International Journal of Computer Applications
Foundation of Computer Science (FCS), NY, USA
|
Volume 187 - Issue 23 |
Published: July 2025 |
Authors: Benjamin Aidoo, Frederick Kwame Minta, Abdul-Aziz N-Yo, Derrick Attoh Tettey, Osbert Kasiimbura, Albert Essilfie |
![]() |
Benjamin Aidoo, Frederick Kwame Minta, Abdul-Aziz N-Yo, Derrick Attoh Tettey, Osbert Kasiimbura, Albert Essilfie . A Scalable IoT–Cloud Architecture with Deep Q-Learning for Air Quality Monitoring and Alerting. International Journal of Computer Applications. 187, 23 (July 2025), 44-51. DOI=10.5120/ijca2025925461
@article{ 10.5120/ijca2025925461, author = { Benjamin Aidoo,Frederick Kwame Minta,Abdul-Aziz N-Yo,Derrick Attoh Tettey,Osbert Kasiimbura,Albert Essilfie }, title = { A Scalable IoT–Cloud Architecture with Deep Q-Learning for Air Quality Monitoring and Alerting }, journal = { International Journal of Computer Applications }, year = { 2025 }, volume = { 187 }, number = { 23 }, pages = { 44-51 }, doi = { 10.5120/ijca2025925461 }, publisher = { Foundation of Computer Science (FCS), NY, USA } }
%0 Journal Article %D 2025 %A Benjamin Aidoo %A Frederick Kwame Minta %A Abdul-Aziz N-Yo %A Derrick Attoh Tettey %A Osbert Kasiimbura %A Albert Essilfie %T A Scalable IoT–Cloud Architecture with Deep Q-Learning for Air Quality Monitoring and Alerting%T %J International Journal of Computer Applications %V 187 %N 23 %P 44-51 %R 10.5120/ijca2025925461 %I Foundation of Computer Science (FCS), NY, USA
Air pollution poses a growing threat to public health in rapidly urbanizing cities, particularly in Sub-Saharan Africa, where real-time monitoring infrastructure is limited. This paper presents the design and implementation of a scalable IoT and cloud-based architecture for air quality monitoring and intelligent alerting in Accra, Ghana. The system integrates low-cost ESP32-based sensor nodes with a Deep Q-Network (DQN) to classify pollution severity and issue adaptive, context-aware alerts. Eight key environmental parameters, including PM1.0, PM2.5, PM10, VOCs, CO, LPG, temperature, and humidity, are continuously monitored and analyzed using cloud-based processing. Real-time data is visualized through a web dashboard, while critical alerts are disseminated via SMS to ensure user accessibility. The DQN agent supports decision transparency through Q-values, feature importance, and temporal trend analysis. Experimental results demonstrate a training accuracy of 89% and a field test classification accuracy of 82.9%, confirming the system’s effectiveness for scalable, real-time, and interpretable environmental health monitoring in resource-constrained settings.