Performance Evaluation of Link Metrics in Vehicle Networks: A Study from the Cologne Case

In vehicle ad hoc networks, the link duration is an important concept for clustering and dissemination. Many link metrics are proposed to predict it based on the vehicles' velocities and distance between vehicles. However, there is a lack of comparison of these metrics under the same framework. In this paper, we compare several common adopted link metrics, LLT (link life time), SLS (spatial locality similarity), and SF (similarity function), in the performance of link duration prediction, according to a real vehicular mobility trace of the city of Cologne, Germany. We find that there exists a large gap in between the predicted link duration order and real link duration order for the real traffic trace, and there is no winning link metric in all scenarios. Furthermore, as an example, we show that, by combining existing link metrics intelligently, it is possible to create a new link metric that predicts link duration better.