NeSt-VR: An Adaptive Bitrate Algorithm for Virtual Reality Streaming over Wi-Fi

TL;DR

NeSt-VR is an adaptive bitrate algorithm designed for VR streaming over Wi-Fi, dynamically adjusting video quality based on real-time network metrics to ensure smooth and satisfying user experiences amidst network variability.

Contribution

This paper introduces NeSt-VR, a novel adaptive bitrate algorithm tailored for VR streaming that considers multiple network metrics for improved performance.

Findings

NeSt-VR effectively manages Wi-Fi capacity fluctuations.

It enhances VR streaming quality in both single- and multi-user scenarios.

The algorithm performs well in controlled and real-world environments.

Abstract

Real-time interactive Virtual Reality (VR) streaming is a significantly challenging use case for Wi-Fi given its high throughput and low latency requirements, especially considering the constraints imposed by the possible presence of other users and the variability of the available bandwidth. Adaptive BitRate (ABR) algorithms dynamically adjust the encoded bitrate in response to varying network conditions to maintain smooth video playback. In this paper, we present the Network-aware Step-wise ABR algorithm for VR streaming (NeSt-VR), a configurable algorithm implemented in Air Light VR (ALVR), an open-source VR streaming solution. NeSt-VR effectively adjusts video bitrate based on real-time network metrics, such as frame delivery rate, network latency, and estimated available bandwidth, to guarantee user satisfaction. These metrics are part of a comprehensive set we integrated into ALVR to characterize network performance and support the decision-making process of any ABR algorithm, validated through extensive emulated experiments. NeSt-VR is evaluated in both single- and multi-user scenarios, including tests with network capacity fluctuations, user mobility, and co-channel interference. Our results demonstrate that NeSt-VR successfully manages Wi-Fi capacity fluctuations and enhances interactive VR streaming performance in both controlled experiments at UPF's lab and professional tests at CREW's facilities.