ai-enhanced 5g-r architecture for adaptive connectivity and security in high-mobility rail environments

Modern high-speed rail systems require ultra-reliable, low-latency, and secure wireless connectivity to ensure operational safety and high-quality passenger services at velocities exceeding 300 km/h. existing technologies such as gsm-r and lte-r fail to maintain seamless handovers, adaptive resource management, and adequate protection against dynamic threats. this paper proposes an ai-enhanced and security-aware 5g-r architecture that integrates rnn-based mobility prediction, dynamic resource partitioning, multi-access edge computing (mec), and a secure connectivity layer for real-time optimization. the system continuously predicts train velocity and signal variations to proactively allocate network slices and pre-configure handovers across edge nodes. federated learning synchronizes local mec models with a global cloud controller, enabling continual adaptation to changing conditions. simulation and field evaluations conducted along a 174 km railway segment achieved throughput up to 240 mbps, latency below 20 ms, and 94 % handover success, while introducing only 3.7 % security overhead. results confirm that ai-driven resource control and mec-assisted execution significantly enhance the reliability, continuity, and resilience of 5g-r communications under high mobility.