Deep Learning Application to 5G Physical Layer for Channel Estimation and CSI Feedback Improvement

Starting from the last decades of the 20th century, Machine Learning has been widely applied in many engineering fields, such as communications, speech and image processing, computer vision and robotics, resulting particularly effective and useful in contexts where a rigorous mathematical model of the problem is too hard to be elaborated. Focusing on wireless communication systems, in recent years Machine Learning applications to the upper layers have been minutely explored for various purposes, like the deployment of cognitive radio and Self Organized Networks or the resource management, while its application to the physical layer has been somehow overlooked. The purpose of this thesis is to investigate the potential use of neural networks for the optimization of specific physical layer blocks in a communication system, taking into account the peculiar characteristics of the emerging radio technologies based on 5G standard (e.g. massive Multiple-Input Multiple-Output, beamforming, millimeter Waves) and all their related challenges. In particular, the analysis focuses on channel estimation and Channel State Information feedback reporting blocks, providing data and statistics that are representative of how Machine Learning algorithms introduction affects radio link performance, in terms of BLock Error Rate and throughput.