In collaboration with ITU, GSMA, and IEEE, we are pleased to introduce the AI Agents for RAN Challenge, which invites participants to design AI agents capable of reasoning, using tools, and solving engineering problems in radio access networks.
The challenge is built around Giulia, a radio network simulator that enables agents to investigate network scenarios and test the effects of their decisions. In this webinar, we introduce the challenge, walk through representative tasks, and explain how agents interact with the simulator and how their solutions are evaluated.
Participants will compete for a total prize pool of €30,000 and will be invited to present their solutions at MWC Barcelona 2027 (1-4 March 2027).
About the challenge
Radio access networks are complex systems: a change to one site or cell can affect coverage, interference, user association, throughput, and energy consumption elsewhere in the network. Solving an engineering problem therefore requires more than identifying a poor performance indicator. Agents must form hypotheses, choose informative tests, and account for trade-offs across the network.
The AI Agents for RAN Challenge puts these capabilities to the test. Agents receive a network scenario and an engineering objective, then interact with Giulia to inspect the network, run simulations, and evaluate possible interventions. Tasks span several areas of radio network engineering, including:
- Fault diagnosis and identification of underlying causes
- Outage impact and service quality analysis
- Radio configuration and coverage optimization
- Network topology and site placement
- Energy management and performance trade-offs
The challenge includes both analytical questions and tasks that require agents to find a configuration meeting specific performance conditions. Some tasks limit the number of simulations an agent may run, making the choice of each experiment part of the challenge.
Agent performance is assessed through task-specific verification of the proposed answer and its measured effects. The evaluation also records simulation and tool usage, helping distinguish a correct result from an efficient approach to reaching it.
Learning objectives:
By the end of the webinar, attendees will be able to:
- Describe how agents interact with the Giulia simulation environment
- Identify the main categories of radio network problems in the challenge
- Explain how agents use simulations to test hypotheses and compare interventions
- Understand how solutions are assessed for accuracy and efficient use of simulations
Recommended prerequisites: This challenge is recommended for participants with intermediate to advanced knowledge of AI agents and LLMs, along with a basic understanding of telecom networks.
