Multi-Task Learning as enabler for General-Purpose AI-native RAN
Hasan Farooq, Julien Forgeat, Shruti Bothe, Kristijonas Cyras, Md Moin
TL;DR
The paper addresses the challenge of enabling a general-purpose AI-native RAN by studying multi-task learning (MTL) across four edge-distributed tasks. It analyzes three MTL architectures (MMoE, DSelect-k, CGC) with various loss and gradient balancing strategies under local, global, and partial federated learning topologies. Key findings show CGC with uncertainty-based weighting often performs best or on par with single-task learning, partial federation generally outperforms full federation, and data sparsity shifts the advantage toward global models for some tasks while LoS classification presents inter-task trade-offs. These results provide design guidance for scalable, resource-efficient AI-native RAN deployments at the network edge, informing architecture choice, task grouping, and edge federation strategies.
Abstract
The realization of data-driven AI-native architecture envisioned for 6G and beyond networks can eventually lead to multiple machine learning (ML) workloads distributed at the network edges driving downstream tasks like secondary carrier prediction, positioning, channel prediction etc. The independent life-cycle management of these edge-distributed independent multiple workloads sharing a resource-constrained compute node e.g., base station (BS) is a challenge that will scale with denser deployments. This study explores the effectiveness of multi-task learning (MTL) approaches in facilitating a general-purpose AI native Radio Access Network (RAN). The investigation focuses on four RAN tasks: (i) secondary carrier prediction, (ii) user location prediction, (iii) indoor link classification, and (iv) line-of-sight link classification. We validate the performance using realistic simulations considering multi-faceted design aspects of MTL including model architecture, loss and gradient balancing strategies, distributed learning topology, data sparsity and task groupings. The quantification and insights from simulations reveal that for the four RAN tasks considered (i) adoption of customized gate control-based expert architecture with uncertainty-based weighting makes MTL perform either best among all or at par with single task learning (STL) (ii) LoS classification task in MTL setting helps other tasks but its own performance is degraded (iii) for sparse training data, training a single global MTL model is helpful but MTL performance is on par with STL (iv) optimal set of group pairing exists for each task and (v) partial federation is much better than full model federation in MTL setting.