Virtual Light Sensors in Industrial Environment Based on Machine Learning Algorithms

Internet of Things (IoT) has become the backbone of current and future emerging applications both in the public and the private, industrial sector. The IoT paradigm, enhanced with intelligence and big data analytics, has found applications in a wide range of solutions such as smart home, smart city, industrial IoT etc. Even though IoT implies that cheap motes can conduct a specific task, thus a large number of them can be deployed, we aim to minimize the installed hardware while we still have a high level of quality of service. Machine Learning algorithms can support this challenge by generating virtual data via utilization of real data from a smaller subset of sensors. The generated data can replicate sensor behavior which would otherwise be difficult or impossible to track. It is also possible to use simulation models for data analysis model validation, by generating new data under varying conditions. In this paper, we propose a concept of an IoT testbed which allows virtual IoT resources to be immersed and tested in real life conditions, which are met in everyday life. Additionally, the features of the implemented testbed prototype are discussed while taking into account specific use cases, regarding luminosity scenarios in industrial environments.