An Overview and Solution for Democratizing AI Workflows at the Network Edge

TL;DR

This work addresses the challenge of democratizing AI/ML workflows at the network edge by introducing NAOMI, a modular, architecture-agnostic platform that orchestrates data preparation, training, model management, deployment, and continuous operation using open-source tools. By analyzing O-RAN WG2 AI/ML workflow architecture alongside MLOps pipelines, the authors identify eight democratization requirements and design NAOMI to satisfy them, including openness, ease of use, modularity, self-evolving capabilities, heterogeneity, distributed services, virtualization, and scalability. Empirical evaluation against the O-RAN AI/ML Framework shows NAOMI achieves up to $40 ext{ extpercent}$ faster deployment and up to $73 ext{ extpercent}$ faster workflow execution on large datasets, while maintaining competitive inference performance, especially in distributed edge setups. The results demonstrate NAOMI’s viability for edge-native AI applications across heterogeneous hardware, facilitating rapid service instantiation, distributed training, and scalable inference, with potential impact on 6G and beyond cellular networks. Future work will enhance reliability, GPU-accelerated edge training/inference, and intelligent task distribution to further strengthen edge democratization.

Abstract

With the process of democratization of the network edge, hardware and software for networks are becoming available to the public, overcoming the confines of traditional cloud providers and network operators. This trend, coupled with the increasing importance of AI in 6G and beyond cellular networks, presents opportunities for innovative AI applications and systems at the network edge. While AI models and services are well-managed in cloud systems, achieving similar maturity for serving network needs remains an open challenge. Existing open solutions are emerging and are yet to consider democratization requirements. In this work, we identify key requirements for democratization and propose NAOMI, a solution for democratizing AI/ML workflows at the network edge designed based on those requirements. Guided by the functionality and overlap analysis of the O-RAN AI/ML workflow architecture and MLOps systems, coupled with the survey of open-source AI/ML tools, we develop a modular, scalable, and distributed hardware architecture-independent solution. NAOMI leverages state-of-the-art open-source tools and can be deployed on distributed clusters of heterogeneous devices. The results show that NAOMI performs up to 40% better in deployment time and up to 73% faster in AI/ML workflow execution for larger datasets compared to AI/ML Framework, a representative open network access solution, while performing inference and utilizing resources on par with its counterpart.

Figures (17)

• Figure 1: AI/ML data preparation ETL process.
• Figure 2: AI/ML training pipeline, adapted from b1.
• Figure 3: AI/ML model inference on O-RAN.
• Figure 4: Architecture diagram of the proposed NAOMI - AI/ML workflow democratized solution.
• Figure 5: Workflow diagram of actors interacting with the proposed NAOMI AI/ML workflow solution.