ML-Assisted Hybrid Channel Prediction Model based on the Ray Tracing Approach

With the advancement of artificial intelligence (AI) and machine learning (ML), data-driven approaches have emerged as powerful tools for accurate wireless channel prediction. However, existing methods often rely on satellite or point cloud images for environmental feature extraction, which can introduce redundant information and lead to overfitting, compromising prediction accuracy. To address these challenges, this paper proposes a novel AI-assisted hybrid model to enhance path loss (PL) prediction accuracy in complex environments, which is consisted of an OpenStreetMap (OSM)-based ray tracing (RT) approach and a ML-based path loss prediction model. Performance of the proposed approach is validated with using the measurement data at campus, which demonstrates the superiority of the proposed model over traditional methods, highlighting its potential for future communication system design and optimization.