1. To minimize the video freeze and the latency of response, the writer proposed a layered encoding algorithm(SVC).
2. Using enhancement layers at the client allows more flexibility in the adaptation process, while the storage requirement on the server-side can be reduced, and caching performance is improved.
3. To reduce the viewport prediction error and enhance the respond time, the writer provide the base layer for all tiles in a single file so the client can fetch it in one request. As for all high quality requests, the enhancement layer is made available in a separate file to be delivered by the server.
4. When the adaptation process begins, the client fills the buffer with base layer, then it begins to predict user’s orientation, the tiles covered by the prediction results are delivered with enhancement layers.
5. By doing this, the client can buffer B(buffer size) seconds of video as opposed to the existing methods where the client can not buffer videos beyond the prediction accuracy, besides, once the bandwidth drops, the proposed method can survive by continuing playback of base quality.
6. Conclusion: the proposed method fixed the problems about reducing video freeze, reducing storage requirement and avoiding unnecessary bandwidth waste.