With a growing trend in usage of mobile navigation devices, there is also a need for an intelligent pedestrian navigation system. As an example, one might think of a visit to a large commercial center or a theme park, where the visitor must be efficiently guided to accomplish his tasks. In order to efficiently schedule a visit in such a scenario, the system not only needs to simulate and determine the waiting times across each destination, but also to integrate the entire system into the geometry of the scenario, to provide spatial context to non-spatial data. In this paper, we present a framework for a pedestrian simulation that models the movement of pedestrians by embedding the simulation into a geometric context. The resulting model considers the dimensions and architectural constraints of the scenario, simulates the pedestrian movement over a 3D graph network and gives an overall view of the congestions along the paths, waiting times at the destinations, etc.