Latency-Aware Federated Learning Framework for Efficient Communication in Internet of Vehicles (Iov)
Latency-Aware Federated Learning Framework for Efficient Communication in Internet of Vehicles (Iov)
1Nagelli Akshitha
Computer Science and Engineering Hyderabad Institute of Technology and Management
Hyderabad, India 22e51a0580@hitam.org
4Peddi Pranay
Computer Science and Engineering Hyderabad Institute of Technology and Management
Hyderabad, India 22e51a0594@hitam.org
2rMsSure Mamatha
Assistant Professor Computer Science and Engineering Hyderabad Institute of Technology and Management Hyderabad,India sure.mamatha@gmail.com
5Shaik Shaheen
Computer Science and Engineering Hyderabad Institute of Technology and Management
Hyderabad, India 22e51a05b1@hitam.org
3Patlolla Sanjanai
Computer Science and EngineeringHyderabad Institute of Technology and Management
Hyderabad, India 22e51a0593@hitam.org
6Sunki Reddy Ajay Reddy
Computer Science and Engineering Hyderabad Institute of Technology and Management
Hyderabad, India 22e51a05b7@hitam.org
Abstract: Internet of Vehicles (IoV) focuses on enhancing communication efficiency and model performance in distributed vehicular networks. Federated Learning (FL) allows multiple vehicles to collaboratively train machine learning models without sharing raw data, thus preserving privacy and security. However, IoV environments experience latency bottlenecks caused by intermittent connectivity, heterogeneous devices, and uneven data distribution.These two reasons slow down the model aggregation and degrade the learning performance. To mitigate the above problems, a adapted model update aggregation is applied, which dynamically adjusts to the network changing conditions. Methods including asynchronous model updates, edge caching, and adaptive communication scheduling are leveraged to reduce latency and enhance real- time responsiveness. It also improves the convergent speed, scalability, and reliability of FL in vehicular edge networks. The optimized architecture enables speedy and accurate decisions in applications such as autonomous driving, traffic management, and intelligent transportation systems with privacy preserving mechanisms and efficient network utilization.