A Study on Quality-of-Experience Index for OTT Video Service and Its Applications

• Main Authors: DU, QIAN-YU, 杜千煜
• Other Authors: CHEN, CHING-YUNG
• Format: Others
• Language: zh-TW
• Published: 2018
• Online Access: http://ndltd.ncl.edu.tw/handle/s964n6

碩士 === 國立高雄第一科技大學 === 電腦與通訊工程系碩士班 === 106 === The emerging over-the-top (OTT) video service has become the most popular and bandwidth-consuming Internet application. For this kind of service, adaptive HTTP video streaming is widely deployed along with the content delivery network (CDN) for delivering multimedia resources to customers. Conventional measurement of network performance (such as throughput, delay and jitter, etc.) fails to give a judgement of the user experience which is also affected by the video quality, player behavior and the display size. There are needs of quality measurement, especially the quality of experience (QoE), for service providers, network operators and even customers to evaluate and improve the system performance as well as the user satisfaction. To investigate QoE measurement, an objective QoE model, the video mean opinion score (vMOS), consisting of application-level performance metrics (APMs) is presented in this paper. In this model, the video resolution, display size, initial buffering time and stalling events are taken into account with individual weightings to make a quantitative measure (between 1 to 5) of the user experience. Subjective experiments are performed to estimate the optimum weightings for each APM. The proposed vMOS is then used to implement an App on the Android platform as an evaluation tool. Some experiments are performed using the developed App for evaluating the QoE of watching YouTube clips, and we observe that the vMOS reports similar results comparing to subjective test evaluated by human observers in most of the cases. Therefore, the proposed vMOS and the associated App can be adopted to automatically assess the QoE of OTT video service, replacing the expensive and time-consuming subjective tests.