Saved in:
| Main Author: | |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2506.17259 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866916803433725952 |
|---|---|
| author | Barros, Sebastian |
| author_facet | Barros, Sebastian |
| contents | As artificial intelligence capabilities rapidly advance, Telco operators face a growing need to unify fragmented AI efforts across customer experience, network operations, and service orchestration. This paper proposes the design and deployment of a horizontal federated AI operating system tailored for the telecommunications domain. Unlike vertical vendor-driven platforms, this system acts as a common execution and coordination layer, enabling Telcos to deploy AI agents at scale while preserving data locality, regulatory compliance, and architectural heterogeneity. We argue that such an operating system must expose tightly scoped abstractions for telemetry ingestion, agent execution, and model lifecycle management. It should support federated training across sovereign operators, offer integration hooks into existing OSS and BSS systems, and comply with TM Forum and O-RAN standards. Importantly, the platform must be governed through a neutral foundation model to ensure portability, compatibility, and multi-vendor extensibility. This architecture offers a path to break the current silos, unlock ecosystem-level intelligence, and provide a foundation for agent-based automation across the Telco stack. The case for this horizontal layer is not only technical but structural, redefining how intelligence is deployed and composed in a distributed network environment. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2506_17259 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | The Case for a Horizontal Federated AI operating System for Telcos Barros, Sebastian Networking and Internet Architecture As artificial intelligence capabilities rapidly advance, Telco operators face a growing need to unify fragmented AI efforts across customer experience, network operations, and service orchestration. This paper proposes the design and deployment of a horizontal federated AI operating system tailored for the telecommunications domain. Unlike vertical vendor-driven platforms, this system acts as a common execution and coordination layer, enabling Telcos to deploy AI agents at scale while preserving data locality, regulatory compliance, and architectural heterogeneity. We argue that such an operating system must expose tightly scoped abstractions for telemetry ingestion, agent execution, and model lifecycle management. It should support federated training across sovereign operators, offer integration hooks into existing OSS and BSS systems, and comply with TM Forum and O-RAN standards. Importantly, the platform must be governed through a neutral foundation model to ensure portability, compatibility, and multi-vendor extensibility. This architecture offers a path to break the current silos, unlock ecosystem-level intelligence, and provide a foundation for agent-based automation across the Telco stack. The case for this horizontal layer is not only technical but structural, redefining how intelligence is deployed and composed in a distributed network environment. |
| title | The Case for a Horizontal Federated AI operating System for Telcos |
| topic | Networking and Internet Architecture |
| url | https://arxiv.org/abs/2506.17259 |