transmission strategies, unnecessary resource waste is reduced, operating costs are lowered, and the cost-effectiveness of the entire system is improved. Description of the Drawings

[0052] Figure 1It is a schematic diagram of a cloud classroom live transmission method based on the Internet provided by an embodiment of the present application. Detailed implementation manners

[0053] Next, in combination with the accompanying drawings and specific implementation manners, the present application will be further described. It should be noted that, on the premise of no conflict, the following-described embodiments or technical features can be arbitrarily combined to form new embodiments.

[0054] Refer to Figure 1 , an embodiment of the present application provides a cloud classroom live transmission method based on the Internet, and the method includes:

[0055] Obtain user parameters and live event parameters, where the user parameters include the user's network condition and data transmission quality, and the live event parameters include the content type of the live event, the platform network condition, the number of participants, and the first live data;

[0056] Allocate the first broadband resource of the live event according to the content type of the live event, the platform network condition, and the number of participants;

[0057] Based on a transmission policy set, transmit the first live data from a first device set to a second device set, where the transmission policy set includes a plurality of different transmission policies, the first device set includes one or more first devices, and the second device set includes one or more second devices;

[0058] Dynamically adjust each transmission policy according to the user's network condition and data transmission quality, where the transmission policy includes intelligently selecting a data transmission link, dynamically adjusting media quality parameters, and dynamically adjusting the buffer size and the time interval between two adjacent adjustments.

[0059] Among them, obtaining user parameters and live event parameters includes:

[0060] In response to account password recognition or biometric recognition, obtain user information; the user identification includes user basic information, device information, and geographical location; allocate corresponding functions and permissions according to the user basic information;

[0061] Obtain the first live data of the lecturer and user interaction information, where the interaction information includes user interaction information, operation logs, and feedback evaluations.

[0062] The working principle of the above technical solution is:

[0063] First, perform user identity verification through account password recognition or biometric recognition technology to obtain user information, including the user's basic information, the information of the used device, and the geographical location. These information are used for subsequent service personalization and permission management.

[0064] Collect relevant parameters of the live event, including the content type of the live broadcast (such as lectures, seminars, experimental demonstrations, etc.), the current network status of the platform, the expected number of participants, and the first live data provided by the lecturer. According to the collected live event parameters, the system intelligently allocates the first broadband resources. This process takes into account the content type (which affects the required bandwidth), the platform network status (which may limit the actual available bandwidth), and the number of participants (which affects the bandwidth demand) to ensure the smooth progress of the live event.

[0065] The first live data is transmitted from the first device set (usually the lecturer's device) to the second device set (the participants' devices); this process adopts a set of transmission strategies, which includes a variety of different transmission strategies to adapt to different network conditions and user needs. The set of transmission strategies includes but is not limited to intelligently selecting the data transmission link, dynamically adjusting media quality parameters (such as video resolution, frame rate, etc.), and dynamically adjusting the buffer size and the adjustment time interval. These strategies can be dynamically adjusted according to the real-time monitored user network status and data transmission quality to maintain the optimal live experience.

[0066] Continuously monitor the user's network status and data transmission quality, and dynamically adjust the transmission strategy according to these changes. For example, if it is detected that a user's network connection deteriorates, the system may reduce the media quality transmitted to the user or increase the buffer size to prevent playback interruption.

[0067] The effects of the above technical solutions are as follows:

[0068] By dynamically adjusting the transmission strategy according to the user's transmission link network status and data transmission quality, such as intelligently selecting the data transmission link, dynamically adjusting media quality parameters, etc., it can ensure that the video stream remains smooth under different network conditions, significantly improving the user's viewing experience; according to the user's basic information and geographical location and other information, the system can provide more personalized services and recommendations, such as adjusting the live content push according to the user's preferences, enhancing user satisfaction. Allocating the first broadband resources according to the content type of the live event, the platform network status, and the number of participants can ensure that the live event can obtain the optimal bandwidth support under limited resources, avoid resource waste, and improve resource utilization efficiency. Dynamically adjusting the buffer size and the time interval between two adjacent adjustments can further optimize the stability of data transmission, reduce the stuttering phenomenon caused by network fluctuations, and at the same time ensure the continuity and integrity of the data.

[0069] By obtaining user interaction information, such as user interaction information, operation logs, and feedback evaluations, the system can understand the feedback and needs of users in real time, thereby adjusting the live content and form, enhancing the interactivity between teachers and students, and the participation of students. The feedback evaluations of users can help lecturers adjust teaching strategies in a timely manner and improve teaching quality; at the same time, the operation logs of users also provide valuable data support for system optimization. Obtaining user information through account password recognition or biometric technology improves the security of the system and the accuracy of user identities, preventing the risks of illegal access and data leakage. The combined use of multiple transmission strategies and dynamic adjustment according to network conditions ensure the stability and reliability of data transmission, reducing the risks of live interruption or data loss caused by network problems.

[0070] In some embodiments, allocating the first broadband resource of the live event according to the content type of the live event, the platform network condition, and the number of participants includes:

[0071] Establishing a first broadband resource prediction model through historical data; the historical data includes the content type of the live event, the peak number of people, the average number of online people, the platform network condition, the type and total amount of live events in the same time period, and the total number of participants;

[0072] Predicting the first broadband resource of the live event corresponding to the (t + 1)th moment through the broadband resource prediction model according to the real-time data at the tth moment.

