Real-Time Traffic Density Estimation: Putting on-Coming Traffic to Work
Ryan Florin, Stephan Olariu
Abstract
Estimating highway traffic density in real-time is an important goal of Intelligent Transportation Systems. The main contribution of this work is to propose a simple and entirely V2V-based methodology to estimate, in real-time, traffic density based on data collected by moving observers in co-directional and on- coming traffic. To estimate traffic density in a privacy-preserving manner, our methodology uses tallies consisting of the number of times a vehicle is passed by other vehicles, minus the number of times it passes other vehicles. As it turns out, keeping tallies is a non-trivial task since vehicles are allowed to vary their speed in arbitrary ways and, as a result, the same two vehicles may pass each other any number of times. We provide a detailed proof of correctness of our methodology; to assess its accuracy, we have performed extensive simulations and sensitivity analyses using SUMO-generated synthetic traffic traces over a wide range of penetration rates and traffic flows.