A virtual augmentation for air quality measurement sensor networks in smart cities

Distributed measurement systems are widely employed in many domains, particularly in the smart city domain. Anyway, because of reasons like expensiveness of devices and installation issues, the sensor coverage is often inadequate to the accomplish the goals of a modern smart city. To surpass such issues, a novel paradigm becomes attractive that, once a specific domain is given, allows the creation of a system capable to virtually augment the sensor network, i.e. providing measurement estimation where sensors are unavailable. This kind of approach s particularly focused on training a target Neural Network model, which outlines environmental issues, on urban areas provided with large sensor network, and then to make other areas, equipped with poor sensor networks, take advantage of the same model. The present work has been specialized on the domain of air pollution that, because of the complex urban environment and the huge costs of measurements devices, represents a case study extremely fitting the problem.