FURI | Fall 2019

Designing Better Wireless Experiments through Screening

Today’s computer networks are complex engineered systems. Due to the large number of parameters available, an important question is Which parameters should be selected for experimentation?  Screening experiments are conducted to answer this question. The approach used here is based on a locating array, a new combinatorial design whose size grows logarithmically. We set up an experiment comparing two MAC protocols: The 802.11 Wi-Fi standard and an auction-based protocol called REACT. Audio is sent between nodes and quality-of-service (QoS) metrics are measured. The goal is to determine the influential parameters in the experiment so that they can be used in follow-on experimentation.