Mobile networks improve large model training with smarter data traffic shaping
Federated Learning for LLMs over Mobile Networks: Issues and Solutions in the RAN Transport
Networking and Internet ArchitectureArtificial Intelligence
Summary
Training large language models on many mobile devices creates lots of small data updates that need to be gathered efficiently. The authors explain that current mobile networks treat these updates individually, which is inefficient and causes delays. They propose a system where the network combines many updates early and manages their delivery together, making better use of high-capacity optical links. This approach can help mobile networks handle the demands of training large models while devices move and have different speeds.
What this means in practice
- •For mobile network operators: Design mobile networks that combine model updates inside the radio access network to improve training efficiency for distributed language models.
- •For optical network engineers: Create dynamically scheduled optical paths that handle the aggregated data flows from mobile devices training large models collaboratively.
A position paper. It proposes an approach and reports no results.
Authors
Emilio Paolini, Andrea Pinto, Flavio Esposito, Luca Valcarenghi
Abstract
Federated LLM fine-tuning enables large models to be adapted using private and geographically distributed data at the network edge, creating recurring and deadline-sensitive communication workloads across access and transport networks. This challenge is particularly relevant in mobile RANs, where wireless variability, mobility, and device heterogeneity cause model updates to arrive asynchronously. Although these updates belong to the same learning round and share a common destination and deadline, conventional transport networks treat them as independent device-originated flows, hiding their underlying structure and limiting the ability to efficiently provision transport resources. This mismatch is particularly problematic for optical circuit switching and all-photonics transport, which benefit from predictable and schedulable traffic demands. We argue that future RANs should act as learning-aware traffic shapers by exposing the communication structure of distributed model adaptation to the transport layer. Through in-network aggregation at the gNB, asynchronous UE updates can be transformed into fewer aggregate transfers with bounded size and delivery requirements. Once shaped in this way, federated LLM traffic becomes a suitable candidate for selectively provisioned optical connectivity, where high-capacity paths can be established during aggregate-transfer windows and released between learning rounds. The resulting architecture combines the flexibility of packet-based mobile access with dynamically provisioned optical capacity, illustrating a broader approach for coordinating distributed AI workloads across programmable access and transport networks.