[0073] The working principle of the above technical solution is as follows: First, a large amount of historical data is collected, which covers various key indicators of different live events, including the content type of the live event (such as lectures, seminars, courses, etc.), the peak number of people (i.e., the maximum number of people online simultaneously during the live broadcast), the average number of online people, the platform network condition (such as network bandwidth, latency, packet loss rate, etc.), the type and total amount of live events in the same time period, and the total number of participants, etc.

[0074] The collected historical data will go through preprocessing steps, including data cleaning (removing errors or outliers), data conversion (such as converting timestamps to a unified date and time format), data standardization or normalization (ensuring that data with different dimensions can be fairly compared in the same model), etc., for subsequent modeling use.

[0075] Based on the preprocessed historical data, a prediction model is established using machine learning algorithms (such as linear regression, decision tree, random forest, neural network, etc.) or statistical methods. The goal of this model is to predict the amount of broadband resources required for a live event given the input features (such as the content type of the live event, the expected peak number of people, the average number of online people, the platform network status, etc.). The training process of the model involves adjusting the model parameters to minimize the prediction error (such as mean squared error, absolute error, etc.), and evaluating the generalization ability of the model through techniques such as cross-validation to ensure that the model can also perform well on unseen data.

[0076] At time t, the system collects real-time data on the upcoming or ongoing live event, including the current network status, the number of registered or online participants, the live content type, etc.; this real-time data is input into the trained broadband resource prediction model, and the model calculates the first amount of broadband resources required for the live event at time t+1 (i.e., when the live event is about to start or is ongoing) based on the input features.

[0077] According to the prediction results of the model, the corresponding broadband resources are allocated for the live event. If the prediction results show that the required resources exceed the current available resources, resource scheduling may be carried out in advance, such as borrowing resources from other non-peak live events, or notifying users to adjust the live settings to reduce the bandwidth demand. During the live broadcast, continuously monitor the network status and transmission quality, and dynamically adjust the allocated broadband resources as needed to ensure the smoothness and stability of the live broadcast. After the live broadcast ends, the actually used broadband resources will be compared with the predicted values to calculate the prediction error. These error data will be used to further optimize the prediction model and improve the accuracy of future predictions. At the same time, the system will also collect users' feedback opinions and live effect evaluation data as a reference for improving the live transmission method and optimizing resource allocation.

[0078] The effects of the above technical solution are as follows: Through the broadband resource prediction model based on historical data, the amount of broadband resources required for live events can be predicted more accurately, thus avoiding over-allocation or shortage of resources; this helps to ensure the effective utilization of resources, reduce resource waste, and lower operating costs. The prediction model can dynamically adjust the broadband resource allocation according to real-time data to ensure a stable network connection and a smooth viewing experience for users during the live broadcast. Even in the case of poor network conditions, it can reduce stuttering and latency through intelligent adjustment, improving user satisfaction; this solution enables the system to quickly adapt to changes in live events, such as a sudden increase in the number of participants or fluctuations in network conditions. Through real-time data-driven resource allocation, the system can respond quickly and make adjustments to ensure the smooth progress of live events. The prediction model established using machine learning algorithms and statistical methods can continuously collect data from actual usage and optimize itself, thereby improving the accuracy of prediction. Over time, the prediction ability of the model will become stronger, providing more accurate resource allocation suggestions for live events. On large educational platforms, it is often necessary to support multiple live events simultaneously. Through the intelligent broadband resource allocation mechanism, this technical solution enables the system to handle the resource requirements of multiple live events simultaneously, ensuring that each event can obtain sufficient bandwidth support to meet the needs of large-scale concurrency. When resource shortages are predicted, the system can perform resource scheduling in advance, such as borrowing resources from other live events during off-peak hours, or distributing traffic to multiple servers through load balancing technology. This helps to balance the system load, reduce the risk of single-point failures, and improve the stability and reliability of the system. By collecting and analyzing the actual broadband resource usage data, user feedback, and live broadcast effect evaluation data, the system can provide data-driven decision support for decision-makers. This helps to optimize the live broadcast transmission strategy, improve the resource allocation plan, and promote the continuous improvement and innovation of educational services.

[0079] In summary, when implemented in the live broadcast transmission of cloud classrooms, this technical solution can significantly improve resource utilization efficiency, optimize the user experience, enhance system flexibility and response speed, improve prediction accuracy, support large-scale concurrent live broadcasts, promote resource scheduling and load balancing, and provide data-driven decision support.

[0080] In some embodiments, real-time scoring is performed on multiple types of first live data; transmission priorities are obtained based on the real-time scoring; a transmission queue is set according to the transmission priorities; the types of the first live data include file data, video data, audio data, and / or writing data;

[0081] In response to changes in the file data, the first device only transmits the changed part of the file data to the second electronic device;

[0082] Set the transmission frequency of the first live data type of the corresponding type according to the refresh frequency of video data, audio data, or interactive data;

[0083] Conduct a comprehensive assessment of the network environment and intelligently select a data transmission link; dynamically adjust media quality parameters according to network conditions, user information, and environmental information;

[0084] Establish a multi-level caching mechanism and dynamically adjust the buffer size and the time interval between two adjacent adjustments according to the user's geographical location, network conditions, and node scores.

[0085] The working principle and effect of the above technical solution are as follows: First, conduct a real-time assessment of different types of first live data (file data, video data, audio data, writing data). This score takes into account factors such as the timeliness, importance, and user attention of the data.

