Spatio-Temporal Variations of Vehicle Traffic in VANETs: Facts and Implications

Via statistical analysis of several sets of empirical data collected from realistic scenarios, we realize that exponential model is a good fit for highway vehicle traffic. This model provides a single parameter (the exponent of vehicle density lambda(s) over bar) to describe traffic variation. We reveal that highway traffic drastically varies over time (of different scales), across geographic locations and over technology adoption phases. Considering multi-hop geocast and single-hop broadcast, we show via both mathematical analysis and simulations the performance of path availability in the former, and packet delivery ratio in the latter, with respect to short- and long-term traffic variation. Therefore, we argue that the critical challenge in vehicular networking is dealing with traffic variation over the small scale (due to differences in vehicular density over space and time) and over the long term (due to market penetration).