Cross-User Viewport Prediction for 360-Degree Video Streaming

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

High bitrate and resolution characteristics of 360-degree virtual reality videos hinder their widespread delivery over the Internet, particularly in bandwidth-restricted networks, resulting in low perceptual quality due to inefficient viewport prediction methods.

Innovation Solution

A cross-user-based viewport prediction method and an adaptive bitrate algorithm that predicts the future viewport location and selects optimal bitrates for tiles, maximizing video quality by considering both user and other users' viewing patterns.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Manufacturing precision

If 360-degree video is delivered with high bitrate and resolution, then video quality is improved, but network bandwidth consumption increases and delivery becomes difficult in bandwidth-restricted networks

Engineering Contradiction:

Improvevideo qualityVSAvoidbandwidth consumption

Core Design Contradiction:

Manufacturing precisionVSQuantity of substance

Solution Approach 1:

The 360-degree video is divided into multiple tiles that can be independently encoded and delivered. This segmentation allows the system to allocate bandwidth selectively to different tiles based on their importance and predicted viewport relevance, rather than transmitting all tiles at high quality simultaneously.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

Different tiles are encoded at different bitrates based on their predicted importance to the user's viewport. Tiles likely to be in the viewport receive higher bitrate and resolution, while tiles unlikely to be viewed receive lower bitrate, optimizing overall quality while reducing total bandwidth consumption.

Inventive Principle:

Principle #3Local quality

2Measurement precision

If traditional viewport prediction methods are used, then system complexity is reduced, but prediction precision is insufficient leading to low perceptual quality

Engineering Contradiction:

Improveviewport prediction precisionVSAvoidprediction system complexity

Core Design Contradiction:

Measurement precisionVSDevice complexity

Solution Approach 1:

The system merges multiple prediction approaches by combining cross-user viewport probability distributions with individual user tracking data. This integration leverages collective viewing patterns from multiple users to improve prediction accuracy for any given user, achieving higher precision without requiring complex individualized models for each user.

Inventive Principle:

Principle #5Merging (Combining)

Solution Approach 2:

The system performs preliminary viewport probability distribution calculations based on cross-user data before actual video delivery. This pre-computed probability information is then used to guide bitrate allocation and tile prioritization, enabling more accurate predictions without adding complexity during real-time delivery.

Inventive Principle:

Principle #10Preliminary action

Data Source

PatentUS11310516B2Adaptive bitrate algorithm with cross-user based viewport prediction for 360-degree video streaming

Publication Date: 2022.04.19 HULU LLC

AI summary

In some embodiments, a method receives a probability distribution of a likelihood that a user might view one or more tiles in one or more segments of a video. The video is encoded in different profiles that are associated with different bitrates. An available bandwidth is determined. Then, the method selects a profile for each tile in a segment of the video based on the available bandwidth and the probability distribution for the tiles and sends a request for a respective profile for each tile for playback of the segment of the video.