DeepWFFS: Enhancing Fog Computing Efficiency Through Multiqueue Architecture and Intelligent Controller for Task Prioritization

This article introduces an innovative multiqueue fog architecture coupled with an intelligent controller, aimed at enhancing the efficiency and adaptability of fog computing environments. Unlike conventional single-queue fog architectures that typically rely on basic first-in-first-out (FIFO) task execution models in fog servers, our approach offers heightened granularity and flexibility in task scheduling. This feature enables effective task management, catering specifically to Internet of Things (IoT) applications characterized by varying degrees of time-sensitivity and resource requirements. Our proposed deep weighted-fair fog servers (DeepWFFS) scheme comprises two key elements: 1) the weighted-fair fog server (WFFS) framework and 2) an intelligent deep controller (DC) leveraging deep reinforcement learning (DRL) for task prioritization. The WFFS framework adopts multiple queues within fog servers, each assigned a predefined weight representing task priority. This prevents task starvation and promotes equitable task execution. Meanwhile, the DC continuously monitors task workloads and priorities, ensuring optimal task allocation to the most suitable queues within fog and cloud servers. Through simulation, our results exhibit the superior performance of DeepWFFS compared to benchmark schemes. This advancement showcases the potential of our architecture to efficiently manage diverse tasks in fog computing environments.