Federated Learning-Enabled Air Quality Monitoring System for Safe Driving in IoT-Integrated Vehicles

Abstract

This research introduces a connected car system that checks air quality and protects driver data through federated learning, thus increasing on-road safety. The system uses AirComp-based personalized federated learning which lets vehicles cooperate on air quality modeling while keeping their information private. Integrating edge computing and in-vehicle sensors in the method makes monitoring the air quality of both outside and inside the car possible in real time. Clever use of AirComp decreases communication between devices by gathering basic updates on the same topics across the entire network. The Flower framework is used because it supports both flexibility and training with a large number of participants in a decentralized way. Apart from keeping data secure, this architecture is able to work properly in many different settings and with different driving patterns. It gives alerts about air quality and suggests safe driving actions whenever pollution levels are high. They prove the performance, speed and ability of the proposed solution to deal with larger data.