The current state-of-the-art solutions try to predict the user viewport for a short duration in the future and request the segments that provide this viewport at higher quality. To minimize the occurrence of such mismatches, the video is not buffered for durations longer than the duration for which the viewport can be predicted accurately, thus increasing the probability of video freeze in the event of network bandwidth fluctuation. This paper uses SVC in adaptive 360-degree video streaming which can mitigate the problem.
This paper discuss several projection methods and it concludes that CMP provides smooth quality degradation with lower processing complexity than pyramid and lower bitrate overhead than ERP. Therefore, this paper uses CMP as the video projection method.
In this method, each segment contains six faces of the cube. Each face of the cube is sliced into m x m (2 x 2) tiles. Each tile is encoded using SVC.
- The client always requests the base layer for all tiles to ensure that all viewports are available for viewing if the user’s viewport changes quickly.
- Depending on the availability of bandwidth, enhancement layers can be requested for the tiles where the user’s viewport may shift.
- The range of the surrounding tiles to request depends on the accuracy of viewport prediction.
By using SVC, it:
- provides longer buffered video which can avoid playback freeze.
- reduces the storage requirement for 360 video. (Remove the redundancy between different levels)
- provides better use of in-network caching capabilities.