Attention-aware Resource Allocation Framework Enables Scalable VR-Cloud Gaming with Low Latency in 6G Networks
Virtual Reality Cloud Gaming presents a significant challenge for future mobile networks, demanding both substantial bandwidth and minimal delay to deliver truly immersive experiences. Gabriel Almeida, João Paulo Esper, and Cleverson Nahum, from the Instituto de Informática at Universidade Federal de Goiás, alongside Audebaro Klautau from Universidade Federal do Pará and Kleber Vieira Cardoso, address this challenge with a new framework for managing network resources in 6G systems. Their research introduces a multi-stage optimisation approach that intelligently allocates resources, places game engines strategically, and prioritises visual quality based on where a user is looking within the virtual environment. This innovative method demonstrably improves user experience by up to 50%, reduces communication resource usage by 75%, and delivers substantial cost savings, all while operating quickly enough for real-time application in next-generation networks.
Network Slicing for XR and Gaming
This document comprehensively explores research directions and technical specifications for next-generation wireless communication networks, including 5G, 6G and beyond, with a strong focus on cloud gaming and extended reality (XR) applications. It investigates how to optimise network performance, resource allocation, and quality of experience (QoE) for these demanding services, encompassing efficient network slicing, multi-connectivity, edge computing, and exploration of future wireless technologies. The research encompasses 5G NR, enhancements in 5G Advanced, and early investigations into 6G technologies, detailing concepts like RAN virtualisation, network slicing, multi-access edge computing, and virtual network functions. Scientists are developing methods for functional split, multi-connectivity utilising various radio access technologies, accurate channel modelling, and sophisticated user association techniques, all within the context of XR applications and popular gaming platforms.
The document highlights key research areas and challenges, including optimal resource allocation for cloud gaming and XR, strategic placement of edge computing resources, network slicing for diverse applications, and effective management of multi-connectivity. Scientists are also addressing privacy and security concerns in edge computing environments, ensuring seamless connectivity during user movement, developing accurate channel models for high-frequency bands, and ensuring fairness in resource allocation. Optimising functional split and RAN function placement, alongside guaranteeing service level agreements, are also central to the research. The study decomposes the complex joint resource allocation problem into three interdependent stages: user association and communication resource allocation, VR-CG game engine placement with adaptive multipath routing, and attention-aware scheduling with wireless resource allocation based on motion-to-latency. Specialized heuristic algorithms were designed for each stage, achieving near-optimal performance while significantly reducing processing time, a critical factor for real-time immersive experiences. The research team developed a novel user-centric Quality of Experience (QoE) model grounded in visual attention to virtual objects, guiding adaptive selection of resolution and frame rate.
This approach enables prioritisation of video quality in areas where the user is focused, while intelligently compressing peripheral regions, effectively reducing transmission requirements and optimising bandwidth usage. Experiments employing a dataset-driven evaluation demonstrate a substantial improvement in QoE, up to 50%, alongside a 75% reduction in communication resource usage and up to 35% cost savings, all while maintaining an average optimality gap of only 5%. To address scalability, scientists formulated the VR-CG resource allocation problem as a complex optimisation challenge, proving it to be NP-hard. The team then developed a heuristic solution designed to solve large-scale scenarios in under 0. 1 seconds, highlighting its potential for real-time deployment in next-generation mobile networks. The work decomposes the complex problem into three interconnected stages: user association and communication resource allocation, VR-CG game engine placement with adaptive routing, and attention-aware scheduling with wireless resource allocation based on motion-to-latency. Specialized heuristic algorithms were designed for each stage, achieving performance comparable to optimal solutions while solving large-scale scenarios in under 0. 1 seconds.
Experiments demonstrate that this framework improves Quality of Experience (QoE) by up to 50% compared to existing approaches, reducing communication resource usage by 75% and achieving up to 35% cost savings, while maintaining an average optimality gap of only 5%. This breakthrough relies on a new user-centric QoE model based on visual attention to virtual objects, guiding adaptive selection of resolution and frame rate. Building on this foundation, scientists explored semantic video transmission techniques, developing an object-aware encoding mechanism that combines scene information with user eye-tracking data. This approach prioritises video quality in the user’s focus area while compressing peripheral regions, effectively reducing transmission requirements. The team developed a multi-stage approach, encompassing user association, game engine placement, and adaptive scheduling, to address the demanding requirements of immersive experiences. Results demonstrate significant improvements in quality of experience, achieving up to a 50% increase compared to existing methods, reducing communication resource usage by as much as 75% and delivering up to 35% cost savings, while maintaining a small optimality gap of 5%. The developed heuristics solve large-scale scenarios rapidly, completing calculations in under 0. 1 seconds and highlighting their potential for real-time deployment. The authors acknowledge that further refinement is possible, and future work will explore the integration of artificial intelligence and machine learning techniques to enhance adaptability, validate their quality of experience model through user studies, and investigate the use of physiological signals to create a more personalised assessment of immersive experience quality.
👉 More information
🗞 Toward Scalable VR-Cloud Gaming: An Attention-aware Adaptive Resource Allocation Framework for 6G Networks
🧠 ArXiv: https://arxiv.org/abs/2512.11667
