Research groups develop Deep Learning Radio Map for MIMO. #AIResearch

The content discusses a presentation on deep learning-based radio maps for MIMO and cell-free massive MIMO systems. The technology aims to reduce the dependency on channel state information by generating radio maps directly from raw CSI data. The presentation introduces two baseline schemes and proposes an end-to-end architecture using deep neural networks. Results show the potential to replace MIMO CSI with location data and highlight the advantages of the approach. A cell-free massive MIMO beamforming framework is also proposed, utilizing deep learning to construct a comprehensive radio map. The integration of space-time block coding helps mitigate the random phase gap issue in cell-free MIMO. Simulation outcomes demonstrate the superior performance of the radio map method over other approaches, with STBC-enabled carrier aggregation and multi-BS coordination improving communication performance significantly.

The bio of Wei Zhang, the presenter, highlights his research interests in space information networks, mmWave communications, and massive MIMO. He is currently a Professor at The University of New South Wales and holds leadership positions in the IEEE Communications Society. Zhang has also served as an editor for various IEEE publications and has received fellowships from both the IEEE and the IET.

Source link

Source link: https://www.imperial.ac.uk/a-z-research/intelligent-transmission-and-processing-laboratory/workshops–seminars/imperial-college-seminars-series-on-future-communications/8th-seminar—deep-learning-based-radio-map-for-mimo-and-cell-free-massive-mimo/