Abstract:
Due to the narrow mmWave beamwidth and high user mobility, efficient beam selection in vehicle-to-infrastructure (V2I) communication is crucial but difficult. Data-driven methods use contextual information from vehicle LIDAR sensors to reduce the search overhead of iterative beam discovery procedures.
This project proposes a lightweight neural network (NN) architecture and LIDAR preprocessing that significantly outperforms previous works. Our solution improves model convergence and accuracy with multiple novelties.
In particular, we define a novel loss function inspired by knowledge distillation, introduce a curriculum training approach that exploits line-of-sight (LOS)/non-line-of-sight (NLOS) information, and propose a non-local attention module to improve performance for the more difficult NLOS cases.
Our NN-based beam selection scheme can achieve 79.9% throughput of an exhaustive beam sweeping approach without beam search overhead and 95% by searching among as few as 6 beams. Our method reduces the beam search time needed to achieve a desired throughput in a typical mmWave V2I scenario.
Note: Please discuss with our team before submitting this abstract to the college. This Abstract or Synopsis varies based on student project requirements.
Did you like this final year project?
To download this project Code with thesis report and project training... Click Here
