Context-aware Codebook Design for 6G Extreme MIMO Systems

Beam management is the defacto approach for configuring the antennas in 5G MIMO communication systems. Extending the beam management framework to larger arrays— also known as extreme MIMO systems—is challenging as the overheads grow with the array dimensions. One solution is to make use of the wealth of sensor data that is becoming available in integrated sensing and communication (ISAC) systems. In this paper, we propose a neural architecture for codebook design using environmental context derived from sensor data. In particular, we combine beamspace transformations with local occupancy grids obtained through network sensing to maximize the achievable rate in vehicular operations. Our results show significant performance gains over traditional codebooks while requiring less overhead than standard 5G beam management.