360 Video Streaming Adaptation via RTT and Bandwidth Prefetching
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing 360-degree video streaming technologies face challenges in managing high bandwidth demands, leading to video freezes and degraded user experience due to unpredictable viewport changes and network limitations, especially under challenging conditions.
Innovation Solution
A method that measures network round trip time (RTT) and bandwidth to determine a future Field of View (FoV) and prefetches video streams, adjusting FoV size and data based on these measurements to ensure smooth playback without significant increases in network bandwidth.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of energy
If high-quality video is streamed only for the user's viewport to reduce bandwidth consumption, then bandwidth usage is reduced, but playback becomes prone to video freezes due to rebuffering when viewport changes are unpredictable
Solution Approach 1:
The system performs preliminary actions by predicting future viewport positions and prefetching video data for those predicted viewports before they are actually requested. This allows the video player to have buffered data ready when the user actually looks in that direction, preventing rebuffering freezes while avoiding the need to stream high-quality video for all possible viewports simultaneously.
Solution Approach 2:
The system dynamically adapts the prefetching strategy based on real-time analysis of user head movement patterns, network conditions, and playback buffer status. The prediction model continuously learns and adjusts to individual user behavior, making the prefetching both bandwidth-efficient and reliable for maintaining continuous playback.
2Duration of action of stationary object
If the buffering duration is extended beyond the viewport prediction interval, then future video chunks can be buffered longer, but the system complexity increases
Solution Approach 1:
The system extends buffering by performing preliminary prefetching actions for multiple future video chunks beyond the immediate next viewport. By predicting a sequence of future viewports and prefetching data for all of them in advance, the system creates a longer effective buffering duration without requiring complex manual buffer management.
Solution Approach 2:
The system uses feedback from actual user viewport transitions to continuously refine its prediction accuracy. By monitoring whether predicted viewports matched actual user behavior, the system adjusts its prediction model to improve future prefetching decisions, effectively managing buffer duration through learned patterns rather than fixed complex rules.
3Loss of energy
If viewport prediction is used to prefetch video data, then bandwidth consumption is reduced, but video quality may degrade when predictions are inaccurate
Solution Approach 1:
The system applies partial prefetching by selectively prefetching video data only for the most likely future viewports based on prediction confidence levels. For highly predictable viewport transitions, it prefetches high-quality data, while for less certain predictions, it may prefetch lower priority or use adaptive quality levels, thus balancing bandwidth usage with maintaining video quality where it matters most.
Solution Approach 2:
The system dynamically changes video quality parameters based on prediction accuracy and network conditions. When predictions are highly accurate and network bandwidth is sufficient, it streams higher quality video for predicted viewports. When prediction confidence is lower or bandwidth is constrained, it adjusts quality parameters to maintain acceptable viewing experience while preserving bandwidth for more certain predictions.
Data Source
AI summary
The disclosure relates to technology for a 360 degree video stream in a network. A network round trip time (RTT) is measured for a requested FoV of the 360 degree video stream, where the RTT indicates a time interval between sending a FoV interest message and receiving a response message in reply to the requested FoV. The network bandwidth is measured for the requested FoV of the 360 degree video stream, where the measured network bandwidth represents an amount of data the network transmits in the response message. A future FoV of the 360 degree video stream is then determined based on the requested FoV, the measured network bandwidth and the measured network RTT, and a video stream is prefecthced for the future FoV of the 360 degree video stream.