An LSTM-based Mobility Prediction Mechanism in the ICN-based Vehicular Networks

Conventional mobility prediction mechanisms use the global positioning system (GPS) location history of a user. However, the exiting mechanisms did not focus enough on the timedependent sequential location history for mobility prediction. On the other hand, information-centric networking (ICN) is a recent paradigm for future Internet architecture, and the built-in caching mechanism is one of the main characteristics of ICN. In this paper, our objective is to use the sequential and time-dependent location history of mobile devices and propose an LSTM-based mobility prediction mechanism to predict the probable next position of a mobile device. Moreover, accurate mobility prediction may assist the proper handover prediction so that the caching capability of ICN can be proactively used in the ICN/cell-based vehicular network. We simulate the proposed mechanism with the real dataset, which was collected using a mobile application for six months. Our predicted simulation results ensured the correctness of the proposed mobility prediction algorithm, which leads to the end-to-end seamless content delivery by using the proactive caching mechanism of ICN and predicting the future movement direction of the mobile device.