Year: 2025
Type: conference-paper
Abstract: In the 6G era, the Management and Orchestration (MANO) layer must become more intelligent to support advanced use cases with stringent Quality of Service (QoS) requirements. These often span multiple domains, terrestrial, non-terrestrial, and edge, requiring efficient coordination across heterogeneous networks. To meet these demands, Mobile Network Operators (MNOs) are evolving their Operations Support Systems (OSS) toward a "network of networks" vision, built on standardized and interoperable APIs. Standards bodies such as 3GPP, ETSI, and TM Forum are addressing this by developing numerous OSS API specifications and intent-based request models in JSON or YAML. However, creating these intents remains challenging for users without deep domain knowledge. We present OSS-GPT, an agentic AI framework that enables users to interact with OSSs via natural language. By leveraging Large Language Models (LLMs), OSS-GPT translates high-level intents into executable sequences of OSS API calls and automates their execution.
Institutions: EURECOM, University of Patras
Language: English
Subfield: Computer Networks and Communications
Field: Computer Science
Domain: Physical Sciences
Open Access status: gold
Funder: European Commission
OpenAlex ID: W4414109790