Sensors and MAchine learning for RemoTe Smart EnviroNmental monitoring SystemS (SMARTSENS)
This project aims to design an innovative IoT measurement network using gas micro-sensors. This involves integrating Machine Learning techniques to improve the accuracy of air quality measurements associated with the monitoring of industrial emissions, control processes and safety. The targeted network will decentralize these Machine Learning techniques to implement them at the Edge Computing level, in order to place the measurement functions at the local level. This objective contains two components, studied respectively by the two project partners: 1- Development of Machine Learning models by UNamur. A methodology giving models adapted to both micro-sensors and the hardware constraints of computing resources is targeted. The models will be designed taking expert knowledge into account by integrating constraints on, for example, the evolution of the quantity measured in relation to the temperature/humidity. 2- Development of the network architecture supporting these models by CeREF. This network must integrate Edge modules allowing the local use of Machine Learning models developed by UNamur. The use of Machine Learning in the Edge context involves innovation in terms of optimizing energy consumption and software libraries in order to target the use of energy harvesting and energy autonomy.