Analysis of Greedy, Semi-greedy, and Random Scheduling Heuristics with DVFS for Heterogeneous Fog Computing Platform

Internet of things (IoT)-based real-time applications, demanding energy-aware and low latency responses, are leading to a wider interest of researchers toward fog computing platforms to complement cloud computing by provisioning support for real-time computations near the end devices. However, the development of an optimized full-fledged fog platform, to meet latency, energy, and other quality of service (QoS) requirements of users and applications, is quite challenging. For real-time deadline-constrained tasks, energy efficiency and timeliness are two crucial parameters. Judicious scheduling of tasks can play a vital role in the effective utilization of such resource-constrained heterogeneous computing platforms. Greedy, semi-greedy, and random scheduling approaches are suggested by researchers to address energy efficiency and complexity-related issues. In this work, we employ dynamic voltage and frequency scaling (DVFS) technique on these three heuristics to ascertain its impact on reducing network energy consumption further without violating task deadline constraints. Simulations done on wide variety of task sets, on heterogeneous fog nodes, indicate that active energy consumption can be reduced with DVFS, with random heuristic exploiting the maximum benefit of DVFS followed by semi-greedy and greedy approach.