Site-specific millimeter-wave compressive channel estimation algorithms with hybrid MIMO architectures

In this paper, we present and compare three novel model-cum-data-driven channel estimation procedures in a millimeter-wave Multi-Input Multi-Output (MIMO) Orthogonal Frequency Division Multiplexing (OFDM) wireless communication system. The transceivers employ hybrid analog-digital architecture. We adapt techniques from wide range of signal processing methods, such as detection theories, compressed sensing, Bayesian inference, to learn the unknown virtual beamspace domain dictionary, well delay-and-beamspace sparse channel. train model-based algorithms with site-specific training dataset generated using realistic ray tracing-based simulation tool. assess performance proposed same site's test data. benchmark our terms normalized mean squared error against an existing fast greedy method empirically show that approaches combined data-driven customization unanimously outperform state-of-the-art by large margin. were selected top solutions "ML5G-PHY Channel Estimation Global Challenge 2020" organized International Telecommunication Union.