Adaptive video streaming algorithm for optimized model predictive control
Adaptive bitrate techniques are used to select bitrates of segments to download when streaming video content. Different bitrate paths may be evaluated to optimize an objection function. Filtering bitrate paths that are unlikely to be optimal increases the computational speed, facilitating deployment.
1 . A method, comprising:
requesting a manifest file corresponding to a media presentation of a live event;
receiving the manifest file, wherein the manifest file comprises information identifying a set of video fragments for each segment of the media presentation, each set of the video fragments comprising different bitrate versions of the corresponding segment of the media presentation;
requesting a first video fragment for a first segment of the media presentation at a first bitrate;
receiving the first video fragment in a buffer;
determining a plurality of bitrate paths in a window of sequential segments of the media presentation starting with the first segment, wherein each bitrate path of the plurality of bitrate paths is a monotonic trajectory of bitrates for segments in the window;
determining a quality of experience (QoE) score for each bitrate path of the plurality of bitrate paths, wherein the QoE score for a bitrate path is determined from:
a utility of the bitrate path based on the bitrates for each segment in the bitrate path,
a rebuffering cost of the bitrate path based on a buffer level of the buffer after downloading segments in the bitrate path and the bitrate of segments in the bitrate path, and
a bitrate switching cost based on the bitrates of sequential segments in the bitrate path;
identifying the bitrate path of the plurality of bitrate paths having the highest QoE; and
requesting a second video fragment at a second bitrate along the identified bitrate path for a second segment of the media presentation sequential to the first segment.
2 . The method of claim 1 , wherein the window comprises two to ten segments.
3 . The method of claim 1 , wherein each QoE score is additionally based on an estimated bandwidth available to download video fragments and a predetermined bandwidth error factor.
4 . The method of claim 1 , wherein each bitrate in the plurality of bitrate paths is within a predetermined number of bitrate levels of the bitrate level of the first video fragment.
5 . A method, comprising:
receiving information identifying a set of video fragments for each segment of a media presentation, each set of the video fragments comprising different bitrate versions of the corresponding segment of the media presentation;
requesting video fragments corresponding to segments of the media presentation using an adaptive bitrate technique, wherein the adaptive bitrate technique comprises:
determining, for a segment, a quality of experience (QoE) score for each bitrate path of a plurality of bitrate paths in a window of segments, the window comprising the segment, wherein each bitrate path of the plurality of bitrate paths is a monotonic trajectory of bitrates for segments in the window,
identifying a first bitrate path of the plurality of bitrate paths as having the highest QoE,
identifying a first bitrate for the segment corresponding to the first bitrate path, and
selecting a video fragment corresponding to the segment based on the first bitrate.
6 . The method of claim 5 , wherein the information identifying the set of video fragments is part of a manifest.
7 . The method of claim 5 , wherein the video fragments represent a live event, linear playout, or video on demand (VOD).
8 . The method of claim 5 , wherein the size of the window is two to ten segments.
9 . The method of claim 5 , wherein each bitrate in the plurality of bitrate paths is within a predetermined number of bitrate levels of the bitrate level of a prior video fragment prior to the segment.
10 . The method of claim 9 , wherein the predetermined number is two to eight bitrate levels.
11 . The method of claim 5 , further comprising determining a QoE score for the first bitrate path based on:
a utility of the first bitrate path based on the bitrates for each segment in the first bitrate path,
a rebuffering cost of the first bitrate path based on a buffer level of the buffer after downloading segments in the first bitrate path and the bitrate of segments in the first bitrate path, and
a bitrate switching cost based on the bitrates of sequential segments in the first bitrate path.
12 . The method of claim 11 , wherein the QoE score is based in part on an estimated bandwidth available to download video fragments and a predetermined bandwidth error factor.
13 . A system, comprising one or more processors and one or more memories configured for:
receiving information identifying a set of video fragments for each segment of a media presentation, each set of the video fragments comprising different bitrate versions of the corresponding segment of the media presentation;
determining a plurality of bitrate paths in a window of segments, the window comprising segments of the media presentation including a first segment, wherein each bitrate path of the plurality of bitrate paths is a monotonic trajectory of bitrates for segments in the window,
determining a quality of experience (QoE) score for each bitrate path,
identifying a first bitrate path of the plurality of bitrate paths as having the highest QoE, and
requesting a first video fragment of the media presentation at a first bitrate, the first video fragment corresponding to the first segment, wherein the first bitrate corresponds to the bitrate of the first segment in the first bitrate path.
14 . The system of claim 13 , wherein the information identifying the set of video fragments is part of a manifest.
15 . The system of claim 13 , wherein the video fragments represent a live event, linear playout, or video on demand (VOD).
16 . The system of claim 13 , wherein the size of the window is two to ten segments.
17 . The system of claim 13 , wherein each bitrate in the plurality of bitrate paths is within a predetermined number of bitrate levels of the bitrate level of the video fragment prior to the segment.
18 . The system of claim 17 , wherein the predetermined number is two to eight bitrate levels.
19 . The system of claim 13 , further comprising determining a QoE score for the first bitrate path based on:
a utility of the first bitrate path based on the bitrates for each segment in the first bitrate path,
a rebuffering cost of the first bitrate path based on a buffer level of the buffer after downloading segments in the first bitrate path and the bitrate of segments in the first bitrate path, and
a bitrate switching cost based on the bitrates of sequential segments in the first bitrate path.
20 . The system of claim 19 , wherein the QoE score is based in part on a predicted bandwidth available to download video fragments and a predetermined bandwidth error factor.