A two-level hidden Markov model for characterizing data traffic from vehicles

With the popularization of intelligent transport and mobile internet services, vehicles and people on board generate increasing amounts of data. To match future networks with this use case, tools are needed to analyze the requirements set for the network. In this paper, we study the characteristics of data traffic in the context of networked vehicles. We generate data traffic based on real-world vehicle traces and reported data patterns of end-user applications and vehicles. Based on this, we propose a two-level hidden Markov model to describe both large and small temporal characteristics of data traffic from vehicles aggregated on base stations. We evaluate the proposed model by comparing the original and synthesized data. The results show that the proposed model can well characterize the data traffic from vehicles.