Distributed Intelligence for 5G and Beyond
Cellular and Internet of Things (IoT) massive deployment is increasingly questioning the possibility to handle several network functionalities, such as resource allocation and service dissemination, by using few centralized network entities. The analysis and design of scalable and robust solutions for decentralized network management, characterized by highly-performing learning schemes, is thus needed, aiming to satisfy end-user demands in terms of Quality of Service and Experience (QoS/QoE).
The analysis of decentralized network management requires the study of the interactions across heterogeneous entities, e.g., network operators, access nodes, and end-user devices, that may cooperate or not towards discovering optimal configurations. For example, from a user perspective, this may be the selection of the best network to connect among several candidates. Game Theory (GT) and Multi-Agent Learning (MAL) can be jointly used to analyze such scenarios and propose practical solutions. As part of MAL, the so-called Multi-Agent Reinforcement Learning (MARL) branch aims to extend RL solutions, originally tailored for single-agent scenarios, into multi-agent settings. Being at early research stage, GT+MA(R)L analysis of network functionalities presents several open challenges that can be investigated in the context of this Project.
The goal is to analytically address several challenges within the broad topic of cognitive networking, focusing in particular on the analysis and design of schemes for decentralized management and organization of cellular and IoT systems.
Knowledge of cellular technologies (4G, 5G), knowledge of Machine and Reinforcement Learning, Application of game theory to networking scenarios.
Foundations of wireless communications, initial knowledge of reinforcement learning and game theory, Python programming, Matlab knowledge (optional).