Impact of slow and fast channel fading and mobility on the performance of AODV in ad-hoc networks

Abstract

Network simulation tools are commonly used to analyze the performance of Mobile Ad-hoc Networks (MANETS) protocols and their applications. The radio propagation models used for these simulations strongly influence the results. Most of the MANETS routing protocol performance studies have used unrealistic (for urban structures) radio propagation models such as the Two-Ray Ground (TRG) or Free Space model for path loss. In this study, the performance of the ad-hoc on demand distance vector (AODV) routing protocol has been investigated for two different types of fading (Lognormal Shadowing & Rayleigh) typical of those found in urban environments. The effect of four different mobility models, Random Waypoint (RWP), Reference Point Group Mobility (RPGM), ManhattanGrid (MG), GaussMarkov (GM) and changing node density have also been considered. The results indicate that slow fading (i.e. Lognormal Shadowing) channel degrades AODV performance much more severely than fast fading for the tested cases. The performance of AODV is best when the nodes tend to cluster (i.e. with RPGM), while it is particularly poor for both GM and MG models (i.e. when the nodes are farther apart).