Hybrid Machine Learning and Federated Learning Framework for Intelligent Task Scheduling, Load Balancing, and Resource Allocation in Fog Computing

Abstract

Fog computing has become a successful technology which processes data at edge locations for Internet of Things (IoT) systems that require quick response times. The fog environment encounters challenges in task scheduling and load distribution and resource management because of its fluctuating operational demands and its various node operational strengths and its limited resource capacity and increasing privacy concerns. The research paper presents a Federated Learning system which uses machine learning with hybrid technology to manage resources efficiently in fog computing environments. The hybrid model uses predictive algorithms together with classification algorithms to forecast workload requirements and determine task importance and assess fog node capabilities which support better scheduling choices. Through Federated Learning multiple distributed fog nodes can train a global model together without sharing their original local data which protects user privacy and minimizes required data transfer. The trained model establishes a dynamic scheduling system which allocates tasks while balancing workloads and adapting to network changes in real time. The proposed framework operates to reduce latency while increasing throughput and achieving maximum resource efficiency and delivering improved system scalability. The experimental results show that our approach achieves better performance than traditional resource management methods which improve response times and load distribution and system performance to meet the requirements of advanced smart technologies used in healthcare and smart cities and industrial automation.