Study On Quality Of Experience Of Network Video

With the rapid development of multimedia technology and network technology,the demand for network video is also growing. But the video quality of the network terminal does not necessarily make the customer satisfied, namely the need to improve the quality of the user experience. So the relationship between network parameters in network transmission and the quality of terminal experience has become a hot topic in academia and industry.This thesis focuses on the research and simulation of impact of network parameters and video content on the video quality, the quality of experience model for video content and model of QoE and QoS based on machine learning, using principal component analysis, regression analysis, deep learning and other methods. Evalvid and NS2 simulation tool are used to develop user experience quality evaluation research of network video quality and model construction; the following are specific research contents and results:(1)The impact of network parameters and video content on the video quality;Principal component analysis is used to analyze the network parameters. Therefore,we can get the characteristic value and contribution of first four principal components.We got the difference of video quality on different video contents in distortion network by experimental investigation.(2)Video quality prediction model based on adaptive-content; The paper presented an evaluation model considering the network parameters and video content at the same time. By analyzing the changing video frame, we represented motion coefficients, which can reflect the content of the video. Motion coefficients and network parameters are combined. Utilizing IQX hypothesis and using non-linear regression algorithm, we got an evaluation model. The assessment model has a high predictive accuracy and low computational complexity, which can be used in real-time applications. Through training and testing of the model, the results show a better performance. Relevance of predicted and actual values is above 0.90.(3)Mapping model of QoE and QoS based on the deep learning; Taking into account the QoS parameters of network layer and application layer, the paper established a model between quality of experience and quality of service by utilizing deep learning algorithm, which laid the foundation on research about influence of network parameters to video quality, quantifying the relationship between QoE and QoS and improve the quality of video experience.This paper focuses on the study of quality of experience about network videos.The author is to explore the impact of network parameters and video content on the quality of experience, and built a real-time and effective evaluation model; On this basis, the application layer parameters is added, combined with the network layer parameters, using deep learning algorithm. Through the establishment of deep belief network model, we obtained mapping model of QoE and QoS. The studies implemented in this paper can provide technical support for network performance evaluation, operation and maintenance, communication quality monitoring, media consumption rating, optimization of network video and many other fields.