Experimental design, modeling and testing of geographic routing protocols

Abstract

The purpose of this paper is to introduce techniques appropriate for the process of experimental design, modeling and testing of geographic routing protocols. The stages of the experimental design process are outlined along with a discussion of experimental validity including internal validity, external validity, use of a control, and confounds in the test process illustrated with examples from current research. The difficulty of selecting an appropriate network simulation platform is discussed. The decision to write a custom geographic routing network simulation application is justified and the resulting application is presented. Selection of appropriate methods of statistical analysis including parametric tests, non-parametric tests, confidence intervals, and descriptive statistics is discussed. To support this analysis a custom graph/statistics software application is also presented.