Machine Learning based approaches for safe and secure communications at the Physical Layer

The topics covered in this work are a proposal for Machine Learning-based authentication scheme based on Wireless Channel fingerprinting and an application of Machine Learning to channel prediction in a URLLC scenario. The authentication scheme considers both classification and anomaly detection-based solutions to authenticate devices in a wireless network and detect spoofing attempts. The proposed channel forecast application aims to counteract CQI aging with an early retransmission solution to prevent packet loss and meet URLLC's QoS requirements.