Experimenting with a Multi-Radio Mesh Networking Testbed

Experimentation is an important next step to take in the study of multi-radio mesh networks, but it involves many obstacles preventing a stable and reproducible testbed environment. Using off-the-shelf equipment we demonstrate the impact that specific design choices for access points may have in the performance of a 2-hop network. We find that only after careful calibration with simulation and analytical results can we identify baseline node configurations with predictable behaviour. In particular, we observe that multi-radio wireless platforms face limitations due to increased interference among the several radios inside a single node. Because of this, our mesh routers cannot hold more than 2 radios while requiring a minimum antenna separation of 35db. Throughout this paper we present a methodical approach that, despite being simple, offers a way in which wireless testbeds can be calibrated before experimental deployment and evaluation.