An Integrated Fog Computing Approach to Improve Quality of Service in IoT Environment with Dynamic Task Scheduling and Resource Allocation

Exponential growth in IoT application involves managing latency, resource utilization and reliability in traditional cloud based architectures. In this research, we present a dynamic IoT-Fog-Cloud architecture using fuzzy logic, DQN and MOEA/D to optimize QoS metrics in real time. The real-time tasks generated by simulated IoT devices are falling through fuzzy logic at the Fog Master node and classified according to urgency. DQN schedules tasks dynamically so that during runtime it discovers optimal tasks solution on a set of Fog nodes. Resource allocation is enhanced by MOEA/D balancing computational loads with trade-off between latency, reliability and resource utilization. Here, we compare static and cloud-only architectures and show comparative simulations in iFogSim tool that achieve 28% reduction in latency, 19% increase in reliability and 22% improvement in resource utilization. The system’s hierarchical adaptive nature provides for an efficient task processing that is scalable and with superior QoS for IoT applications. The proposed framework is a scalable, robust and efficient means by which one can run real-time IoT tasks in fog-cloud computing environments by synergizing advanced deep learning and evolutionary algorithms.