Research On Bandwidth Measurement Method Based On Real Time Video Transmission

In the application of real-time video transmission,such as video conference,remote control,remote driving and so on,users have higher requirements for real-time video transmission delay and video quality.For example,in 5G remote driving,the transmission delay of real-time video usually needs to be controlled within 100 ms.However,the transmission bandwidth and inherent propagation delay of the network are dynamic,especially in the rapid movement of nodes.Existing solutions usually adjust the video rate dynamically based on the dynamic bandwidth to ensure the stable end-to-end delay.However,the premise of this work is to provide accurate and reliable bandwidth prediction.In wireless networks,how to provide accurate and reliable bandwidth prediction is still a difficult problem.In this paper,real-time video transmission is taken as the application scenario,and the subject that how to use real-time video transmission message to predict bandwidth is mainly studied,which can provide accurate and reliable bandwidth prediction without affecting real-time video transmission.In the traditional real-time video transmission system,active measurement is usually used to measure the bandwidth,which needs to send additional detection messages,which will bring traffic burden to the video transmission system and affect the quality of video transmission.This paper proposes a new bandwidth measurement method for real-time video transmission system,which uses the transmitted video stream to measure the bandwidth.It detects the time-varying network transmission bandwidth by setting different packet intervals in the process of video stream transmission,and optimizes the network congestion problem by adjusting the packet interval.The experimental results show that this method can accurately measure the bandwidth without sending additional detection messages,and the error ratio of bandwidth estimation is controlled below 15%.According to the research,most of the existing bandwidth detection methods send detection packets,and have not been able to solve the problem of high end-to-end delay of video transmission.On the basis of using video stream packet to detect bandwidth,this paper optimizes the structure of video stream packet sequence and sets different packet intervals to send video stream sequence for bandwidth detection,so as to optimize the end-to-end delay.Specifically,this paper divides the video stream message sequence used to detect bandwidth into two parts: bandwidth detection video packets and balanced transmission rate video packets,which are used to detect bandwidth and reduce end-to-end delay respectively.By establishing a video transmission link model,the transmission interval of each video packet is determined with the goal of maximizing video transmission rate,so as to achieve bandwidth detection and improve the video transmission quality at the same time.The experimental results show that this method can accurately measure the link bandwidth value and effectively improve the quality of video transmission without sending additional detection packets.At the same time,the overdue loss rate of video transmission is improved to 3.51% compared with the above algorithm.The algorithms proposed in this paper are tested and verified on the actual platform.The proposed algorithm can be effectively applied to real-time video transmission system under heterogeneous wireless network,and can provide certain guidance for bandwidth measurement technology in real-time video transmission.