Research On Quality Of Service For Enterprise Video Conference

Video conference can provide real-time multimedia interactive environment for people in distance, making remote communication more efficient, and thus being widely used in business, education, government management, etc.Video conference of the next generation uses Internet as transfer platform, and needs real-time data transfer for high definition video, audio and text. This needs the transfer platform can provide higher network bandwidth at lower loss rate and with low delay. However, the Internet fails to provide this kind of guarantee ability, which poses the challenge of QoS for video conference of the new generation.This paper focuses on video conference of new generation, under the condition of enterprise network communication environment. The main research content and contributions of this thesis are as follows:1. Propose a system framework for guarantee QoS of Enterprise Video Conferences(EVCQoS), construct its formalized architecture, and analyze the connectivity and behavioral characteristics of the model mathematically. The model is used as theoretical basis for the solution to network-based QoS and rate adaptation.2. Construct a network-based mechanism of QoS guarantee for video conference. According to the mechanism, a resource matrix(resource model) is built to organize conferences and network resources in order to meet unicast or multicast requirements proposed by users. In order to guarantee the QoS of video stream, the network resource of enterprise network is monitored and controlled by the system. At the edges of network, the entrance of legal flow is controlled with the switch that supports the SAVI technology, at the core of network, resource reservation method is used to allocate the network resource, therefore guaranteeing the QoS of video stream.3. When the transfer network fails to provide enough network resource, application-based QoS mechanism is used. Aimed to solve this problem, this study proposes a QoE-based rate adaptation algorithm. H.264 SVC is used as the encoder because it can provide three dimension scalability: resolution, frame rate and video quality. For given bandwidth, PSQA evaluation algorithm is used to evaluate the QoE of video, based on the network resource constraint, and the optimal parameters are selected for encoding.4. Video conference QoS mechanism uses ANN-based QoE model to optimize the video coding, and the model is also used to monitor the performance of video conference, since the ANN-based QoE model needs a large mount of subjective test data, and it is expensive to build QoE model. This paper proposes an adapted method for QoE model, which performs a linear transformation to the input vector before it is the input of the ANN network. This method can produce a model for the new condition with small amount of training data.Based on the work above, we also implement the EVCQoS. Experiments suggest that the EVCQo S guarantee system can ensure the transmission quality and application service of enterprise video conferencing on CERNET2. To be more specific, two complementary technologies enable the QoE-based bit rate adaptation techniques to ensure QoS for video conferences, despite of the insufficient network resources which may induce failures of network QoS guarantee.