2nd Reinforcement Learning Applied to Networking (RLAN) workshop

RLAN 2026 - The Second International Workshop on Reinforcement Learning Applied to Networking (RLAN) Workshop will be held in conjunction with the 22nd edition of the International Conference on Network and Service Management (CNSM) 2026 in Alcalá de Henares (Madrid), Spain, on October 26 - 30. 

The RLAN Workshop will address critical technical issues at the intersection of Reinforcement Learning (RL) and networking. Key research areas include:

• Network Service Management and Orchestration: RL can be used to intelligently manage and orchestrate network services according to the network status and user requirements.

• Network Resource Allocation: RL can dynamically adapt to changing conditions to distribute network resources efficiently.

• Traffic Engineering and Management: RL is applied to develop intelligent routing strategies.

• Network Resilience and Security: RL enhances real-time anomaly detection and response.

The workshop will explore RL applications in various networking scenarios, advancing autonomous network management to create self-organizing, self-healing networks. Addressing cutting-edge topics that

align with current academic and industry trends, the RLAN Workshop will particularly appeal to the CNSM community. Its interdisciplinary approach bridges machine learning and networking, offering

practical insights and real-world applications that attendees can apply to their work. The workshop will feature distinguished speakers, a panel discussion, and numerous networking opportunities, fostering collaboration and knowledge exchange among researchers, practitioners, and industry experts. 

Workshop General co-chairs

• José Santos (Ghent University - imec, Belgium)

• Guillaume Fraysse (Orange Research, France)

Technical Program Committee (TPC)

• Mauro Tortonesi (University of Ferrara, Italy)

• Filippo Poltronieri (University of Ferrara, Italy)

• Tim Wauters (Ghent University - imec, Belgium)

• Jaime Galàn-Jimènez (University of Extremadura, Spain)

• Marco Zambianco (FBK, Italy)

• Jaime Llorca, (New York University, NY, USA)

• Roberto Rodrigues Filho (Federal University of Santa Catarina, Brazil)

• Francesca Fossati (Sorbonne Université, France)

• Raphaël Feraud (Orange Research, France)

• Yoichi Matsuo (NTT Network Service Systems Laboratories, Japan)

Title: From Explanations to Actions: Making Deep Reinforcement Learning Trustworthy in Mobile Networks 

by Prof. Claudio Fiandrino (IMDEA Networks Institute, Madrid, Spain)

Abstract

Deep reinforcement learning (DRL) is becoming a common way to control mobile networks: it schedules radio resources, slices capacity and manages energy, often as xApps in the Open RAN. Operators still hesitate to deploy it. A DRL agent maps high-dimensional network states to actions through a policy nobody can read, so when it makes a poor decision, nobody can say why or how to correct it. Standard explainable AI tools were built for classifiers on images and text. They do not handle sequential decisions, delayed rewards or the domain semantics of network KPIs, and their explanations rarely tell an engineer what to change. In this talk I argue that explainability in networking should be judged by whether it lets us act on an agent, not only understand it. I show this through three systems from our group. EXPLORA (ACM CoNEXT 2023) links an agent's actions to their effects on KPIs and replaces harmful actions at runtime. SymbXRL (IEEE INFOCOM 2025) turns agent behavior into symbolic rules that engineers can read and uses them to steer actions toward an operator's intent. SIA (IEEE INFOCOM 2026) explains forecast-driven agents and refines their actions online, improving performance by up to 25% without retraining. Across the three systems, explanations work as a control interface: they find failure modes, encode operator intent and add a runtime safety layer to learned controllers.

Speaker Bio

Claudio Fiandrino is a Research Assistant Professor at IMDEA Networks Institute, Madrid, Spain, where he leads the Resilient AI Networking Lab. For his research, Claudio has been awarded with several fellowships, including the Spanish Ramón y Cajal Fellowship and Fulbright scholarship, as well as several Best Paper Awards (e.g., IEEE INFOCOM 2024). His primary research interests include explainable and robust AI in 5G/6G networks, 5G network performance characterization with measurements, and AI/ML for the Open RAN.