Know your air before you breathe it.
AirWater Sense is a connected home sensor that predicts air pollution levels using artificial neural networks and the Internet of Things — so families can act before the air turns unhealthy.
Six pollutants that shape the air in our cities.
Particulate matter
Tiny dust and smoke particles small enough to be breathed deep into the lungs.
Dust · construction · trafficCarbon monoxide
A colourless, odourless gas that reduces the blood's ability to carry oxygen.
Vehicle exhaust · burning fuelCarbon dioxide
A greenhouse gas whose levels also signal how well a space is ventilated.
Combustion · crowded roomsSulphur dioxide
A sharp-smelling gas that irritates the airways and eyes.
Burning fuels with sulphurNitrogen dioxide
A traffic-related gas linked to breathing problems and asthma.
Road traffic · enginesGround-level ozone
Formed when sunlight reacts with other pollutants; irritates the lungs.
Sunlight + traffic emissionsTested on real data from three major cities in Jordan.
The neural network models were developed and evaluated for three major Jordanian cities. Prediction strength varied by pollutant — and the research is open about where it was weaker.
Carbon monoxide
CO showed the strongest relationship with weather and city factors.
Prediction accuracy varied for these pollutants.
Weakest correlation, because particulate matter comes from many different sources.
Why this suits a home device
The research presents ANN-derived equations as a computationally efficient forecasting tool for places where monitoring networks are limited or costly — exactly where a low-cost home sensor can fill the gap.
Source: Bani Hani et al., Environmental Engineering and Management Journal 24(8), 1657–1665 (2025). doi:10.30638/eemj.2025.129
Small enough for a shelf. Smart enough to forecast.
AirWater Sense brings the neural network forecasting model into a connected device built for homes.
Predicts, not just measures
Neural network models forecast pollutant levels instead of only reporting the present.
Internet of Things
Connects to the internet to combine its readings with local weather conditions.
Timely alerts
Warns the household when the air is forecast to become unhealthy.
Made for homes
Affordable local sensing where public monitoring stations are scarce.
Getting even sharper.
The research sets out how predictions can improve, especially for pollutants that were harder to forecast.
ANN forecasting of six pollutants from weather and city data.
Adding further explanatory variables to the model.
Expanding data across longer periods and more locations.
Combining ANNs with other machine learning methods.
Engineers who design it and build it.
AirWater Sense was founded in 2025 by a research team uniting environmental engineering and modelling with the hands-on skills to manufacture the device.
Faculty of Engineering
Develops the neural network models and the environmental science behind the pollutant forecasts.
Faculty of Engineering Technologies
Designs and manufactures the sensor hardware, electronics and IoT connectivity.
Frequently asked questions
Which pollutants does AirWater Sense predict?
Six: particulate matter (PM10), carbon monoxide (CO), carbon dioxide (CO2), sulphur dioxide (SO2), nitrogen dioxide (NO2) and ground-level ozone (O3).
How does it predict pollution?
An artificial neural network learns how wind speed, temperature, humidity, population density and vehicular traffic relate to pollutant levels, and uses those relationships to forecast them.
How accurate is it?
Accuracy depends on the pollutant. Carbon monoxide was predicted best, with an R² approaching 1 and a mean squared error of 7.2 × 10⁻³. PM10 was the hardest, because particles come from many different sources.
Where was the model tested?
On data from three major cities in Jordan.
Why use neural network equations?
They are computationally efficient, which makes forecasting practical on affordable devices in places where official monitoring networks are limited or costly.