Current students

In the context of 6G Vehicle-to-Everything (V2X), multiple-input multiple-output (MIMO) systems will require flexible and adaptive methods to enable efficient communication. Embedding prior knowledge about the application context within the inference procedures, deep learning models can widely improve standard physical layer communication protocols. In the last few years, many machine learning-based solutions have been proposed in the related literature, concerning, e.g., signal detection, channel estimation, positioning, modulation recognition, and coordinated beamforming tasks. As far as the whole communication system is concerned, proposed models extend from model-based approaches, up to channel agnostic end-to-end learning. We examined the channel estimation task, extremely challenging within the high mobility settings emerging in V2X communication scenarios. At millimetre and terahertz waves, multiple-input multiple-output (MIMO) channels show a sparse impulse response in the angular and delay domains, also jointly referred to as the Space-Time (ST) domain. Exploiting the prior knowledge on the propagation conditions, and the recurring patterns emerging in vehicular networks, we designed a deep neural network to infer the ST subspace spanned by the channel underlying an Unconstrained Maximum Likelihood (U-ML) channel estimate. To examine the performance of the developed algorithm, we simulate realistic vehicular traffic and electromagnetic propagation in urban scenarios by means of specialized simulation software. With respect to standard U-ML channel estimation, the developed deep learning model showed remarkable benefits in terms of Mean Squared Error (MSE), attaining on average the theoretical lower bound over simulated vehicular trajectories unseen at training time. Encouraged by the obtained results, this PhD research aims at exploiting the representational power of machine learning models in order to provide novel signal processing solutions for the physical layer of V2X communication systems in the context of 6G wireless communications.