AI_r: Transforming Air Quality Monitoring through Cost-Effective AI Solutions

Abstract

Air quality monitoring is vital for public health, particularly in resource-limited areas. However, traditional monitoring systems are often costly, leading to inadequate data and unequal access to air quality information. To address this issue, the South African Consortium of Air Quality Monitoring (SACAQM) has developed a more affordable air quality monitoring system tailored to resource-constrained regions. This system uses IoT technology and cost-effective sensors to create a wide network that provides real-time data on air pollution. The network consists of grouped Wireless Sensor Networks (WSNs) that communicate using LoRa and LTE technologies. The data is stored in a NoSQL database and is easily accessible through a user-friendly dashboard. Calibration against existing air quality systems ensures the data’s accuracy and reliability. Additionally, the system employs Artificial Intelligence (AI), particularly Graph Neural Networks (GNNs), to enhance air quality modeling and prediction capabilities. After a successful pilot deployment at schools in Soweto, Johannesburg, the system is now being expanded to hospitals and other community hubs. This expansion underscores the system’s potential to democratize access to critical air quality data, aiding public health strategies and improving air quality in vulnerable communities.