Experimental method for testing networks

Abstract

To address problems of scale and validity in testing networks, we propose application of experimental method in the testing process. This means that testing will be an iterative process of running tests, developing models of the network, examining properties of the models to make predictions about the network, and validating (or invalidating) the predictions with further tests. We also propose the use of self-similar structures in networks to reduce the scale of the test effort. Identification of self-similar structures in a network allows the tester to understand the behavior of large networks by experimenting on smaller networks.