Border Effects in the Simulation of Ad Hoc and Sensor Networks

In ad hoc and sensor networks, spatial communication constraints have a tremendous influence on many aspects of network performance. The network behavior becomes easily unpredictable with an increasing number of nodes and connections. Therefore, simulation is the evaluation method of choice for ad hoc and sensor networks. Simulations of wireless networks, however, encounter two challenges: First, computation and time constraints lead to the use of ldquosmallrdquo values for simulation parameters. Researchers then tend to simply extrapolate these results for larger values. Second, the simulation model is an abstraction of the real world, often not including the effects of borders and obstacles. The contribution of this work is a technique to deal with both challenges, namely a mechanism to correct ldquosmall sizerdquo simulations to make them amenable to extrapolation via a correction that adds the border effects to the simulation results. The system model used is based on static networks and the unit disc graph assumption.