[0086] Based on the real-time score, the system assigns a transmission priority to each data stream; the data stream with a higher priority will be transmitted first to ensure the timely delivery of key information. According to the transmission priority, the system establishes a transmission queue. Each position in the queue corresponds to a data stream, and the transmission is carried out in the order of priority. This helps to maximize the transmission efficiency of key information under limited network resources.

[0087] For file data, the system adopts a differential transmission strategy. When the file data changes, the system only transmits the changed part to the second electronic device instead of the entire file. This greatly reduces the amount of data transmitted and improves the transmission efficiency.

[0088] Transmission frequency based on the refresh frequency: The system dynamically sets the transmission frequency of the corresponding type of data according to the refresh frequency (i.e., update speed) of video data, audio data, or interactive data. Data with a high refresh frequency will be transmitted more frequently to meet real-time requirements; while data with a low refresh frequency will reduce the number of transmissions to save network resources.

[0089] Conduct a comprehensive assessment of the current network environment, including indicators such as bandwidth, latency, and packet loss rate. Based on the assessment results, intelligently select the optimal data transmission link to ensure the stability and efficiency of data transmission.

[0090] Dynamically adjust media quality parameters (such as video resolution, frame rate, audio bit rate, etc.) according to network conditions, user information (such as device performance, network package, and device battery health), and environmental information (including environmental noise and environmental brightness). When the network conditions are poor, reduce the media quality to ensure smoothness; when the network conditions are good, improve the media quality to enhance the viewing experience.

[0091] Establish a multi-level caching mechanism, and dynamically adjust the buffer size and the time interval between two adjacent adjustments according to the user's geographical location, network condition, and node score (i.e., performance, stability, and other indicators of the caching node). When the network is congested or the performance of the user device is low, increase the buffer size to reduce the lag caused by network latency or insufficient processing power; when the network is unobstructed or the performance of the user device is high, reduce the buffer size to release the caching resources.

[0092] In summary, through strategies such as real-time scoring, differential transmission, intelligent link selection, dynamic media quality adjustment, and multi-level caching mechanism, this transmission policy set realizes the comprehensive optimization of live data transmission. These strategies cooperate with each other to ensure the efficient and stable transmission of live data and improve the user experience.

[0093] In some embodiments, comprehensively evaluate the network environment and intelligently select the data transmission link; including:

[0094] According to the user's geographical location, obtain multiple communication links between the sender and the receiver through the network topology structure;

[0095] According to the first link score, geographical distance, and link load at time t, predict the available links at time t + 1 through constraint conditions.

[0096] In some embodiments, the step of predicting the available links at time t + 1 according to the first link score, geographical distance, and link load at time t through constraint conditions includes:

[0097] Collect the real-time performance indicators of the available links, where the performance indicators include packet loss rate, packet loss residual, packet loss rate residual change rate, real-time bandwidth, latency time, and link reliability;

[0098] Output the first link score at time t through a machine learning model;

[0099] Use time series analysis to predict the link performance indicators at time t + 1, and combine the first link score at time t to predict the first score at time t + 1;

[0100] Calculate the second link score at time t + 1 through the prediction of the geographical distance and load of the link;

[0101] Define the constraint conditions, where the constraint conditions include the maximum acceptable packet loss rate, minimum bandwidth requirement, maximum latency, and load balancing;

[0102] Predict the available links at time t + 1 according to the second link score and constraint conditions of the link at time t + 1;

[0103] According to the redundancy requirements of the service, select multiple available links; and select the second-highest scoring available link as the primary link; allocate the data in the transmission queue to different available links based on the data volume in the transmission queue and the link load.

[0104] The working principle of the above technical solution is as follows: According to the user's geographical location, first use the existing network topology information to determine multiple possible communication links between the sending end (such as a live server) and the receiving end (such as a user device). These links may include different network operators, different physical paths, or different transmission technologies (such as wired, wireless).

[0105] For each discovered link, the system collects its performance metrics in real time. These metrics include, but are not limited to, packet loss rate, packet loss residual, change rate of packet loss rate residual, real-time bandwidth, latency time, and link reliability. These metrics reflect the current state and quality of the link.

[0106] Using a machine learning model, the system scores each link based on the collected real-time performance metrics (referred to as the first score). This score comprehensively considers multiple aspects of the link to evaluate its advantages and disadvantages at the current moment.

[0107] The system uses time series analysis techniques, combines historical data and the current state, and predicts the performance metrics of the link at time t+1. Then, combined with the first score of the link at time t, through a machine learning model or statistical method, predicts the first score of the link at time t+1; which helps to understand the future link conditions in advance.

[0108] In addition to performance metrics, also consider the geographical distance and current load of the link. Geographical distance affects the latency and cost of data transmission, while the load reflects the congestion level of the link. By comprehensively considering these factors, calculate the second score of the link at time t+1. The second score can be a weighted score of the first score, geographical distance score, and current load score;

[0109] Define a series of constraint conditions, such as the maximum acceptable packet loss rate, minimum bandwidth requirement, maximum latency, and load balancing, etc. These conditions reflect the minimum requirements of the service for link quality; according to the second score of the link at time t+1 and these constraint conditions, the system filters out the available links.

[0110] To improve the reliability and fault tolerance of transmission, the system selects multiple available links according to the redundancy requirements of the service. Among the multiple available links, select the link with the second-highest score as the primary link to ensure the transmission quality of the main data stream.

[0111] Dynamically allocate data to different available links according to the data volume in the transmission queue and the load conditions of each link. This helps to balance the load of each link and improve the overall transmission efficiency. At the same time, the system will also monitor the real-time performance of each link and make dynamic adjustments as needed.

