A two-tier machine learning-based handover management scheme for intelligent vehicular networks

Abstract With the increasing demand for real-time road safety services and infotainment applications on vehicles, the development of an efficient wireless mobile communication became crucial for the content delivery of such services in Intelligent Vehicular Networks (IVN). Mobility management enables mobile hosts to communicate over the Internet from foreign networks. However, vehicles' high mobility and the rapid shifts in network topology affect the performance of traditional mobility management protocols. Hence, raising the challenge for seamless wireless communications over IVN. In this paper, we propose a two-tier Machine Learning-based scheme for handover management in intelligent vehicular networks. In the first tier, we use a recurrent neural network model to predict the receiving signal strength of Access Points (APs), to derive a handover trigger decision. In the second tier, a stochastic Markov model is used to select the next access point by utilizing the vehicle flow projections. The performance of the proposed protocol is evaluated using NS-2 simulator and generated vehicles mobility. Simulation results show that the proposed ML-based model outperformed related work in term of prediction accuracy, while the integration of the handover trigger scheme and the access point selection method improved network performance.