Abstract-Driving in an urban canyon can be frustrating when your GPS teller keeps telling you to make a turn at the place that you just passed, because the information transmission is deferred by the wireless signal reflecting off of buildings and other interfering objects. In this paper, we provide a practical solution for turn-to-turn guidance with inter-vehicle communication in vehicle ad-hoc networks (VANETs). Vehicles collect information from neighbors and catch the snapshot to describe the global impact of traffic congestions, in the presence of unpredictable changes of topology and vehicle trajectory. Without any centralized control, the information can be aggregated along the traffic flow and be disseminated in a minimal area, while sufficiently guiding each vehicle to achieve a global optimization on its path, and to remain on a non-blocked route. The information constitution is implemented in the proactive model, saving the delay of reconstruction in the reactive model (on-demand). Its substantial improvement on the elapsed time will be shown in the experimental results, compared with the best results known to date in both proactive and reactive information models.