Predicting the performance of virtual reality video streaming in mobile networks

Abstract

The demand of Virtual Reality (VR) video streaming to mobile devices is booming, as VR becomes accessible to the general public. However, the variability of conditions of mobile networks affects the perception of this type of high-bandwidth-demanding services in unexpected ways. In this situation, there is a need for novel performance assessment models fit to the new VR applications. In this paper, we present PERCEIVE, a two-stage method for predicting the perceived quality of adaptive VR videos when streamed through mobile networks. By means of machine learning techniques, our approach is able to first predict adaptive VR video playout performance, using network Quality of Service (QoS) indicators as predictors. In a second stage, it employs the predicted VR video playout performance metrics to model and estimate end-user perceived quality. The evaluation of PERCEIVE has been performed considering a real-world environment, in which VR videos are streamed while subjected to LTE/4G network condition. The accuracy of PERCEIVE has been assessed by means of the residual error between predicted and measured values. Our approach predicts the different performance metrics of the VR playout with an average prediction error lower than 3.7% and estimates the perceived quality with a prediction error lower than 4% for over 90% of all the tested cases. Moreover, it allows us to pinpoint the QoS conditions that affect adaptive VR streaming services the most.

Effects of Adaptive Streaming Optimizations on the Perception of 360° Virtual Reality Video

“…As for viewport prediction, there is no generic solution on how to select the qualities of the different tiles. In our approach, we select the quality-zones approach presented in [2]. The tiles are distributed in zones according to their distance to the viewport, in such a way that only the tiles within the viewport are selected at the highest quality, while the edges are at the second highest quality, and so on.…”

“…This action is performed for each of the quality levels wanted. Finally, the MP4Box is employed to pack the HEVC streams into mp4 containers and to prepare the videos to be streamed using HAS [2]. During the streaming, the server predicts the viewport position for the next segment (based on the received position from the client, using a UDP channel) and assigns the qualities to the tiles.…”

“…The demand of streaming of Virtual Reality (VR) video to mobile devices is booming as Head Mounted Devices (HMDs) increase their penetration rate, with more effective and affordable solutions. Examples of such are the Google's Cardboard 1 or the Samsung Gear VR 2 . In fact, the VR video streaming traffic is expected to increase 20-fold by 2021 [1].…”

“…A VR video contains a full 360 • panoramic view [2]. Thus, its streaming demands significantly higher bandwidths than traditional 2D video applications, even if only a fraction of the spherical view, namely the viewport (or Field of View-FoV), is visible at any given instant.…”

“…For example, considering that only the predicted viewport's tiles are downloaded in advance, predictions errors may lead to application level degradation (i.e., video stall, quality switch) and, consequently, QoE degradation. However, the degree in which each of the optimizations might influence the QoE is still an open topic of research and discussion [2].…”