[0112] In summary, by comprehensively evaluating the network environment, intelligently selecting data transmission links, and combining real-time performance metrics, geographical distance, link load, and business constraints, the efficient and flexible selection and management of live data transmission links are achieved. This helps to improve the stability, reliability, and efficiency of data transmission, thereby enhancing the user experience.

[0113] The effects of the above technical solutions are as follows: By collecting real-time performance metrics and using machine learning models for scoring, the system can accurately evaluate the quality of the current link. Combining time series analysis to predict future link states can detect and avoid potential network problems in advance, thus significantly improving the stability of data transmission. Selecting the link with the optimal performance as the main link and reasonably allocating the data in the transmission queue to different available links can reduce problems such as latency and packet loss, ensuring that users can smoothly receive live content or data and enhancing user satisfaction and experience.

[0114] Select multiple available links according to the redundancy requirements of the service and quickly switch to the backup link when the main link fails, ensuring the continuity and reliability of data transmission. This redundant design greatly reduces the risk of system interruption due to single-point failures. Allocating data by considering the load conditions of the links helps to achieve load balancing and avoid performance degradation of some links due to overload. This can not only improve the utilization rate of the overall network but also extend the service life of the equipment.

[0115] In some embodiments, the method further includes:

[0116] Establish a closed-loop feedback mechanism to continuously monitor the link performance and adjust the parameters of the scoring model and prediction algorithm according to the actual performance;

[0117] Adaptive adjust the constraint conditions and scoring weights according to historical data and newly emerging network conditions; introduce an anomaly detection algorithm to identify and isolate abnormal behaviors and / or faulty links in the network.

[0118] The working principle and effects of the above technical solution are as follows: By deploying monitoring points or agents in the network, real-time performance metric data of the link is continuously collected, such as packet loss rate, bandwidth, latency, etc. The collected data is input into a scoring model and a prediction algorithm for real-time evaluation of the link performance. The evaluation results are not only used for the selection of the current link but also serve as the basis for subsequent adjustment of the parameters of the scoring model and the prediction algorithm. According to the actual performance of the link, the system dynamically adjusts the parameters of the scoring model and the prediction algorithm. These adjustments aim to improve the accuracy of scoring and prediction to better adapt to changes in the network environment. The system analyzes historical data to understand the link performance characteristics under different time periods and different network conditions. This data provides an important basis for the adjustment of constraint conditions and scoring weights. At the same time, the system also continuously monitors new network conditions, such as network congestion, equipment failures, etc., which will affect the link selection and scoring.

[0119] Based on historical data and newly emerging network conditions, the system adaptively adjusts the constraint conditions and scoring weights. For example, during network congestion periods, the importance of bandwidth and latency may be increased, and the tolerance for packet loss rate may be reduced.

[0120] First, the system preprocesses the collected link performance data, such as denoising, normalization, etc., to improve the accuracy of anomaly detection. Then, it uses an anomaly detection algorithm to analyze the processed data to identify data points that deviate significantly from the normal pattern, i.e., abnormal behaviors or faulty links. Once abnormal behaviors or faulty links are identified, the system immediately takes measures to isolate them to prevent them from affecting the overall network performance. At the same time, it also triggers a fault recovery mechanism to attempt to restore the normal operation of the faulty link or switch to a backup link.

[0121] Through the above working principle, the method can establish a dynamic and adaptive network environment evaluation and link selection mechanism. The closed-loop feedback mechanism ensures the continuous optimization of the scoring model and the prediction algorithm; the adaptive adjustment of constraint conditions and scoring weights enables the system to flexibly respond to changes in the network environment; the anomaly detection algorithm enhances the stability and reliability of the system. Under the combined action of these mechanisms, it can provide users with high-quality and stable data transmission services.

[0122] In some embodiments, the data in the transmission queue is allocated to different available links according to the data volume and link load in the transmission queue; including:

[0123] The data with the same priority is divided into different data blocks; a hash function is used to determine the boundaries of the data blocks; it is ensured that data blocks with the same content have the same hash value;

[0124] The data block with the highest priority is allocated to the primary link until a preset load threshold is reached;

[0125] The remaining data blocks are sequentially allocated according to the second score of the remaining selection link and the load of the corresponding link.

[0126] The working principle of the above technical solution is as follows: The data sets with the same priority are divided into multiple data blocks according to a certain strategy (such as fixed size, content characteristics, etc.). The purpose of doing this is to facilitate management and transmission, and at the same time improve the flexibility of data processing.

[0127] Each part of the data is processed using a hash function, and the boundaries of the data blocks are determined according to the hash values. The hash function has the characteristic of transforming an input of any length into an output of a fixed length (i.e., the hash value) through a hashing algorithm. Due to the determinism of the hash function, data blocks with the same content will generate the same hash value, thus ensuring the consistency and identifiability of the data blocks. The integrity and consistency of the data blocks are verified by comparing the hash values. If the hash values of two data blocks are the same, it can be considered that the two data blocks are consistent in content.

[0128] Sort according to the priority of the data. The data blocks with higher priority will be processed first. The data blocks with the highest priority are allocated to the main link for transmission. The main link is usually the link with the best performance and the highest stability, which can ensure the fast and reliable transmission of high-priority data blocks. During the allocation process, it is necessary to monitor the load of the main link to ensure that it does not exceed the preset load threshold. Once the threshold is reached, the allocation of data blocks to the main link will be stopped. For the remaining data blocks, they are sequentially allocated according to the second score of the remaining selection link (i.e., the score obtained by comprehensively considering factors such as link performance, geographical distance, load, etc.) and the current load of the corresponding link; during allocation, the link with a high score and a low load is preferred to ensure the efficiency and stability of data transmission.

