Self-adaptive Control Study Of Real-time Video Transmission

As a multimedia stream, real-time video stream is carried in a series of successive packets with timestamp. Video transmission is the application of the streaming media. The streaming media is a continuous time-based media using streaming transmission over the Internet/Intranet. Transmission of the real-time video has a higher demand in timely, network bandwidth and delay, so the application must provide the service quality assurance, as packet loss rate, throughput, delay, jitter and so on. And the retransmitted mechanism is not suitable for video transmission. However, due to network bandwidth changing constantly and random packets lost inevitably, the current best-effort network does not offer quality of service (QoS) for real-time video transmission. The lacks of bandwidth and network congestion are the bottlenecks of video transmission. Using self-adaptive control mechanisms to achieve the desired service quality assurance and reduce and prevent network congestion becomes a research focus of streaming media transmission.By means of adjusting the transmission speed to change the bandwidth can reduce the loss of packet and solve the problem of network congestion. And adjusting the transmission speed need on the basis of the specific network condition.Gathered video data can be transmitted in the fashion of the compressed package. Video compressed in the mode of MPEG-4 contains three different types of video frame:I-frame, P-frame and B-frame. Their significances are from high to low.This paper adopts the feedback mechanism based on RTP/RTCP protocol. The sending end estimate the network condition according to the message in RTCP RR report. Then put forward a mechanism of initiative dropping frame. In the process of transmission, we can use this mechanism to abandon the partial non-essential frame according as the specific conditions of the network and protect the more important frame as I-frame and P-frame. On the account we could adjust the transmission speed to make the network load matching with available bandwidths, eliminate the adverse impact deriving from the random loss, and slow down network congestion. In addition, we deduce a modified algorithm to ensure the homogenization when the system cannot but drop some frames in the process of transmission. This method can advance the quality of playback.Experiment suggests that strategy of video self-adaptive transmission adopting the initiative dropping frame can lower packet loss rate effectively, reduce delays and jitter, and significantly improve the quality of the video received.