LAUSR

In this letter, we propose a two-stage traffic load prediction scheme for network slices (NSs) in high-speed railway (HSR) wireless networks, where in the first stage, the K-means algorithm is leveraged to cluster traffic flows, and in the second stage, the long-short term memory (LSTM) algorithm is applied to predict the traffic load. Based on the obtained traffic features (including traffic volume and user velocity) and the network radio resource characteristics (including coverage performance and capacity), we design a service-tailored resource reservation mechanism. Simulation results show that our proposed scheme can significantly improve the traffic load prediction accuracy to ensure the NS resource reservation performance.