Evaluating Wi-Fi Performance for VR Streaming: A Study on Realistic HEVC Video Traffic

Ferran Maura, Francesc Wilhelmi, Boris Bellalta · 2026

Read Evaluating Wi-Fi Performance for VR Streaming: A Study on Realistic HEVC Video Traffic by Ferran Maura, Francesc Wilhelmi, Boris Bellalta (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2601-16950.txt.

Introduction

Ferran Maura, Francesc Wilhelmi, and Boris Bellalta test Wi-Fi for cloud VR streams. Cloud VR sends rich shots to a head set by air. That stream needs high flow and low lag at once. One user can strain a home net. Many users can drown the air band fast. Their 2026 preprint gauges that strain with real film traffic.

The film used a top codec, HEVC, for true load shape. The team counts how many users one net can hold. Those counts guide net design for VR halls and homes. They also warn where 802.11 nets hit their wall. Such facts help plan gear for smooth VR.

Catalog record

CREATORFerran Maura, Francesc Wilhelmi, Boris Bellalta
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE22 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2601-16950

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Evaluating Wi-Fi Performance for VR Streaming: A Study on Realistic HEVC Video Traffic Thanks: This work is partially funded by Wi-XR PID2021-123995NB-I00 and TRUE-Wi-Fi PID2024-155470NB-I00 (MCIU/AEI/FEDER,UE) (MCIU/AEI/FEDER, UE), by MCIN/AEI under the Maria de Maeztu Units of Excellence Programme (CEX2021-001195-M), and by the Spanish Ministry of Science, Innovation and Universities under an FPI fellowship (PRE2022-101847). Ferran Maura, Francesc Wilhelmi, Boris Bellalta Abstract Cloud-based Virtual Reality (VR) streaming presents significant challenges for 802.11 networks due to its high throughput and low latency requirements. When multiple VR users share a Wi-Fi network, the resulting uplink and downlink traffic can quickly saturate the channel. This paper investigates the capacity of 802.11 networks for supporting realistic VR streaming workloads across varying frame rates, bitrates, codec settings, and numbers of users. We develop an emulation framework that reproduces Air Light VR (ALVR) operation, where real HEVC video traffic is fed into an 802.11 simulation model. Our findings explore Wi-Fi’s performance anomaly and demonstrate that Intra-refresh (IR) coding effectively reduces latency variability and improves QoS, supporting up to 4 concurrent VR users with Constant Bitrate (CBR) 100 Mbps before the channel is saturated. I Introduction EXtended Reality (XR) —which encompasses both Virtual Reality (VR) and Augmented Reality (AR) —is expected to be a transformative technology in the coming years. While XR is already well-established in gaming, it is poised to disrupt a wide range of emerging use cases, including immersive remote collaboration, virtual training and education, remote medical procedures, and teleoperation of industrial machinery [ 2 ] . Cloud VR streaming is a bandwidth-intensive XR application in which interactive VR content is rendered on powerful remote servers (e.g., in the cloud) and streamed to less capable end devices. These servers must provide sufficient CPU and GPU resources to render scenes at high frame rates and run the VR streaming application in real time. Specifically, the server encodes and transmits video and audio streams—at their respective frame and sampling rates—to the user’s Head Mounted Display (HMD) . Simultaneously, the HMD tracks the user’s pose and movements, sending tracking data back to the server to update the player’s state. Based on this feedback, new frames are rendered at the server and streamed back to the HMD for display [ 3 ] . Cloud rendering offers the advantage of reducing client-side hardware requirements. As long as the end-to-end latency remains sufficiently low, VR streaming enables high-fidelity graphics that would otherwise exceed the rendering capabilities of typical HMD devices. While wired connections can support such applications, wireless technologies—particularly IEEE 802.11 (commercially known as Wi-Fi) , the most widely used standard for indoor connectivity—are increasingly preferred for their support of user mobility and comfort. However, meeting the strict latency and reliability requirements of real-time VR streaming is challenging for wireless networks. Wi-Fi, in particular, faces several limitations. First, 802.11 implements random channel access through the Distributed Coordination Function (DCF) , meaning that devices must compete for access to the medium. Second, the server, Access