Reproducing Consistent Wireless Protocol Performance across Environments

Abstract

Full scale experimentation with wireless networks in deployment environments is difficult, so a common validation technique is to test a prototype network in a convenient environment prior to deployment. In this paper, we consider the problem of obtaining comparable protocol performance when the test and deployment environments differ in RF propagation environment and/or inter-node spacing. To achieve comparable protocol behavior in the two settings, we propose the concept of “link usage spectrum”. Based on the hypothesis that the link usage spectrum is a gross predictor for network performance, we show how to replicate in the test setting the link usage spectrum of the protocol that is expected in the deployment setting. We illustrate our technique for achieving comparable protocol behavior via experiments and simulations in multiple indoor and outdoor propagation environments. The link usage spectrum is protocol specific; we illustrate for a family of protocols how the link usage spectrum is calculated analytically, from the protocol metric for choosing forwarding links in the network, and how power scaling can be used to match the link usage spectrum across networks.