Channel Estimation via Digital Twins with Limited a Priori Knowledge

Digital Twin (DT) has emerged as a promising solution for channel estimation. By leveraging high-resolution 3D models of the scenario and ray-tracing simulations, DT could provide valuable site-specific prior knowledge on the channel's space-time (ST) invariant features of the multipath environment, such as angles of arrival, angles of departure, and propagation delays. However, the real-time characterization of these features imposes computational constraints on ray-tracing simulations, hence limiting the prior knowledge of the multipath environment and corresponding ST features, and degrading estimation accuracy. In this paper, we propose and investigate, for the first time, three distinct DT-empowered low-rank methods for channel estimation, under different degrees of prior knowledge corresponding to limited number of paths provided by DT. Specifically, these methods perform modal projection onto a joint space-time, a spatial, and a temporal subspace. We compare our proposed methods with state-of-the-art techniques, and evaluate their performance in a synthetic scenario. Numerical results show that robustness, when prior knowledge is limited to few paths, is achieved when exploiting only temporal features, while estimation accuracy is attained when joint space-time features are considered.