Quality Of Experience Evaluation For Video Streaming Service

With the rapid growth of network services, it is urgent to focus on user-oriented evaluation for the service quality in order to study how to provide higher quality and ensure the users’approval. Quality of experience (QoE), which extends the concept of traditional Quality of service (QoS) and directly reflects subjective perception of users, is the users’recognition degree for services as the standard to evaluate services. As a result, QoE evaluation becomes a research hotspot in the field of network management.In this paper, the research is focused on QoE evaluation for the video services. Firstly, in order to evaluate QoE more accurately and effectively, video streamings are classified into different types before QoE evaluation. Then QoE assessment is specific to each video content type. Based on analysis of the typical characteristics of video content and consideration of the users’perception for the video scene features, this paper proposes the user-oriented video feature information extraction mechanism. The authors employ fast, accurate and effective video content clustering algorithm to establish user-oriented video content type classification mechanism.On the basis of video content classification, this paper analyzes the typical features of video streaming over UDP and objective parameters affecting user perception of video quality. Then high level parameters are extracted from three layers, including network layer, application layer and bit-stream layer, to build hierarchical objective performance index system. We design Back Propagation Neural Net (BPNN) to predict QoE quantization value (MOS) from the hierarchical objective performance metrics. The performance evaluation demonstrates a high accuracy in real time monitoring video streaming over UDP. While QoE evaluation of video streaming over UDP achieves high precision and efficiency, the authors distinguish the key differences between TCP and UDP transmitted video streaming and present novel application-layer objective metrics influencing user perceived quality of video over TCP. Then Back Propagation Neural Net is utilized to model the correlation between the selected objective parameters and QoE. The experimental study shows that the proposed method performs high accuracy in assessing QoE of video streaming over TCP.In conclusion, this paper evaluates QoE of video streaming over UDP and TCP with a certain degree of real-time and high accuracy in consideration of reducing time overhead and computational expense. Finally the authors achieve the expected goal of the researchs.