Performance Evaluation of Mobility in Non-Terrestrial Networks

Low Earth Orbit (LEO) satellites exhibit high mobility, leading to frequent handover challenges. Addressing these handover issues is crucial for maintaining seamless and stable service connections. This paper presents a novel D1 event handover scheme that integrates Kalman filtering with artificial neural networks(ANN) to enhance handover performance in Non-Terrestrial Networks (NTN). In the proposed method, Kalman filtering provides precise UE position estimates and generates smooth trajectory predictions, which serve as inputs to a neural network. The neural network adaptively adjusts handover parameters (hysteresis and threshold) and intelligently selects the optimal target cell based on predicted positions, signal-to-interference-plus-noise ratio (SINR) and network load. Through modelling and simulation experiments, we demonstrate that the proposed scheme significantly reduces handover failure rates and ping-pong events. The results show that, compared to traditional D1 handover mechanisms and RSRP-based handover methods, the proposed scheme improves handover stability and reliability, offering an effective solution for intelligent mobility management in dynamic NTN environments.