Affordable new devices use artificial intelligence to monitor access points in real time

Air quality has become one of the most important public health problems in Africa. Poor air quality kills more people worldwide each year than HIV, tuberculosis and malaria combined. And that’s just the tip of the iceberg. Air pollution makes people less productive because they suffer from headaches and feel tired.

For example, India has poor air quality. The impact of poor air quality in India on its gross domestic product is about US$100 billion per year.

The health risks of poor air quality are known. However, setting up monitoring stations to measure it regularly has always been expensive.

I am a particle physicist and the director of the Institute for Collider Particle Physics at the University of the Witwatersrand and the iThemba Laboratories for Accelerator Based Sciences, a unit of the National Research Foundation in South Africa. As part of our technology transfer activities, we have launched the South African Consortium of Air Quality Monitoring, a multidisciplinary team of scientists who have strong links with policy makers and are interested in improving air quality in South Africa.

We set out to create, for the first time in South Africa, a cost-effective air quality monitoring system based on sensors, the Internet of Things and Artificial Intelligence. We called this system Ai_r.

Our 25-person team includes over 20 years of experience as particle physicists working with sensors, communications, and artificial intelligence. In particle physics, we create complex systems with different technologies and different disciplines. Particle physics includes, among others, electrical and computer engineering, data science, and artificial intelligence.

There are only 130 large air quality measurement stations in South Africa. They only measure air quality in the vicinity of the stations. Therefore, we need cost-effective, dense networks of Ai_r systems around these stations to measure air quality over a much larger area.

Our vision is to deploy tens of thousands of these devices across South Africa.

Use of artificial intelligence

Ai_r consists of a collection of small boxes that cost about $100 to make. The boxes can be mounted on the windowsill of any building, where they take air samples and send that data back to the cloud in real time. Modeling and predictions are made using AI connected to the cloud.

AI doesn’t do magic. It’s a collection of mathematical tools that scientists control to get the job done. It integrates data sets, learns from data, and creates automatic models. This saves a huge amount of resources.

In air quality monitoring, AI will learn from the vast wealth of data these sensors produce and then make predictions. For example, Ai_r will be able to tell us how weather changes affect air quality, so we know which areas will have more polluted air.

Without AI, it would be very difficult to create a cost-effective system that could monitor air quality and give us forecasts.

Air quality measurement in Johannesburg

The purpose of measuring air quality is to identify hotspots (places where the air quality is the worst), not just to determine the average air quality in a city.

The hotspot of very polluted air may be quite close to a cold spot where the air quality is better. For public health reasons, we need to know and be able to predict where the air quality will be bad on a given day. This is the only way we will be able to convince authorities to establish air pollution control measures in severely affected areas.

We have already introduced around 20 new devices in Soweto and Braamfontein in Johannesburg, with another 120 coming in the coming months. Both parts of the city have tens of thousands of vehicles passing through them every day, putting them at high risk of air pollution.

We have also placed air quality sensors at the University of the Witwatersrand and iThemba LABS. We have an agreement with the Gauteng Department of Education to implement them in schools in the Soweto area. Hospitals in the private group Netcare have also agreed to place air quality measuring devices.

Our website has a public panel where you can view all the data we collect.

The project was launched in June 2024. The government has responded very positively to this and we are currently integrating the system into the official South African Air Quality Scoring System, SAAQIS. We have also had strong support from the City of Johannesburg to roll it out further.

The Ai_r system measures particulate matter 2.5. These are extremely small particles of polluting matter such as dust, dirt, soot or smoke. The problem with these particles is that people inhale them. They enter people’s bloodstreams and can cause heart attacks, cancer and other diseases. That’s why particulate matter 2.5 is usually used as a benchmark for air quality worldwide.

How it works

The air quality measuring device is in a box. It has a small laser that emits light into the air. Depending on how the light scatters and reflects, it can measure the concentration of particulate matter. The measurements are then sent to a small computer inside the box, which communicates through an antenna every five minutes, via IoT technology, with a cloud-based system where the data is stored. So no one has to be there to collect air quality samples anymore.

We have also optimized the devices to use very little power. In the event of a power outage, the devices have economical batteries that last for several days. They are very easy to maintain. They are inexpensive to roll out throughout South Africa.

The existing network will continue to expand to become the largest, cost-effective, automated air quality measurement system in Africa by the end of 2024.

We really need to measure pollution and find out how bad it is so that we can design mitigation strategies that will target problem areas very effectively. The goal is to improve air quality for those most affected by air pollution.

This article is reprinted from The Conversation under a Creative Commons license. Read the original article.