Ali Khalesi - Communication systems
Date: - Location: Eurecom
Abstract: Modern distributed systems increasingly need to serve many users simultaneously, each requesting different computations or learning tasks from shared data. This talk introduces General Multi-User Distributed Computing, a unified perspective to reason about such systems beyond classical “single task / single user” models. I will explain (i) how to model heterogeneous user demands as general functionals of common sub-computations, (ii) what fundamentally limits performance when we are constrained by communication, computation, and latency/energy budgets, and (iii) how the network topology and task assignment shape achievable trade-offs. A key message is that, even when user demands are not neatly decomposable, one can still obtain principled design rules and sharp insights by combining information-theoretic thinking with learning-theoretic tools. The talk will highlight the main concepts, the most important trade-offs (communication vs. computation vs. accuracy), and the practical implications for scalable distributed and federated systems, including settings motivated by edge intelligence and aeronautical/satellite networks. Short bio: Ali Khalesi is an enseignant-chercheur (Assistant Professor) at IPSA (Ivry-sur-Seine, Paris region). He earned his B.Sc. in Electrical Engineering (Telecommunications) from K. N. Toosi University of Technology, Tehran (2014–2018, with honors), and his M.Sc. in Electrical Engineering (Secure Telecommunications & Cryptography) from Sharif University of Technology, Tehran (2018–2020, with high distinction). He received his Ph.D. in Telecommunications Systems from Sorbonne Université (EDITE) (2021–2024), with research carried out at EURECOM under Prof. Petros Elia. His Ph.D. thesis, “Multi-User Linearly-Decomposable Distributed Computing,” was shortlisted (semi-finalist) for the 2025 “Prix solennels” of the Chancellerie de Paris (Sorbonne Université), and he received the 2nd Prize of the EDITE Best Thesis Competition (2025). His research focuses on the fundamental limits of distributed computing and learning algorithms, and on information-theoretic analysis of aeronautical and satellite communication systems, with emphasis on communication–computation trade-offs, coding-theoretic methods, and reliability/latency constraints. He also holds a research affiliation with LINCS Lab (Paris).
