Accelerating AI-RAN at AWS Agentic AI Hackathon

The Amazon Web Services (AWS) Breaking Barriers for Agentic Networks Hackathon in partnership with Telecom Infra Project brought together 377 participants across 68 projects to explore the role of agentic AI in solving challenges and capturing new opportunities in the telecom industry. 45 in-person teams attended the hackathon at Dogpatch Labs in Dublin, Ireland, while others streamed in from around the world.

VIAVI’s TeraVM AI RSG tool was made available to all participants to generate data for their projects. Use cases supported included anomaly detection, mass event,  A1 policy and control, high-throughput scenarios, energy optimization and network slice management. In addition, VIAVI representatives provided insights into the TeraVM AI RSG tool and supported the teams using AI RSG.

Three out of the four winning teams used TeraVM AI RSG to train agentic networks, for example, an anomaly detection agent as part of a self-healing architecture designed to tackle real-world telecom challenges. The TeraVM AI RSG dataset generator played a crucial role in bringing the agentic systems with multi-party agentic core to life.

The RSG accurately simulates system-level RAN behavior, creating a RAN Digital Twin (RDT) that mimics real network conditions and behaviors. By turning network data into high-fidelity test environments using both synthetic and real network KPIs, the TeraVM AI RSG enables modeling, stress-testing and optimization of AI-powered applications before changes hit the live network.

Supporting 4G, 5G, emerging 6G and NTN networks, the AI RSG helps train AI-driven models for optimization of technologies like massive MIMO.

During the Hackathon, VIAVI also introduced an R1 App development kit as an abstraction of O-RAN for AI developers.

With the help of VIAVI’s R1 ADK, AI developers can start developing RAN applications today. VIAVI is working to ensure that apps using this API will work with minimal integration effort on SMOs exposing R1 interface.