Expose learning intent to the transport network

Determine what learning-related information a radio access network should expose to the optical transport layer, including whether source, destination, size, readiness time, and delivery deadline are sufficient and how transport should adapt when wireless variability prevents a planned transmission window from being met.

Background

The paper proposes that a learning-aware RAN expose federated-learning aggregates to the transport network as bounded, deadline-constrained transfers. Although the transport layer needs information such as the aggregate source, destination, volume, readiness time, and latest useful delivery time, the paper leaves unresolved whether this metadata is sufficient for effective coordination or whether interfaces must also convey learning priorities, aggregation policies, and tolerance for delayed updates. It also leaves open how transport control should respond when wireless variability disrupts the expected transmission window.

References

Several research questions remain before this vision can become part of future AI-native RANs.

RQ1: How should learning intent be exposed to the transport network? The optical layer does not need to understand the details of the learning algorithm, but it does require a compact description of each aggregate transfer. What information should the RAN expose? Is the source, destination, size, readiness time, and delivery deadline sufficient, or should future interfaces also communicate learning priorities, aggregation policies, or tolerance to delayed updates? Moreover, how should transport adapt when wireless variability prevents the planned transmission window from being met?

— Federated Learning for LLMs over Mobile Networks: Issues and Solutions in the RAN Transport  (2610.01304 - Paolini et al., 1 Oct 2026) in Section 6, “Toward Deployable AI-Native RANs,” RQ1