QoE-driven Delivery Optimization Methods For Dynamic Adaptive Streaming Over HTTP

The increasing demand for video services has driven the continuous evolution of streaming media technologies.For video services,viewers' satisfaction is a key performance indicator.Therefore,achieving optimal adaptive streaming media transmission under limited network resources to improve users' Quality of Experience(QoE)is a common concern of the industry and academia.The HTTP dynamic adaptive streaming media technology has the advantages of good compatibility,strong scalability,and ease of deployment.It can also adaptively adjust the bit rate according to the network fluctuation characteristics,enhance the bandwidth utilization,and improve the user experience quality.Therefore,HTTP dynamic adaptive streaming media has been widely concerned and applied.We study the QoE-based dynamic adaptive streaming media transmission optimization method,which is of important research significance and application value.Since the existing QoE utility model cannot fully reflect the quality of adaptive streaming media,we define a novel QoE utility model based on video segment information by comprehensively considering the segment encoding quality,bitrate switching and playback interruptions.The defined model can effectively describe segment quality fluctuations during video playback.Based on the defined QoE utility model,we propose a QoE-based rate adaptation scheme for HTTP adaptive streaming.We take the QoE as the optimization object,use the Markov network model and future segment size to accurately predict the future segment transmission,and obtain optimal bitrate decision based on long-term user experience.Simulation results show that the proposed QoE-based rate adaptation method can adaptively adjust the segment rate according to the network fluctuations,achieving good bandwidth adaptability and video smoothness as well as improving the user's QoE.In view of the fact that the current rate adaptation methods for HTTP live streaming do not consider the impact of time-varying factors in the network on the performance of live video transmission,we study the QoE-based rate adaptation method for HTTP live streaming.We define a new live streaming transmission model after analyzing the influence of the segment duration and live latency on the characteristics of real-time streaming transmission.Then,a QoE-based rate adaptation method for HTTP live streaming is proposed using QoE utility model defined above and the transmission characteristics.Analytical and simulation results show that the proposed rate adaptation method can adapt to live streaming transmission under different live latency and segment durations,effectively improve the bandwidth utilization rate and ensure the high quality and smoothness of live video at the same time.Since there are drawbacks such as the over idealized network scenario and the ill-suited QoE utility function in current cross-layer video delivery optimization schemes,we investigate the adaptive streaming delivery optimization in the cognitive radio network under imperfect channel conditions.Based on the defined QoE utility model,we set the video clients' quality of experience as the optimization objective.A cross-layer video delivery optimization method is proposed by jointly considering the resource allocation problem under inaccurate channel estimation and the rate adaptation problem for video users.Analytical and simulation results show that the proposed cross-layer delivery optimization method can effectively mitigate the negative influence of channel uncertainty on transmission performance,and obtain improved transmission performance at the lowest communication cost between the user and the base station.The telecom operators and content providers can adaptively adjust the transmission scheme to achieve the fairness of the system level and the user's viewing experience according to the proposed QoE-driven delivery optimization methods for adaptive streaming and the characteristics of the wireless network.Therefore,the methods proposed in this paper is meaningful for improving the performance of adaptive streaming services.