[0129] The effects of the above technical solution are as follows: Prioritizing the allocation of high-priority data blocks to the main link with better performance can ensure that these key data are processed quickly, thereby shortening the overall transmission time and improving the transmission efficiency. Allocating the remaining data blocks according to the second score of the link and the current load can ensure the reasonable allocation of resources, avoid overloading some links while other links are idle, and thus improve the resource utilization rate of the entire network. Determining the data block boundaries through the hash function and ensuring that data blocks with the same content have the same hash value helps to verify the data integrity and consistency during data transmission, thereby enhancing the stability and reliability of the system.

[0130] This strategy has the ability to dynamically adjust, and can adjust the data allocation scheme in real time according to changes in network conditions (such as link load, performance fluctuations, etc.) to ensure the continuity and stability of data transmission. By decomposing complex data transmission tasks into multiple manageable data blocks and processing them based on clear priorities and allocation rules, the complexity and difficulty of data transmission management can be greatly reduced.

[0131] In some embodiments, dynamically adjusting media quality parameters according to network conditions, user information, and environmental information includes:

[0132] Collect the first score of the user - side transmission link, user information, live broadcast time, environmental information, and media quality parameters; the user information includes user device information, device battery life and power, geographical location, and preferences; the media quality parameters include audio quality, media clarity, and interaction response speed; if the live broadcast data is transmitted to a certain user - side through multiple links, the first scores of the multiple links of the user - side are fused to obtain the link - fused score of the user - side, and the fused score is used as the input of the deep - learning model;

[0133] Among them, the fused score is:

[0134]

[0135] Where Rj is the fused score of n transmission links of the j - th user - side, Pij is the first score of the i - th transmission line of the j - th user - side; wij is the weight of the i - th transmission line of the j - th user - side; Pij is the maximum load of the i - th transmission line of the j - th user - side, and Dij is the transmission distance of the i - th transmission line of the j - th user - side;

[0136] Train the deep - learning model with historical training data to learn the relationship between media quality and user satisfaction under different conditions;

[0137] According to the real - time network conditions, user information, live broadcast time, and environmental information, use the deep - learning model to predict the optimal media quality parameters.

[0138] The working principle of the above - mentioned technical solution is as follows:

[0139] For the case of transmitting to the user - side through multiple links, first collect the first score (Pij) of each transmission link, and this score is comprehensively obtained based on the performance indicators of the link (such as bandwidth, delay, packet loss rate, etc.) and environmental information.

[0140] Collect relevant information of the user side, including user device information (such as device model, processing capacity), device battery life, geographical location, and user preferences, etc. Record the specific time of the live broadcast because the network conditions may vary at different time periods. Collect the environmental information where the user side is located, such as network coverage, weather conditions, etc. These information may affect the performance of the transmission link.

[0141] Set the initial media quality parameters, including audio quality, media clarity, and interactive response speed, etc. These parameters will be used as the basis for subsequent adjustments. For the case of multiple transmission links, use the given formula to calculate the link fusion score; this way of weight assignment takes into account the load capacity and transmission cost (distance) of the link to more comprehensively evaluate the performance of the link.

[0142] Utilize historical training data, including network conditions, user information, live broadcast time, environmental information, and corresponding media quality parameters under different conditions, to train a deep learning model. The model learns the relationship between media quality and user satisfaction and establishes a prediction model to predict the optimal media quality parameters under given conditions. During the live broadcast, according to the real-time network conditions, user information, live broadcast time, and environmental information, use the trained deep learning model for prediction. The model outputs the optimal media quality parameters, including audio quality, media clarity, and interactive response speed, etc., to achieve dynamic adjustment of media quality. According to the prediction results of the deep learning model, adjust the media quality parameters to match the current network conditions and user needs.

[0143] For example, when the network condition is poor, reduce the media clarity to reduce lags, or when the user device has high performance, improve the picture and sound quality; reduce the media clarity when the user's battery power is low and the battery health is poor; when the battery power of the user device is lower than a certain threshold (such as 20%), the system can automatically reduce the media clarity, reduce the video frame rate or adjust the audio quality to reduce power consumption and extend the device's usage time. For example, switch from high definition to standard definition, or reduce from 60 frames per second to 30 frames per second. At the same time, the system can prompt the user that the current battery power is low and ask whether the user is willing to continue watching in the current media quality, or automatically switch to a more power-saving viewing mode. If the battery health of the user device is poor (such as the capacity is reduced due to battery aging), similar measures can also be taken to reduce power consumption. This includes reducing media quality parameters, adjusting the screen brightness, etc. to relieve the battery burden and prevent the battery from running out too quickly. Regularly detect the battery health and automatically adjust the media quality parameters according to the detection results to ensure that the user can still obtain a relatively smooth viewing experience when the battery condition is poor. More complex adjustment strategies can also be formulated according to the comprehensive situation of the battery power and battery health. For example, when the battery power is low and the battery health is poor, the system can further reduce the media quality parameters to extend the device usage time to the greatest extent. The system can also consider the user's preferences and viewing habits. For example, if the user often turns off the screen to save power when watching a live broadcast, the system can remember this habit and automatically adjust the media quality parameters to adapt to this scenario when the user watches a live broadcast again.

