A Traffic-Aware Task Mapping Problem in Fog-Cloud of Things

The Internet of Things (loT) is rapidly advancing, but its limited resources are often supplemented by integrating fog computing to address these constraints. Despite this, fog devices encounter challenges such as heterogeneity, widespread distribution, dynamic nature, and resource limitations. To tackle these issues, a robust task scheduling strategy is crucial for efficiently utilizing fog resources and enhancing Quality of Service (QoS). Moreover, tasks generated are dynamic in nature with disparate requirements such as deadline, cost, priority, etc. Fog resources could be utilized effectively with proper mapping of tasks based on ingress traffic on each node to meet the requirements. This research formulates the task mapping problem (TMP) with the goal of minimizing cost while improving QoS by reducing response times and deadline violations for loT tasks. It uses SVM (Support Vector Machine) to predict fog node traffic and categorizes nodes into peak-traffic and low-traffic groups. The Quantum-inspired PSO-based TM Algorithm (QPSO) has been implemented to map the tasks on these nodes. Simulation results reveal that the proposed method offers significant improvements, achieving up to 19.66% better response times, 30% cost reduction, and a 26% deadline violation rate i.e., an increase in meeting task deadlines compared to other approaches.