Abstract: Modern applications are no longer isolated or static, but deeply interconnected. They traverse multiple layers, spanning from cloud to edge, and interact with a growing. The size and complexity of their building parts, i.e., data, logic, and infrastructure, are unprecedented. This evolution requires learning mechanisms for automating and coordinating the functioning of such wide systems. Still, these learned models have to be not only dynamic and precise, but — very importantly — trustworthy and transparent. It is essential to develop tools and strategies to understand how decisions were made and build a solid trust in the automated components that manage such systems to achieve this objective. This keynote will look into these aspects, highlighting methods for explainability for modern distributed systems, spanning from network traffic to distributed applications on cloud infrastructure, looking at where the understanding of their functioning, plus the trust, play an essential role.

Biography: Andrea Morichetta is a postdoctoral researcher at the Distributed Systems Group of TU Wien, where he leads both Bachelor and Master-level courses in Distributed Systems. His research is contributing to the study of complex, large-scale distributed systems, through modular, multi-modal coordination. Dr. Morichetta holds a Ph.D. in Electrical, Electronics, and Communication Engineering from Politecnico di Torino and has collaborated with leading institutions and industry partners such as Cisco (San Jose, CA, US), AIT (Vienna, AT), Futurewei Technologies (Seattle, WA, US), and Tsinghua University (Beijing, CN).

Abstract: The data-intensive and time-sensitive communication demands of modern networked intelligent systems far exceed the capabilities of traditional connectivity models, which focus on transferring data between digital devices. Today’s cyber-physical and autonomous systems are capable of acquiring, processing, and transmitting vast amounts of distributed, real-time data. In this talk, we introduce a novel communication paradigm envisioned for future 6G networks, one that accounts for the significance and usefulness (i.e., semantics) of the information being generated, processed, and transmitted. We will explore potential theoretical metrics and KPIs that could help define and implement this paradigm shift. The discussion will include relevant case studies that leverage statistical signal processing, dynamical systems, reinforcement learning, and control theory. The central aim is to highlight the scientific challenge of achieving mathematical convergence between signal processing and goal-oriented communication. This involves exploiting source and signal characteristics, process variability, and semantic attributes of information. If successfully realized, this approach could substantially reduce unnecessary data traffic, thereby lowering the communication, processing, and energy demands of future intelligent systems.

Biography: Dr. Stavrou (Senior Member IEEE) received the D.Eng. degree from the Department of Electrical and Computer Engineering (ECE), Faculty of Engineering, Aristotle University of Thessaloniki, Greece, in 2008, and the Ph.D. degree from the Department of ECE, Faculty of Engineering, University of Cyprus, Cyprus, in 2016. From November 2016 to September 2022, he held postdoctoral research positions at Aalborg University (Denmark), the KTH Royal Institute of Technology (Sweden), and EURECOM (France). Since September 2022, he has been an Assistant Professor in the Communication Systems Department at EURECOM. His research interests include information and communication theory, networked control systems, goal-oriented communication, as well as feedback and privacy in communication, optimization, and game theory. Dr Stavrou is a co-recipient of the Best Paper Award in Industry Innovation at ACP/POEM 2023 and of the Best Poster Award at MenaML 2025. He has authored or coauthored more than 75 peer-reviewed conference and journal papers, along with four book chapters.