[0144] The effects of the above technical solutions are as follows: By evaluating the network condition, user information, and environmental information in real time and adjusting the media quality parameters accordingly, it can ensure that users can obtain the best viewing experience under different conditions.

[0145] When transmitting through multiple links, by calculating the link aggregation score and selecting the optimal link or combination, network resources can be utilized more effectively, avoiding performance bottlenecks caused by overloading of a single link. At the same time, allocating weights according to the load capacity and transmission cost of the link helps to achieve balanced utilization of resources. When the network condition is poor, reduce the load of data transmission by reducing the media quality, thereby effectively reducing the occurrence of lags and delays. When the network condition is good and the user device has high performance, improve the media quality parameters (such as media clarity and audio quality) so that users can obtain a clearer and more realistic viewing experience. By optimizing the interactive response speed part of the media quality parameters, it can ensure that users can interact smoothly when watching a live broadcast, improving the overall user experience.

[0146] In some embodiments, the establishment of the multi-level cache mechanism, which dynamically adjusts the buffer size according to the user's geographical location, network condition, and node score, includes:

[0147] Construct a distributed cache network; the distributed cache network includes multiple nodes; and obtain node scores;

[0148] According to the node scores, adjust the size of the buffer and the time interval between two adjacent buffer size adjustments.

[0149] Among them, the node score is:

[0150] If any one of the values is greater than 0, then

[0151]

[0152] If all are less than or equal to 0, then:

[0153]

[0154] Among them, Qk is the current data volume of the kth node, Qka is the average data volume of the kth node within a preset time period under the current buffer area size setting; Fk is the current load of the kth node, Fka is the average load of the kth node within a preset time period under the current buffer area size setting; Tk is the current delay of the kth node, Tka is the average delay of the kth node within a preset time period under the current buffer area size setting; max() is to take the maximum value;

[0155] If the node score is greater than the first preset threshold, perform the first adjustment to adjust the size of the buffer;

[0156]

[0157] If the node score is less than the second preset threshold, perform the second adjustment to adjust the size of the buffer;

[0158]

[0159] Among them, skt1 is the buffer size of the kth node after the first adjustment; skt2 is the buffer size of the kth node after the second adjustment; α is the adjustment coefficient, with a range of (0 to 3); Skc is the current buffer size of the kth node, Jy1 is the first threshold; Jy2 is the second threshold; both skt1 and skt2 are within the range of [skmin, skmax], skmin is the minimum required buffer size; skmax is the maximum required buffer size.

[0160] If the interval from the previous adjustment time point to the current time point is less than the minimum time interval, no adjustment is made currently, and the next adjustment will be made.

[0161] The working principle of the above technical solution is as follows: A distributed cache network is constructed, which consists of multiple nodes. Each node is responsible for storing and distributing data. These nodes are distributed in different geographical locations to optimize the data access speed and reliability.

[0162] Calculate the score (Jk) of each node regularly (or triggered by a specific event); the score is based on three key metrics: data volume, load, and latency.

[0163] Compare the current data volume (Qk) with the average data volume (Qka) within a preset time period.

[0164] Compare the current load (Fk) with the average load (Fka) within a preset time period.

[0165] Compare the current latency (Tk) with the average latency (Tka) within a preset time period.

[0166] For each metric, if the current value is greater than the average value, calculate the growth rate of the metric (i.e., (current value - average value) / average value), and take the maximum value among all growth rates as the score (Jk) of the node. If the growth rates of all metrics are less than or equal to 0, take the minimum value as the score. Although in this case, the score may be a negative number or very close to 0, indicating that the node state is relatively stable or has declined.

[0167] According to the node score (Jk), the system dynamically adjusts the buffer size of each node.

[0168] If Jk is greater than the first preset threshold (Jy1), indicating that the node may face a high load or data volume growth, the system will increase the buffer size to cope with it. The specific adjustment formula is skt1 = (1 + α × (Jk - Jy1) / JK) × Skc, where α is the adjustment coefficient, JK may be a normalization factor or some aggregation value of all node scores (depending on the system design), and Skc is the current buffer size.

[0169] Less than the second preset threshold: If Jk is less than the second preset threshold (Jy2), indicating that the node is currently relatively idle, the system may reduce the buffer size to release resources. The specific adjustment formula is skt2 = (1 - α × (Jk - Jy2) / Jy2) × Skc.

[0170] The adjusted buffer size (skt1 or skt2) must be between the minimum (skmin) and maximum (skmax) required buffer sizes.

[0171] To avoid overly frequent adjustments, the system also takes into account the interval from the previous adjustment time point to the current time point. If this interval is less than the set minimum time interval, the system will postpone the next adjustment until the minimum time interval is reached and then conduct an evaluation and adjustment.

[0172] In summary, this multi-level caching mechanism dynamically adjusts the buffer size by comprehensively considering node status (including data volume, load, and latency), as well as user geographical location and network conditions, in order to optimize cache performance and resource utilization.

[0173] The effects of the above technical solution are as follows: By dynamically adjusting the buffer size, the system can manage cache resources more effectively, reduce the waiting time during data access, and thus reduce latency. During peak periods, by increasing the buffer size, the system can handle more concurrent requests and improve the overall throughput. Dynamically adjusting the buffer size according to the actual load and data volume of the nodes avoids waste and over-allocation of resources. By reducing unnecessary cache resource usage, storage and maintenance costs can be reduced; by dynamically adjusting the buffer size between different nodes, a more balanced load distribution can be achieved, reducing the pressure on individual nodes.

