| In recent years,the rapid development and widespread adoption of realtime video applications have greatly changed peoples lifestyles,entertainment,and work methods.However,as the user base rapidly grows,the cost of CDN bandwidth for real-time video transmission service providers is also increasing.During the high concurrency periods of realtime video transmission services,most providers choose to lower the video bit rate to save bandwidth,but this strategy inevitably leads to a decrease in video quality.Considering that users have increasingly higher requirements for video quality,this solution of degrading user experience quality(QoE)is not a fundamental solution and is gradually becoming obsolete.Recently,the industry has been hoping to change this dilemma by using a P2P transmission mode.However,traditional P2P systems face two critical problems:1.the lack of real-time scheduling algorithms,and 2.high playback delay caused by video transmission protocols,making it difficult to provide real-time video delivery services.To address these challenges,this article proposes two key designs:1.Node scheduling algorithm based on random forest prediction:The algorithm first predicts the state of the link through a random forest to ensure the users transmission quality.Secondly,it improves the scheduling efficiency by optimizing the KM algorithm to ensure real-time performance.Finally,the algorithm is made more adaptable to dynamic network environments through mechanisms such as real-time monitoring and historical information.2.R2P System,a new type of live streaming system based on the SRS architecture and RTP/RTCP protocol:The SRS architecture and RTP/RTCP protocol solve the problem of high playback delay in existing P2P systems through realtime encoding,Simulcast,SVC,and other methods.At the same time,a subscription-based push-pull architecture is implemented in the system to further improve system performance and efficiency.This article verifies the performance of the above scheduling strategies and systems through joint debugging and simulation of the algorithm and system,and validates the performance of the algorithm under a scale of hundreds of thousands of users through a simulated transmission system.Based on experimental results,it can be seen that compared to CDN systems,the R2P system proposed in this article has improved by approximately 64.7%and 0.7%in terms of packet loss rate and throughput,respectively.In terms of frame rate,stutter rate,and jitter,the entire system has improved by 6.5%,50%,and 5.2%,respectively,compared to CDN systems,and the entire system saves approximately 68%of bandwidth.Through experiments,it has been verified that this system and algorithm have the advantages of low playback delay,high performance,high adaptability,and bandwidth savings for CDN. |
