Black-box modeling of connected vehicle networks

In this paper, we propose a black-box modeling framework for connected vehicle networks comprised of conventional vehicles and vehicles equipped with wireless vehicle-to-vehicle (V2V) communication. First, we identify the link length that is the number of vehicles between the broadcasting and the host vehicle. Based on the estimated link length, a linear model is used to approximate the dynamics of the vehicle network. The proposed framework does not require priori information about the dynamics of the vehicle network, and hence can be implemented in real traffic. Numerical simulations are used to demonstrate the effectiveness of the estimators in capturing the link length and predicting the time evolution of the vehicle network. The estimated model can be used when designing connected cruise control (CCC) algorithms.