[0174] This application provides an Internet-based cloud classroom live transmission system, and the system includes:

[0175] A first acquisition module, configured to acquire user parameters and live event parameters, where the user parameters include the user's network condition and data transmission quality, and the live event parameters include the content type of the live event, the platform network condition, the number of participants, and the first live data;

[0176] A first allocation module, configured to allocate the first broadband resources of the live event according to the content type of the live event, the platform network condition, and the number of participants;

[0177] A transmission module, configured to transmit the first live data from the first device set to the second device set based on a transmission policy set, where the transmission policy set includes a plurality of different transmission policies, the first device set includes one or more first devices, and the second device set includes one or more second devices;

[0178] An adjustment module, configured to dynamically adjust each transmission policy according to the user's network condition and data transmission quality, where the transmission policies include intelligent selection of data transmission links, dynamic adjustment of media quality parameters, dynamic adjustment of buffer size, and the time interval between two adjacent adjustments.

[0179] In some embodiments, the first allocation module includes:

[0180] A first model building unit is used to build a first broadband resource prediction model through historical data; the historical data includes the content type of live broadcast activities, peak number of people, average number of online people, platform network status, type and total amount of live broadcast activities in the same time period, and total number of participants;

[0181] The first prediction unit is used to predict the first broadband resource corresponding to the live broadcast activity at time t+1 according to the real-time data at time t through the broadband resource prediction model.

[0182] In some embodiments, the transmission strategy set includes:

[0183] Performing real-time scoring on multiple types of first live broadcast data; obtaining a transmission priority according to the real-time scoring; setting a transmission queue according to the transmission priority; the types of the first live broadcast data include file data, video data, audio data and / or written data;

[0184] In response to a change in the file data, the first device transmits only the changed portion of the file data to the second electronic device;

[0185] According to the refresh frequency of the video data, the audio data or the interactive data, the transmission frequency of the first live data type of the corresponding type is set;

[0186] Comprehensively evaluate the network environment and intelligently select data transmission links; dynamically adjust media quality parameters based on network conditions, user information, and environmental information;

[0187] A multi-level cache mechanism is established to dynamically adjust the buffer size and the time interval between two adjacent adjustments based on the user's geographic location, network conditions, and node scores.

[0188] The working principle and effect of the above technical solution are the same as those in the embodiment of the method of the present application and will not be described in detail here.

[0189] This application is explained from the perspectives of purpose of use, effectiveness, progress and novelty, and has met the functional enhancement and usage requirements emphasized by the Patent Law. The above description and drawings of this application are only the preferred embodiments of this application, and are not intended to limit this application. Therefore, all structures, devices, features, etc. that are similar or identical to this application, that is, all equivalent replacements or modifications made in accordance with the scope of the patent application of this application, should fall within the scope of protection of the patent application of this application.

Claims

1. A cloud classroom live transmission method based on the Internet, characterized in that: The method comprises: Acquire user parameters and live broadcast activity parameters, and allocate a first broadband resource for the live broadcast activity according to the content type, platform network status, and number of participants of the live broadcast activity; Based on the transmission strategy set, transmitting first live broadcast data from the first device set to the second device set, wherein the type of the first live broadcast data includes file data, video data, audio data and / or written data; Dynamically adjust each transmission strategy according to the user's network status and data transmission quality; The transmission strategy set includes: Establish a multi-level cache mechanism to dynamically adjust the buffer size and the time interval between two adjacent adjustments based on the user's geographic location, network conditions, and node score. Specifically, it includes: Constructing a distributed cache network; the distributed cache network includes a plurality of nodes; and obtaining node scores; Among them, the node score Jk is: like , , If any value in is greater than 0, then: ； like , , are less than or equal to 0, then: ； Wherein, Qk is the current data volume of the kth node, Qka is the average data volume of the kth node in the preset time period under the current buffer area size setting; Fk is the current load of the kth node, Fka is the average load of the kth node in the preset time period under the current buffer area size setting; Tk is the current delay of the kth node; Tka is the average delay of the kth node in the preset time period under the current buffer area size setting; max() is the maximum value; min() is the minimum value; If the node score is greater than a first preset threshold, a first adjustment is performed to adjust the size of the buffer; ； If the node score is less than the second preset threshold, a second adjustment is performed to adjust the size of the buffer; ； Among them, skt1 is the buffer size of the kth node after the first adjustment; skt2 is the buffer size of the kth node after the second adjustment; α is the adjustment coefficient, and its value range is (0, 3); Skc is the current buffer size of the kth node, Jy1 is the first threshold; Jy2 is the second threshold; skt1 and skt2 are both within the range of [skmin, skmax], skmin is the minimum required buffer size; skmax is the maximum required buffer size; If the interval from the last adjustment time point to the current time point is less than the minimum time interval, no adjustment is made at the current time and adjustment is made next time.

2. The method according to claim 1, characterized in that: The user parameters include the user's network status and data transmission quality, and the live broadcast activity parameters include the content type of the live broadcast activity, the platform network status, the number of participants, and the first live broadcast data; The first device set includes one or more first devices, and the second device set includes one or more second devices.

3. The method according to claim 2, characterized in that The transmission strategy set also includes: Performing real-time scoring on multiple types of first live broadcast data; obtaining a transmission priority according to the real-time scoring; and setting a transmission queue according to the transmission priority; In response to a change in the file data, the first device transmits only the changed portion of the file data to the second device; According to the refresh frequency of the video data, the audio data or the interactive data, the transmission frequency of the first live data of the corresponding type is set; Comprehensively evaluate the network environment and intelligently select data transmission links; dynamically adjust media quality parameters based on network conditions, user information, and environmental information.

