Research And Implementation Of QoS Technologies In Embedded Video Communication System Based On DM642

With the rapid development of network technology and multimedia technology, IP-based audio and video network communication has been widely used. The big improvement of the digital signal processors DSPs'processing ability provides the possibility for the embedded video communication. Real-time video communication is sensitive to the QoS (Quality of Service) indicators, such as delay, delay jitter, packet loss rate and so on, however, the best-effort IP network can not provides better QoS guarantee. Therefore, how to improve QoS in IP-based video communication has become a challenging research topic.The video communication system in this paper is an embedded multimedia communication terminal based on DM642. TI's DM642 is a specialized multimedia applications oriented DSP chip. In the DSP chip, the internal bus frequency is 600MHz, the external bus frequency is 100MHz, the computational speed is 4800MIPS,and there are eight parallel computing units. The multimedia terminal could provide communication service in text, audio, video through dialing keyboard and real-time display of the QoS parameter statistics, so we can timely analysis the performance of the video service and estimate the level of network congestion.In the embedded multimedia communication terminal based on DM642, this paper deeply studies the technologies to improve the video QoS in several aspects such as flow prediction, noise processing, error control, congestion control and so on, which provide some QoS guarantee when the quality of the network declines.In order to improve QoS of the video in the network with fluctuating bandwidth, proper use and optimization of network bandwidth resources is a very important issue. To research a reasonable forecasting model to predict video traffic is helpful to improve network utilization and video service quality to some extent. Considering video communication systems and comparing several common video traffic predicting models such as the auto-regressive model (AR), moving average model (MA), autoregressive moving average model (ARMA), the gray model, wavelet decomposition model and neural network model, this paper uses the gray model to predict video flow. The experimental results show that the prediction is accurate.In the video communication system, we can alleviate network congestion by sampling the video image to reduce image resolution. However, the aliasing noise is introduced in the down-sampling process, which seriously affects the video quality. This paper adopts frequency domain and spatial domain processing methods to process the video image before down sampling. The experimental results show that the two methods can both reduce the aliasing noise in the video image. The spatial domain processing method is better in real-time and effect.In order to reduce the delay jitter of the video, This paper develops a receiver cache mechanism whose structure is a linear queue. It ensures the stability and continuity of the playing video in the receiver and provides a basis for the retransmission mechanism. The experimental results show that the delay jitter is smaller and more stable after using the cache mechanism which guarantees the video stream to play steadily, continuously and smoothly.The fluctuation and suddenness of the IP network bandwidth introduce the packet loss rate in the video transmission. In order to reduce the packet loss rate on the impact of the video quality, this paper proposes an I-frame retransmission mechanism. It only retransmits the I frames which contain key information and has a smaller number, so it not only improves the video quality, but also reduces the retransmission overhead. The experimental results show that the method can improve the decoding accuracy and improve the picture quality.In order to obtain the QoS information of the real-time video traffic, this paper develops a method of collecting the video service statistical parameters to analyze the video performance and the level of network congestion. The statistical parameters include format, frame rate, flow, predict flow, packet loss rate, delay jitter and number of the retransmitted frames.Generally speaking, network congestion affects the video QoS. Therefore, it is an effective strategy of guaranteeing QoS to develop a reasonable mechanism for congestion control. At present, there are two congestion control solutions: one is a network-based method by increasing the network bandwidth; the other is a terminal-based method by controlling rate. The terminal-based method is divided into a window-based method and a rate-based method. This paper adopts a threshold-based sending rate congestion control method. After sending a RTP packet with sequence number and time stamp in the sender, the receiver calculates the QoS statistics such as delay jitter and packet loss rate according to the sequence and time stamp when receiving the RTP packet, and send a RTCP feedback packet to the sender periodically. The sender compares the feedback loss rate with the threshold to analyze the level of network congestion. By adjusting frame rate, compression rate and resolution rate to change send rate, a closed-loop adaptive congestion control mechanism is achieved. The experimental results show that the video packet loss rate decreases obviously and the send rate is more adapted to the bandwidth fluctuation after using the method.In conclusion, this paper mainly studies the QoS technologies in the video communication system, develops and implements several measures to improve the video QoS in the multimedia terminal based on DM642. Firstly, it develops a method to predict video flow and proposes two methods to process the aliasing noise, then realizes the video cache mechanism and I-frame retransmission mechanism, finally discusses the principles and methods of congestion control in detail. Considering the multimedia terminal, this paper proposes a closed-loop feedback RTCP-based adaptive congestion control mechanism by adjusting frame rate, compression rate and resolution rate according to the level of network congestion to change send rate to adapt the variation of network bandwidth. This paper dose some experiments about these QoS technologies in the embedded multimedia terminal of these, the results show that there is some effectiveness and practicality. At present, there are still some imperfections in the system, the author will be further his study in future work.