A case for good defaults: pitfalls in VANET physical layer simulations

Abstract

Network simulations are often the first choice to design, test, and evaluate novel applications and protocols for vehicular networks. Aiming for higher realism, simulators become increasingly complex, relying on detailed simulation models that are developed by different communities. With this trend, it also becomes difficult to understand all models in detail and researchers might lack the expert knowledge to parameterize such models properly. In this paper, we identify suboptimal default parameter values for physical layer effects in common simulation frameworks and show how they can negatively impact the results. We also review papers that use said simulation models and highlight that this is not simply a theoretical issue: We found that the majority of the papers simply copy these default parameter values or do not mention physical layer parameters at all. Both cases are clearly problematic. We thus argue that we should focus on reasonable default parameter values just as much as on the functional correctness of simulation models.