From a Real Deployment to a Downscaled Testbed: A Methodological Approach

This paper proposes a novel methodology for the spatial downscaling of real-world deployments of wireless networks, running protocols, and/or applications for the Internet of Things (IoT). These networks are often deployed in environments not easily accessible and highly unpredictable, where doing experiments is very expensive and time consuming. The latter calls for the need to develop downscaled testbeds, deployed in controlled environments, where tests can be conducted under predictable conditions. This paper presents a methodology to realize the downscaling of a real deployment on an experimental platform, called controllable testbed that has a much larger number of nodes with respect to the real one. The downscaling procedure proposed is based on the identification of the most appropriate subset of nodes of the controllable testbed, to be used to reproduce the channel gains between each node pair in the real world. The latter, in fact, results in obtaining the same network topologies, bringing to the same performance on average, when the same protocol stack and software are run. After the description of the procedure, an example of its implementation is provided. Comparison of results, in terms of packet loss rate (PLR), network throughput, and topologies achieved on the downscaled testbed and on the real-world deployment, is given; results show a very good fit and demonstrate the efficacy of the proposed methodology.