| With the increasing demand for online communication and the continuous development of network technology,real-time video has become increasingly widely used in fields such as live streaming and online education,bringing huge bandwidth costs to video suppliers.In this context,the use of P2P(Peer to Peer)technology for real-time video transmission through user sharing has received increasing attention.Compared to the server to user network link,the network jitter between user to user links in P2P networks is more frequent,and traditional realtime video congestion control algorithms will face greater dynamic challenges in P2P network environments.Therefore,optimizing the performance of congestion control algorithms in real-time video systems in P2P networks is of great significance for improving the quality of realtime video services.To address this challenge,this article has completed the following tasks:(1)Measure the transmission performance of P2P networks.Measure the network bandwidth and round-trip delay of end-to-end links in current P2P networks in multiple scenarios,and evaluate the performance indicators of mainstream real-time video congestion control algorithms,including the GCC(Google Congestion Control)algorithm.The experiment found that the current real-time video congestion control algorithm does not respond in a timely manner to more frequent network jitter and random packet loss in P2P network environments.(2)Propose the Dynamic-GCC algorithm.On the one hand,through dynamic parameter adjustment mechanism,flexible control of bit rate decisions is achieved,achieving fast response to network dynamics and improving the recovery speed of algorithm throughput;On the other hand,the introduction of packet loss correlation detection mechanism distinguishes between link random packet loss and congestion packet loss,reduces the interference of random packet loss on congestion determination in the link,and avoids erroneous bit rate reduction.(3)Design and implement a P2P based real-time video system MixedRT.The P2P transmission node and signaling server are designed and implemented respectively.The transmission node realizes the cascade forwarding of UDP(User Datagram Protocol)packet granularity,and the signaling server realizes NAT(Net Address Translation)traversal and transmission quality monitoring.The problem of Transmission delay in P2P network is optimized by hierarchical tree network topology optimization,and Dynamic GCC algorithm is deployed in this system.This article organized 40 nodes for evaluation experiments and collected over 200 hours of data.The results showed that Dynamic-GCC improved network transmission performance and user experience by 22.7%and 43.6%,respectively,compared to the most mainstream GCC algorithm. |
