Context-Aware Air Quality Analytics Using Low-Cost Sensors and First-Order Logic

Low-cost sensors changed the pollution monitoring paradigm due to their ability to measure air pollution ad hoc, close to the sources and in great spatio-temporal resolution. Our team has been engaged in the development of highly reliable air quality sensing devices, using low-cost sensors. Air quality measurements are transmitted to a scalable dynamic knowledge-base maintenance system for representing and reasoning with knowledge about Sentient Computing, extended with air quality predicates. The system functions similarly to a two-level cache: the lower layer maintains knowledge about air quality at the sensor level by continually processing a high rate of observations from the measuring devices. The higher layer maintains easily-retrievable, user-defined abstract knowledge about current and historical states of air quality along with temporal properties such as the time of occurrence of observation timeseries and their length. Our approach uses deductive systems in an unusual way: by creating a model of real-world air quality we are able to prove quality aspects of logical predicates both at the sensor level - such as the existence of outliers, stuck-at-zero values or whether a timeseries is additive or multiplicative - and at the derived predicate level - such as AQI index, a person’s aggregated exposure to pollutants or analytical trends and patterns, paving the road to explainability. Furthermore, by embedding the lower layer of the architecture inside the monitoring devices we support in-network processing, lowering the number of published messages and extending the device lifetime. The preliminary results provide useful insights on both pollution level and the intensity of industrial activity in the dairy.