aiqbench analysis
This research introduces RAFC, a lightweight adaptation mechanism for wireless foundation models that avoids updating the pretrained backbone by dynamically routing features from intermediate layers. By generating task-specific weights to combine complementary representations across depths, the method achieves significant performance gains in mobile systems with minimal parameter overhead. The approach also demonstrates improved receiver robustness under real-world interference conditions using portable software-defined radio testbeds.
Lightweight Adaptive Feature Composition for Heterogeneous Downstream Adaptation of Wireless Foundation Models
arXiv:2606.10277v2 Announce Type: replace Abstract: Mobile systems increasingly rely on heterogeneous learning-enabled wireless functions, for which separate taskspecific models incur redundant training and model-management overhead. Wireless foundation models (WFMs) enable these functions to share a pretrained backbone, but existing adaptation either updates the backbone per task or relies on an inflexible final-layer representation. We observe that intermediate WFM layers exhibit distinct depth-dependent correlation structures. Based on this observation, we propose a Routing Adapter for Feature Composition (RAFC), which summarizes selected hidden states into compact descriptors, generates task- and sampledependent routing weights, and combines the original fullresolution features without updating the pretrained backbone. Its gains arise from giving each task adaptive access to complementary representations across multiple depths rather than restricting it to the final layer. Experiments across four task categories and three WFM backbones show improvements in every evaluated backbone-task pair, with mean relative gains across the four task categories ranging from 2.9% to 35.9%, while RAFC adds only 0.003M-0.049M trainable parameters across the evaluated backbones. Experiments on a portable softwaredefined radio testbed further demonstrate improved end-to-end receiver robustness, particularly under partial-band co-channel interference. These results establish RAFC as a lightweight, model-agnostic, and interpretable adaptation interface for WFMs in mobile systems.
Abstract:Mobile systems increasingly rely on heterogeneous learning-enabled wireless functions, for which separate taskspecific models incur redundant training and model-management overhead. Wireless foundation models (WFMs) enable these functions to share a pretrained backbone, but existing adaptation either updates the backbone per task or relies on an inflexible final-layer representation. We observe that intermediate WFM layers exhibit distinct depth-dependent correlation structures. Based on this observation, we propose a Routing Adapter for Feature Composition (RAFC), which summarizes selected hidden states into compact descriptors, generates task- and sampledependent routing weights, and combines the original fullresolution features without updating the pretrained backbone. Its gains arise from giving each task adaptive access to complementary representations across multiple depths rather than restricting it to the final layer. Experiments across four task categories and three WFM backbones show improvements in every evaluated backbone-task pair, with mean relative gains across the four task categories ranging from 2.9% to 35.9%, while RAFC adds only 0.003M-0.049M trainable parameters across the evaluated backbones. Experiments on a portable softwaredefined radio testbed further demonstrate improved end-to-end receiver robustness, particularly under partial-band co-channel interference. These results establish RAFC as a lightweight, model-agnostic, and interpretable adaptation interface for WFMs in mobile systems.
Machine Learning (cs.LG)
arXiv:2606.10277 [cs.LG]
(or arXiv:2606.10277v2 [cs.LG] for this version)
https://doi.org/10.48550/arXiv.2606.10277 Focus to learn more arXiv-issued DOI via DataCite
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Full text under the publisher's open licence · 288 words
Related coverage
A Unified Adaptive Feature Composition Framework for Multi-Task Generalization in Wireless Foundation Models
arXiv · Jun 10, 2026
Federated Self-Supervised Modulation Classification under Non-IID and Imbalanced Data
arXiv · Aug 19, 2026
Fast Cross-Scenario Adaptation of CSI Models via Channel Conditional Parameter Generation
arXiv · Jul 28, 2026
scBatchProx: Federated-Inspired Refinement for Stable Cell-Type Discriminability under Heterogeneous Batch Compositions
arXiv · Jun 3, 2026