4. The method according to claim 1, characterized in that The allocating a first broadband resource for the live broadcast activity according to the content type, platform network status, and number of participants of the live broadcast activity comprises: Establishing a first broadband resource prediction model through historical data; the historical data includes content type of live broadcast activities, peak number of people, average number of online people, platform network status, type and total amount of live broadcast activities in the same time period, and total number of participants; According to the real-time data at time t, the first broadband resource corresponding to the live broadcast activity at time t+1 is predicted through the broadband resource prediction model.

5. The method according to claim 3, characterized in that: The comprehensive assessment of the network environment and intelligent selection of data transmission links include: According to the user's geographical location, multiple communication links between the sender and the receiver are obtained through the network topology; According to the first link score, geographical distance and link load at time t, the available links at time t+1 are predicted through constraints.

6. The method according to claim 5, characterized in that The method predicts the available links at time t+1 according to the first link score, geographical distance and link load at time t through constraint conditions, including: Collect real-time performance indicators of available links, including packet loss rate, packet loss residual, packet loss rate residual change rate, real-time bandwidth, delay time, and link reliability; Through the machine learning model, the first link score at time t is output; Use time series analysis to predict the link performance index at time t+1, and combine the first link score at time t to predict the first score at time t+1; Calculate the second link score at time t+1 based on the geographical distance and load prediction of the link; defining constraints, including maximum acceptable packet loss rate, minimum bandwidth requirement, maximum delay, and load balancing; According to the second score and constraint condition of the link at time t+1, predict the available link at time t+1; According to the redundancy requirements of the business, multiple available links are selected; and the available link with the second highest score is selected as the main link; according to the amount of data in the transmission queue and the link load, the data in the transmission queue is distributed to different available links.

7. The method according to claim 6, characterized in that The method further comprises: Establish a closed-loop feedback mechanism to continuously monitor link performance and adjust the parameters of the scoring model and prediction algorithm based on actual performance; Adaptively adjust constraints and scoring weights based on historical data and emerging network conditions; introduce anomaly detection algorithms to identify and isolate abnormal behaviors and / or faulty links in the network.

8. The method according to claim 6, characterized in that The method allocates the data in the transmission queue to different available links according to the amount of data in the transmission queue and the link load; comprising: Divide data of the same priority level into different data blocks; determine the boundaries of data blocks through hash functions; ensure that data blocks with the same content have the same hash value; Assign the highest priority data blocks to the primary link until the preset load threshold is reached; The remaining data blocks are distributed in sequence according to the second scores of the remaining selected links and the loads of the corresponding links.

9. The method according to claim 6, characterized in that The method of dynamically adjusting the media quality parameters according to the network status, user information and environment information includes: Collecting the first score of the user-side transmission link, user information, live broadcast time, environmental information and media quality parameters; the user information includes user device information, device battery life and power, geographic location and preferences; the media quality parameters include audio quality, media clarity and interactive response speed; Through historical training data, deep learning models are trained to learn the relationship between media quality and user satisfaction under different conditions; Based on real-time network conditions, user information, live broadcast time and environmental information, deep learning models are used to predict optimal media quality parameters.

10. A cloud classroom live transmission system based on the Internet, characterized in that: The system comprises: The first acquisition module is used to acquire user parameters and live broadcast activity parameters; A first allocation module allocates a first broadband resource for the live broadcast activity according to the content type of the live broadcast activity, the platform network status, and the number of participants; A transmission module, configured to transmit first live broadcast data from a first device set to a second device set based on a transmission strategy set, wherein the type of the first live broadcast data includes file data, video data, audio data and / or written data; The adjustment module dynamically adjusts each transmission strategy according to the user's network status and data transmission quality; The transmission strategy set includes: Establish a multi-level cache mechanism to dynamically adjust the buffer size and the time interval between two adjacent adjustments based on the user's geographic location, network conditions, and node score. Specifically, it includes: Constructing a distributed cache network; the distributed cache network includes a plurality of nodes; and obtaining node scores; Among them, the node score Jk is: like , , If any value in is greater than 0, then: ； like , , are both less than or equal to 0, then: ； Wherein, Qk is the current data volume of the kth node, Qka is the average data volume of the kth node in the preset time period under the current buffer area size setting; Fk is the current load of the kth node, Fka is the average load of the kth node in the preset time period under the current buffer area size setting; Tk is the current delay of the kth node; Tka is the average delay of the kth node in the preset time period under the current buffer area size setting; max() is the maximum value; min() is the minimum value; If the node score is greater than a first preset threshold, a first adjustment is performed to adjust the size of the buffer; ； If the node score is less than the second preset threshold, a second adjustment is performed to adjust the size of the buffer; ； Among them, skt1 is the buffer size of the kth node after the first adjustment; skt2 is the buffer size of the kth node after the second adjustment; α is the adjustment coefficient, and its value range is (0, 3); Skc is the current buffer size of the kth node, Jy1 is the first threshold; Jy2 is the second threshold; skt1 and skt2 are both within the range of [skmin, skmax], skmin is the minimum required buffer size; skmax is the maximum required buffer size; If the interval from the last adjustment time point to the current time point is less than the minimum time interval, no adjustment is made at the current time and adjustment is made next time.

Images