frezghi / ef112x-fast-beam-alignment- Goto Github PK

Future data-hungry 6G applications such as holographic telepresence, wireless extended reality, and autonomous vehicles are expected to be supported by higher frequency bands, e.g., millimeter wave and terahertz bands. However, these bands suffer from high penetration loss through blockages. To this end, multi-antenna base station (BS) can be used to establish line-of-sight reflection links, thereby ensuring reliable communications. This requires efficient beam alignment to identify the optimal beams at the base station maximize the achievable communication rate.

Conventional beam alignment methods require a large beam training overhead. The challenge of this project is to consider the trade-off between communication rate and beam training overhead, and understand the impact of different system parameters on beam alignment. This would include a study of the state- of-art work, combined with simulations of different scenarios and various machine learning algorithms.

This project aims at developing a novel machine learning-based scheme to achieve a good beam alignment performance with a small amount of beam training overhead.