PRECEPT: Power-efficient Field-of-View Prediction for VR Video Streaming

Ahmed Abdelreheem; Darijo Raca; Ahmed H.  Zahran

PRECEPT: Power-efficient Field-of-View Prediction for VR Video Streaming

School of Computer Science and Information Technology

University of Sarajevo
University College Cork

Research output: Contribution to conference › Paper › peer-review

Abstract

Field-of-view (FoV) prediction is critical for reducing device energy consumption and enhancing user quality of experience (QoE) in immersive streaming. To address the high computational and energy costs of standard DL-based FoV prediction, we propose PRECEPT, an energy-efficient, system-oriented two-stage framework. PRECEPT splits the prediction pipeline by adding a lightweight, CPU-based classifier to identify tile change. PRECEPT's classifier successfully filters approximately 80% of "no-change" events. PRECEPT activates the resource-intensive DL model only during identified tile change. This design reduces the average inference delay and energy consumption by up to 69% in a real mobile deployment. PRECEPT's two-stage design enables sustainable, high-performance FoV prediction on resource-constrained devices.

Original language	English (Ireland)
Number of pages	6
Publication status	Accepted/In press - Jun 2026
Event	The IEEE 5th International Conference on Intelligent Reality - Pisa, Italy Duration: 25 Jun 2026 → 26 Jun 2026 https://icir.ieee.org/

Conference

Conference	The IEEE 5th International Conference on Intelligent Reality
Abbreviated title	IEEE ICIR 2026
Country/Territory	Italy
City	Pisa
Period	25/06/26 → 26/06/26
Internet address	https://icir.ieee.org/

UN SDGs

This output contributes to the following UN Sustainable Development Goals (SDGs)

SDG 9 Industry, Innovation, and Infrastructure
SDG 13 Climate Action

UCC Futures

Future of Networks, Systems & Cybersecurity (NASC)

Keywords

Field-of-view (FoV) prediction
[ComputerScience]

Access to Document

2026_ICIR_PRECEPT_FoV_PredictionAccepted author manuscript, 1.12 MB

Cite this

@conference{3d7601894cde4462ba3d478649f8c010,

title = "PRECEPT: Power-efficient Field-of-View Prediction for VR Video Streaming",

abstract = "Field-of-view (FoV) prediction is critical for reducing device energy consumption and enhancing user quality of experience (QoE) in immersive streaming. To address the high computational and energy costs of standard DL-based FoV prediction, we propose PRECEPT, an energy-efficient, system-oriented two-stage framework. PRECEPT splits the prediction pipeline by adding a lightweight, CPU-based classifier to identify tile change. PRECEPT's classifier successfully filters approximately 80\% of {"}no-change{"} events. PRECEPT activates the resource-intensive DL model only during identified tile change. This design reduces the average inference delay and energy consumption by up to 69\% in a real mobile deployment. PRECEPT's two-stage design enables sustainable, high-performance FoV prediction on resource-constrained devices.",

keywords = "Field-of-view (FoV) prediction, [ComputerScience]",

author = "Ahmed Abdelreheem and Darijo Raca and Zahran, \{Ahmed H.\}",

note = "{\textcopyright} 2026, the Authors.; The IEEE 5th International Conference on Intelligent Reality, IEEE ICIR 2026 ; Conference date: 25-06-2026 Through 26-06-2026",

year = "2026",

month = jun,

language = "English (Ireland)",

url = "https://icir.ieee.org/",

}

TY - CONF

T1 - PRECEPT: Power-efficient Field-of-View Prediction for VR Video Streaming

AU - Abdelreheem, Ahmed

AU - Raca, Darijo

AU - Zahran, Ahmed H.

PY - 2026/6

Y1 - 2026/6

N2 - Field-of-view (FoV) prediction is critical for reducing device energy consumption and enhancing user quality of experience (QoE) in immersive streaming. To address the high computational and energy costs of standard DL-based FoV prediction, we propose PRECEPT, an energy-efficient, system-oriented two-stage framework. PRECEPT splits the prediction pipeline by adding a lightweight, CPU-based classifier to identify tile change. PRECEPT's classifier successfully filters approximately 80% of "no-change" events. PRECEPT activates the resource-intensive DL model only during identified tile change. This design reduces the average inference delay and energy consumption by up to 69% in a real mobile deployment. PRECEPT's two-stage design enables sustainable, high-performance FoV prediction on resource-constrained devices.

AB - Field-of-view (FoV) prediction is critical for reducing device energy consumption and enhancing user quality of experience (QoE) in immersive streaming. To address the high computational and energy costs of standard DL-based FoV prediction, we propose PRECEPT, an energy-efficient, system-oriented two-stage framework. PRECEPT splits the prediction pipeline by adding a lightweight, CPU-based classifier to identify tile change. PRECEPT's classifier successfully filters approximately 80% of "no-change" events. PRECEPT activates the resource-intensive DL model only during identified tile change. This design reduces the average inference delay and energy consumption by up to 69% in a real mobile deployment. PRECEPT's two-stage design enables sustainable, high-performance FoV prediction on resource-constrained devices.

KW - Field-of-view (FoV) prediction

KW - [ComputerScience]

M3 - Paper

T2 - The IEEE 5th International Conference on Intelligent Reality

Y2 - 25 June 2026 through 26 June 2026

ER -

PRECEPT: Power-efficient Field-of-View Prediction for VR Video Streaming

Abstract

Conference

UN SDGs

UCC Futures

Keywords

Access to Document

Fingerprint

Cite this