Data analysis and AI implementation towards increased accuracy and selectivity of smart autonomous environmental sensors

Details

Supervisors

Legay, Axel ; Francis, Laurent

Faculty

Ecole polytechnique de Louvain

Degree label

Master [120] : ingénieur civil électricien, à finalité spécialisée

Abstract

In view of an advancing IoT market where things are getting ”smarter”, user expectations regarding air quality sensors have increased. More and more people want to track values via smartphones or tablets and want additional evaluations, analyses and alarms. On the other hand, manufacturers of existing infrastructure see a new opportunity to make their products smarter and to develop them by integrating sensors and IoT technology. Driven by the market requirement, the performance of the server has been advanced in terms of speed optimisation and scalability of the computational power. The benchmark performed in this work (figure 3.2) has shown that 30% of the total execution time, corresponding to data process, can be optimised. Thanks to a smart reshaping of the referring method, the total data treatment time for 200 data has been reduced of 9s. The flexibility of the server has been improved with Python extension, and auto-calibration mechanisms are integrated. The set of EnviCam® modules are nowadays automatically calibrated by smart algorithms which remove the offset of each sensor, and correct on the fly the effect of temperature and humidity; moreover the influence of parasitic gas will be corrected in a near future