Mobile Network Optimization in High-Speed Railway Tunnels: An Adaptive Handover Based on Train Position and Signal Quality Prediction

As an emerging branch of mobile communication systems, High-Speed Railway (HSR) mobile networks face unique challenges not present in traditional mobile networks. Especially in tunnel environments, the narrow and closed structure of HSR tunnels causes the waveguide effect in the propagation of wireless signals, resulting in uneven signal coverage. Achieving seamless connectivity and stable communication in HSR tunnels is challenging, especially when the train travels to the Base Station (BS) coverage edge. To address these issues, we propose an adaptive Handover (HO) algorithm comprising three modules: train position prediction, Reference Signal Received Power (RSRP) prediction and target cell selection. The train position prediction module employs an Extended Kalman Filter (EKF) integrated with the Constant Turn Rate and Velocity (CTRV) model to provide accurate location information. The RSRP prediction module builds the Tunnel Radio Map (TR-Map) by incorporating the Dual-Slope Large-Scale Fading Model (DS-LSFM) into the kernel function of Gaussian Process Regression (GPR), providing continuous RSRP. The target cell selection module leverages the Proximal Policy Optimization (PPO) algorithm to select the optimal serving cell based on predicted train positions, target cell RSRP, and Cell Individual Offset (CIO) from cell measurement reports, minimizing HO interruption time and enhancing throughput. Compared to benchmark algorithms, the proposed adaptive HO algorithm significantly improves HO performance, achieving up to a 66.3% increase in throughput.