On the Initial Network Topology Factor in Mobile Ad-Hoc Network

Abstract

The impact of the initial network topology on performance of routing algorithms is explored. Typically researchers use a randomly chosen network topology for performance evaluation of their protocols and algorithms. Here we show that the initial network topology can have a significant impact on algorithm performance and can lead to biased results, in particular, an initial topology that includes a major connectivity obstacle such as low connectivity level (e.g., a tree topology) or bridges. Although users move according to commonly implemented random mobility models, the effect of the initial topology can persist over time. To avoid biased results we recommend using multiple initial topologies instead of one, and/or running the simulation in an initialization phase until the effect of the initial topology fades.