Enhancing MPTCP Performance on High-Speed Trains with Predictive Handover-Aware Packet Scheduling
Min Ki Kim, You-Ze Cho · 2023
In the environment of a high-speed train operating at a speed of 300 km/h, TCP experiences challenges arising from the train's high mobility, such as rapidly changing channel conditions and frequent inter-cell handovers. These challenges result in significant decreases in throughput and temporary connection interruptions. The Multi-Path TCP (MPTCP) using multiple cellular carriers for high-speed train scenarios is being considered as a solution to address these issues caused by frequent inter-cell handovers. However, the default scheduler(minRTT) of MPTCP inadequately handles frequent handovers occurring in the high-speed train environment, leading to a degradation in throughput. In this paper, we propose a location-based predictive handover-aware packet scheduler designed for the high-speed train environment, considering its unique characteristics: fixed path, stationary base station locations, and constant velocity. The proposed scheduler, upon handover occurrence, demonstrated an average instantaneous throughput enhancement of 20.8% and an overall throughput improvement of 1.95%.