Initial Perceived Quality Evaluation for Video Streaming Services

Abstract

Initial perceived quality (IPQ) reflects users' perception when waiting for videos to play, which directly determines the user's willingness to go on watching videos. However, IPQ has never been carefully analyzed and evaluated. In this paper, two types of subjective test methods were elaborately designed, where the IPQ scores were obtained. The accuracy of IPQ scores was checked according to the consistency of the rating results. Then, an objective evaluation model was proposed accordingly to evaluate IPQ. Meanwhile, the relationship between IPQ and users' unacceptability of initial loading delay was also investigated according to users' abandonment behavior during the waiting period. A probability model was proposed to evaluate the unacceptability of users using logistic regression analysis. The experimental results show that the proposed models can accurately evaluate IPQ and users' unacceptability of the initial loading delay. The proposed models can be used as guidelines for service providers to effectively improve the buffering strategy for video streaming services.