Point (AP) and HMD stations (STAs) do not coordinate their Downlink (DL) and Uplink (UL) transmissions, which can further degrade the quality of the VR experience [ 7 ] . Finally, interference from neighboring networks operating on the same channel can introduce unpredictable disruptions, further reducing the performance of VR applications [ 8 ] . In this paper, we study the suitability of running VR streaming applications over Wi-Fi. The main novelty of this study lies in the use of real encoded video transmissions within 802.11 simulations, which is key to accurately reproducing frame sizes resulting from different codec configurations. Our main contributions are the following: • VR streaming emulation over IEEE 802.11: We develop a modular simulation framework that emulates one or multiple VR streaming sessions between a server and an HMD client over a shared Wi-Fi network. The framework includes encapsulated applications for the server and client, a video codec pipeline capable of processing 4K video at configurable frame rates and bitrates, and a queueing and transmission model representing IEEE 802.11 PHY/MAC behavior. • Joint analysis of network and codec-level dynamics: We use our integrated simulation framework to evaluate how key Quality of Service (QoS) metrics (e.g., latency, packet loss) are affected by codec parameters (e.g., frame rate, bitrate), network contention (e.g., number of CBR users), and PHY conditions (e.g., STA -to-AP distance). Our results offer insights into the interplay between video encoding strategies and wireless performance in dense VR scenarios. II Cloud-based VR streaming In this section, we introduce the architecture of cloud VR streaming, provide an overview of the video coding strategies currently used, and discuss the performance requirements for ensuring a good user experience. II-A Server-Client Architecture In VR streaming, rendering typically takes place on a server, which encodes and transmits video, audio, and haptic data to an HMD . The HMD continuously tracks the user’s pose and movements, sending tracking packets to the server. This feedback enables the server to update the user’s position, render the scene accordingly, and transmit the next frames to the HMD for display. Representative examples of this architecture include Air Light VR (ALVR) 1 1 1 https://github.com/alvr-org/ALVR and WiVRn 2 2 2 https://github.com/WiVRn/WiVRn . ALVR is an MIT-licensed application for VR streaming that acts as a wireless bridge between a SteamVR server and an HMD client. The server transmits video, audio, and haptic data, while the client sends tracking information and statistics back to the server. ALVR also supports VR -specific features such as foveated rendering—reducing resolution on image borders for more efficient encoding—and reprojection, which helps compensate for rapid head movements not yet rendered by the server [ 3 ] . II-B Video Coding for Real-Time Streaming Video traffic between server and client is typically compressed, drastically reducing bandwidth requirements compared to raw RGB video. However, compression introduces encoding and decoding latencies for each Video Frame (VF) , which contribute to increasing the end-to-end latency. Modern codecs such as H.264 and HEVC typically encode frames into a Group of Pictures (GoP) , consisting of intra-coded frames (I-frames) and predicted frames (P-frames) [ 15 ] . For real-time streaming, B-frames (bidirectionally predicted frames) are not considered, as they require access to future frames at the encoder side. HEVC’s IR mode offers improved resilience by gradually injecting I-frame coding units into successive frames. This not only enhances robustness under frame loss conditions [ 10 , 5 ] but also results in a less variable frame size distribution [ 13 ] . II-C Requirements for High QoE Due to the proximity of the HMD screens to the user’s eyes, VR streaming requires high resolutions and bitrates to ensure visual fidelity for a good Quality of Experience (QoE) . The bitrate specifies the compression level applied during encoding, and along with the frame rate (FPS), determines the average frame size and visual quality. In this study, we consider bitrates ranging from 10 to 100 Mbps. Regarding latency, we distinguish between: a) Motion-to-Photon (MTP) latency—i.e., the delay between a physical user movement and its visual update in the HMD ; and b) network latency. For (a), ITU-T recommends that MTP latency should remain below 50 ms for 360º VR experiences [ 12 ] . This constraint includes time for rendering,

Read the complete text