QoE-Driven Joint Optimization of Segment Adaptation and Resource Allocation for DASH Clients Over LTE Networks

Abstract

In this paper, we focus on a comprehensive dynamic adaptive streaming over HTTP (DASH) framework, which consists of video content analysis at the HTTP adaptive streaming server, joint optimization of segment switch and resource allocation at eNodeB, and playback information extraction at the client, to improve the quality of experience (QoE) of the DASH clients over long-term evolution networks. We explore the unique characteristics of each segment by establishing the size-time and size-distortion models based on the video packet information extracted from the encoder rather than modeling the entire video sequence by a fixed function. Video quality, stability, and playback continuity can be predicted and reflected by these constructed models. Considering the above three factors, we develop an accumulative QoE model by calculating and predicting the QoE factors of each segment. Then, the entire framework is formulated as a mixed-integer non-linear programming mathematical model to maximize the total QoE of the DASH clients, where decisions on resource allocation and video segment switch are jointly optimized in a semi-synchronous centralized mode. Resource allocation consists of resource block assignment and modulation and coding scheme selection rather than only determining the resource utilization percentage. To perform the optimization, we develop a low-complexity heuristic algorithm that decomposes the original problem into multiple subproblems. The simulation results show that our proposed algorithms can efficiently improve the QoE level compared with other existing algorithms.