Federated Learning for Privacy-Preserving Air Quality Forecasting using IoT Sensors

Abstract

Air quality forecasting is crucial for public health and urban planning. However, traditional machine learning models face challenges with centralized data collection, raising privacy and security concerns. Federated learning (FL) offers a promising solution by enabling model training across decentralized data sources while preserving data privacy. This study presents an FL framework for predicting the Air Quality Index (AQI) using data from many Internet of Things (IoT) sensors deployed in urban areas. The proposed FL framework facilitates model training using diverse sensor data while maintaining data privacy at each source. Local computational resources at the sensor level are used for initial data processing and model training, with only model updates shared centrally, reducing data transmission requirements. The FL model achieved comparable accuracy to centralized approaches while enhancing data privacy. This work represents a significant advancement for smart city initiatives and environmental monitoring, offering a scalable, real-time, and privacy-aware framework for air quality monitoring systems that leverage IoT technology.