![[2602.16174] Edge Learning via Federated Split Decision Transformers for Metaverse Resource Allocation](https://arxiv.org/static/browse/0.3.4/images/arxiv-logo-fb.png){kind=logo}
[2602.16174] Edge Learning via Federated Split Decision Transformers for Metaverse Resource Allocation
Summary
The paper presents Federated Split Decision Transformers (FSDT) for optimizing resource allocation in mobile edge computing for the metaverse, enhancing user experience while minimizing computational load.
Why It Matters
As the metaverse expands, efficient resource allocation becomes critical for maintaining quality user experiences. This research addresses performance challenges in federated learning, proposing a novel framework that balances local adaptability and global cooperation, which is essential for future developments in edge computing and virtual reality.
Key Takeaways
- FSDT improves quality of experience (QoE) by up to 10% in heterogeneous environments.
- The framework offloads nearly 98% of transformer model parameters to the cloud, reducing MEC server burden.
- Combines federated learning with reinforcement learning for enhanced resource allocation strategies.
- Addresses challenges of conventional federated learning, such as performance degradation in multi-radio environments.
- Promotes cooperative training across MEC servers for better adaptability.
Computer Science > Networking and Internet Architecture arXiv:2602.16174 (cs) [Submitted on 18 Feb 2026] Title:Edge Learning via Federated Split Decision Transformers for Metaverse Resource Allocation Authors:Fatih Temiz, Shavbo Salehi, Melike Erol-Kantarci View a PDF of the paper titled Edge Learning via Federated Split Decision Transformers for Metaverse Resource Allocation, by Fatih Temiz and 2 other authors View PDF HTML (experimental) Abstract:Mobile edge computing (MEC) based wireless metaverse services offer an untethered, immersive experience to users, where the superior quality of experience (QoE) needs to be achieved under stringent latency constraints and visual quality demands. To achieve this, MEC-based intelligent resource allocation for virtual reality users needs to be supported by coordination across MEC servers to harness distributed data. Federated learning (FL) is a promising solution, and can be combined with reinforcement learning (RL) to develop generalized policies across MEC-servers. However, conventional FL incurs transmitting the full model parameters across the MEC-servers and the cloud, and suffer performance degradation due to naive global aggregation, especially in heterogeneous multi-radio access technology environments. To address these challenges, this paper proposes Federated Split Decision Transformer (FSDT), an offline RL framework where the transformer model is partitioned between MEC servers and the cloud. Agent-specific components (e...
