AI-Enhanced IOT for Air Quality Forecasting: Offloading LSTM Predictions to Edge Servers

Abstract

Air quality forecasting is critical for dealing with pollution's health and environmental implications, but resource-constrained IoT devices struggle to execute complicated predictive analytics locally. This study presents a distributed architecture-based AI-enhanced Internet of Things system for real-time air quality index (AQI) predictions. An edge server receives Long Short-Term Memory (LSTM) predictions from a Raspberry Pi-based system that gathers multi-sensor data (PM2.5, PM10, CO, temperature, and humidity) through a RESTful API. Sensor data and the predicted AQI are shown in real time on a Node-RED dashboard. According to experimental findings, offloading achieves scalable and effective air quality monitoring by reducing the computing load of the IoT device by 85%. By bridging IoT and AI, this hybrid edge-cloud strategy provides a workable solution for intelligent environmental systems.