A Probabilistic Graphical Model for Social IoT-based Indoor Air Quality Monitoring in Smart Villages

The smart village is a promising approach for achieving socio-economic sustainability in rural areas. This paper utilizes Social Internet of Things (SIoT) methodologies to realize the smart village concept through efficient and cost-effective IoT technology. Each physical sensor and IoT device has a virtual counterpart Digital Twin (DT) at the edge for effective data analysis and optimization. For emerging public health services, monitoring indoor air quality (IAQ) in critical rural buildings is crucial. This paper proposes a probabilistic graphical model to capture IAQ changes using low-cost LoRa end nodes (EN) and gateway devices. These devices measure light intensity, temper- ature, and polluting gas concentration levels. The unsupervised k-means algorithm clusters the real-time IAQ data. At the same time, a Markov-based model visualizes and predicts IAQ changes. The model parameters are updated in real-time using data from a deployed LoRa wireless network. The framework was evaluated in rural areas near Ghaletol, Khuzestan province, Iran, with deployments in schools, agri- cultural warehouses, medical centers, and supermarkets. The best IAQ Markov states were 3 for schools, 3 for agricultural warehouses, 4 for medical centers, and 5 for supermarkets. For instance, the supermarket's IAQ model showed a polluting gas concentration of 862.6 ppm, an indoor temperature of 28.66°C, and a light intensity of 70.05 Lux.