Automation of experiments
One click, one experiment
This project was carried out as part of an internship at the Research Centre for Automatic Control (CRAN) for SenseVid, a video transmission evaluation software developed by my internship supervisor. My mission was to automate the various stages allowing to carry out video transmission experiments on sensor networks and to extract the results.
So, I created several Python clients capable of launching bursts of real experiences on IoT-Lab, a large French platform of sensor networks, but also simulations on a simulator called Cooja.
A unique interaction with one of these customers allows you to instantly start the experiment with full control of the environmental parameters. These experiments are now done more intuitively with a simplified, filtered and more global visualization in a single terminal and with additional features. In particular, we can visualize various elements such as topological trees of the network in real time during the experiment and retrieve results such as video reconstruction, statistics and evaluation metrics and graphs of energy consumption.