Speaker
Description
Air pollution poses a severe and growing public health challenge globally, with the World Health Organization estimating approximately 4.2 million premature deaths annually attributable to ambient air pollution.
Traditional monitoring infrastructure comprised of large, fixed reference
stations is costly, spatially sparse, and incapable of real-time,
high-resolution pollution mapping. This creates critical data gaps,
particularly in urban centres and resource-constrained settings across
Africa and the Global South.
The South African Consortium of Air Quality Monitoring (SACAQM), an
initiative led by Professor Bruce Mellado of the University of the
Witwatersrand and iThemba LABS, addresses these limitations through the
development and deployment of the $AI\_r$ system: a low-cost, IoT-enabled air quality monitoring
node. The system integrates the Sensirion SEN55 environmental sensor which is capable of measuring $\mathrm{PM}_{1}$, $\mathrm{PM}_{2.5}$,
$\mathrm{PM}_{4}$, $\mathrm{PM}_{10}$, VOC, $\mathrm{NO}_{x}$,
temperature, and relative humidity, with the Nordic Semiconductor
nRF9160 system-in-package, running on the Zephyr RTOS. Data transmission
is achieved via LTE/4G for real-time cloud connectivity, with LoRa communication additionally integrated to extend deployment to areas lacking cellular coverage.
This paper reviews the architecture and performance trade-offs of LTE/4G
and LoRaWAN communication protocols in the context of low-cost IoT air
quality monitoring, drawing on the SACAQM deployment experience. LTE/4G offers low latency (30-100 ms) and high throughput, suited to continuous real-time data streaming, whereas LoRaWAN provides long-range, ultra-low-power communication at reduced data rates (0.3-50 kbps), enabling deployment in remote or infrastructure-limited environments. A hybrid communication strategy leveraging both technologies is proposed as an optimal framework for scalable, wide-area monitoring networks. The $AI\_r$ system is calibrated against South African Air Quality
Information System (SAAQIS) reference stations, with active deployments
across Soweto and Braamfontein in Johannesburg, and systems shipped to
partner institutions all over the world. Integration of machine learning models, including
AI-driven $\mathrm{PM}_{2.5}$ forecasting and anomaly detection, further enhances the system's predictive and public health utility. The SACAQM initiative demonstrates that cost-effective, IoT-based,
AI-integrated sensor networks represent a viable and scalable approach
to closing the air quality monitoring gap in low- and middle-income
countries.
| Apply for student award at which level: | MSc |
|---|---|
| Consent on use of personal information: Abstract Submission | Yes, I ACCEPT |