Analysis of vehicular mobility patterns on routing protocols

In this report, we illustrate how the realistic motion patterns introduced
VanetMobiSim [1] affect the velocity, and how new parameters become necessary
to evaluate the performance of routing protocols in Vehicular Ad Hoc
Networks (VANETs). To express our point, we evaluate the performance
of AODV with realistic urban scenarios. We show how new urban specific
parameters have significant impacts on routing, and de-facto replace some
non-urban specific parameters. For example, the average velocity appears
to be irrelevant in urban scenarios and should be replaced by road segment
lengths. Then, we evaluate AODV and OLSR performance in realistic urban
scenarios. We study those protocols under urban-specific metrics such as
road segment length, and cluster effect, or non-urban specific metrics such as
vehicle density, and data traffic rates. We show that clustering effects created
by cars aggregating at intersections have remarkable impacts on evaluation
and performance metrics. We conclude that OLSR is a better candidate than
AODV for routing in VANET in urban areas.