Research On QoE Calculation Method For Streaming Service

Streaming service is the application of streaming media technology in in-ternet information service,including video on demand,live streaming,video conference,distance eduaction,and so on.As a fudamental support for smart city,intelligent medicine,and 5G business architectures,streaming service has also driven up the growth on video communication,video consumption,and video monitoring in recent years.Until now,to match the rapid development in maket demands,streaming service is oriented from a pure technical-driven model towards the one that engages more in service,keeping up with the focus on experience and interaction.Accomplishing accurate Quality of Experience(QoE)calculation is essential to satisfy customers' needs,improve service qual-ity,and thus increase market competiveness.Therefore,providing satisfactory QoE has become the main objective of streaming service.In this dissertation,we elaborate the research on the multidata-driven QoE calculation method for streaming service by taking full advantage of user data,network data,and ser-vice data.Such a research is carried out from three aspects:QoE influences,QoE evaluation,and QoE-based service optimization.The following results briefly summarize the contribution of this dissertation:1)In terms of QoE influences for streaming service,we propose a user preference calculation method based on user's cognitive level in terms of ser-vice experience.The method consists of three steps:(a)employing user-friendly linguistic variables to collect apparent user preferences(AUP),which are converted into standardized fuzzy weights with the triangle fuzzy numbers;(b)performing the AUP defuzzification and weight adjustment according to user's cognitive levels based on historical data analysis;(c)obtaining a com-prehensive weights by intergarting the adjusted AUP weight and the extracted potential user preference(PUP)weights,and the PUP-extraction process is achieved by applying the Rough set theory.In-depth comparative experimental evaluations are performed using two real-world datasets.The results show that the proposed method effectively resolves the vagueness,inaccuracy,and in-completeness of user preference,and provides accurate user personalized in-formation for QoE evaluation.2)In terms of QoE evaluation for streaming service,we propose a multi-data-driven QoE evaluation method,and establish a public visual platform called QoECenter,which provides comprehensive support for subjective and objective combined QoE evaluation for streaming service.Specifically,QoE-Center conducts video classification in terms of the extracted spatial-temporal features,integrates techniques for dynamic adaptive streaming over HTTP(DASH),and implements DASH-based content generation.Then,QoECenter achieves the precise control of repeatable impairments during encoding and transmission,supports multiple algorithms of objective video quality evalua-tion,and provides subjective QoE evaluation based on user opinions on cus-tomized video presentation.Finally,QoECenter realizes a multidata-driven QoE evaluation throughout the streaming lifecycle.Based on widely real ex-periments conducted in QoECenter,a comprehensive QoE dataset is obtained and analyzed.The results verify the effectiveness of the proposed QoE evalua-tion method and platform.3)In terms of QoE-based service optimization,we propose a QoE-aware control method for streaming service adaption.Based on the hierarchy structure of mobile edge computing(MEC)and the file composition structure of DASH,a time-slot system with a look-ahead window is proposed to control the edge node switch and the quality adaption.Then,the local cost and the migration cost are obtained in terms of edge nodes' resource utilization and users' contex-tual information for edge node switch.Finally,based on QoE calculation,video quality adaption is conducted to match the dynamically changing network con-ditions.In-depth comparative experimental evaluations are performed using three real-world datasets.The results show that the proposed method can im-prove QoE performance and network load performance for streaming service over MEC Infrastructures.