Research And Implementation Of Video Conference System Based On Source-based Congestion Control Transmission

With the rapid development of the domestic economy and the widespread adoption of the Internet,there are more and more remote regional office scenarios.In particular,the needs of small and medium-sized enterprises are increasing,and the traditional and expensive hardware video conference systems cannot meet the needs of the market.Therefore,a low-cost,easy-to-implement pure software video conferencing system ushered in the climax of its development.However,the software-based video conferencing system has the characteristics of high bandwidth and low delay due to the real-time media stream it generates,and it has higher requirements for congestion control in the transmission process.The traditional congestion control is mostly based on the improvement of TCP.Th ere are generally problems such as low channel utilization,large queuing delay,and network friendliness(including inter-protocol fairness and TCP-friendliness).To solve these problems,this paper designs a novel congestion control algorithm based on mu ltiple indicators and carries out relevant experiments to verify its effectiveness.In order to meet the requirements of real-time media transmission with high bandwidth and low latency,this paper proposes a hybrid congestion control algorithm based on multiple congestion index fusion.Compared with the traditional single-indicator congestion control method,the congestion control algorithm in this paper uses a variety of congestion indicators such as one-way transmission delay,one-way transmission delay variation,and packet loss rate.Through the combination of one-way transmission delay and one-way transmission delay variation,two delay-based congestion indicators effectively avoid the "late arrival" effect when competing with the same stream and reduce the queuing delay.In addition,this paper designs a fine-grained packet loss-based congestion detection mechanism and a loss-interval model for absorbing aggregated packet loss.It uses the change in packet loss rate to determine congestion and indire ctly according to the average loss interval.Rate adjustments greatly increase channel utilization while maintaining a low loss rate.For the rate adjustment strategy,this paper designs a maximum rate estimation method and a rate convergence detection method,thereby increasing the stationarity of the rate and improving the network friendliness.Finally,experiments and comparisons are carried out in this paper through NS-3 network simulation tools.It is verified that the proposed algorithm not only has higher channel utilization and lower queuing delay,but also has network friendliness and can adapt to more network queue type.For pure software video conferencing system,this article implements a low-cost,easy-to-implement,easily-expanding video conferencing system through detailed design of the protocol stack,network topology architecture,servers,and clients.SIP-based signaling servers and ICE-based NAT traversal servers were designed to provide reliable channel transmission for media.An end-to-end congestion control mechanism is implemented on the client to enhance the control over real-time media transmission and improve Qo E support for video conferencing applications.The video conference system implemented in this paper can meet the small-scale multi-person video conference under time-varying network bandwidth.Compared with the existing system in the market,it has better adaptive variable bandwidth capability,lower implementation cost and more good user experience.