Integrating wireless communication engineering and machine learning
The research in WindMill is about the integration of two research fields: wireless communications and machine learning. The overall research objective is the development of new methodologies based on machine learning in the design of wireless systems, while also contributing to the advancement of applied ML science. The achievement of the ultimate scientific objectives of the WindMill project will be pursued by the accomplishment of the following, more specific goals:
Sub-objective 1: Advancing the field of ML for wireless communications.
The objective is to advance the state of the art in the field of ML where the ML methods respond to specific constraints found in wireless communications. Those constraints are: 1) a highly dynamic environment, 2) distributed wireless systems leading to distributed algorithm implementation, 3) stringent end-to-end delay constraints.
Subobjective 2: Prediction schemes and anticipatory optimisation for fast-varying processes.
The objective is to develop ML methods for the physical (PHY)-layer of wireless systems. One goal is to learn and anticipate spatio-temporal features related to the wireless channel in multiuser multi-antenna networks. Distributed and centralized processing will be addressed, in particular distributed channel estimation and multi-user beamforming in massive MIMO, multi-cell, connected vehicle and robotic networks.
Subobjective 3: Data-driven optimisation schemes for radio access management.
The objective is to develop ML methods to solve radio access problems in networks with massive connectivity and different QoS requirements, which is essential in the upcoming 5G and will remain even more so in subsequent network generations. One part of the work is dedicated to the field of radio resource management, e.g., power control, rate-adaptation. Another part of the work is dedicated to the field of IoT and URLLC where new important communication paradigms have emerged
Subobjective 4: System-wide “cognitive” optimisation schemes.
The objective is to develop distributed and hierarchical ML architectures to enable cognitive network slicing and achieve system-wide optimisation. The goal is to come up with effective approaches to achieve global system optimisation by confederating fundamental ML blocks that operate at different levels (physical, data link, network) and with various time scales.