Computer Simulations of VANETs Using Realistic City Topologies Francisco J. Martinez, Manuel Fogue University of Zaragoza, Spain Email: {f.martinez, m.fogue}@unizar.es C. K. Toh National Tsing Hua University, Taiwan Email: ck away@hotmail.com Juan-Carlos Cano, Carlos T. Calafate, Pietro Manzoni Universitat Polit` ecnica de Val` encia, Spain Email: {jucano, calafate, pmanzoni}@disca.upv.es Abstract Researchers in Vehicular Ad Hoc Networks (VANETs) commonly use simulation to test new algorithms and techniques. This is the case because of the high cost and labor involved in deploying and testing vehicles in real outdoor scenarios. However, when determining the factors that should be taken into account in these simulations, some factors such as realistic road topologies and presence of obstacles are rarely addressed. In this paper, we first evaluate the packet error rate (PER) through actual measurements in an outdoor road scenario, and deduce a close model of the PER for VANETs. Secondly, we introduce a topology-based visibility scheme such that road dimension and geometry can be accounted for, in addition to LOS (line-of-sight). We then combine these factors to determine when warning messages (i.e., messages that warn drivers of danger and hazards) are successfully received in a VANET. Through extensive simulations using different road topologies, city maps, and visibility schemes, we show these factors can impact warning message dissemination time and packet delivery rate. Index Terms Vehicular ad hoc networks; attenuation and visibility schemes; VANET simulation; city maps. I. I NTRODUCTION Vehicular ad hoc networks (VANETs) are wireless communication networks that do not require any sort of fixed infrastructure, offering a novel networking paradigm to support cooperative driving applications on the road. VANETs are characterized by: (a) constrained but highly variable network topology, (b) specific speed patterns, (c) time and space varying communication conditions (e.g., signal transmissions can be blocked by buildings), (d) road-constrained mobility patterns, and (e) no significant power constraints. Deploying and testing VANETs involves high cost and manpower. Hence, simulation is a useful methodology tool prior to actual implementation [1]. Simulations of VANETs often involve large and heterogeneous scenarios. One of the important issues when creating a simulation environment in VANETs is to correctly model how 1