LETO: An Efficient Load Balanced Strategy for Task Offloading in IoT-Fog Systems

Abstract

The resource-constrained IoT devices often offload tasks to Fog nodes (FNs) owing to the intermittent WAN delays and multi-hopping by executing at remote cloud servers. An efficient allocation strategy satisfies the users' requirements by ensuring minimum offloading delays and provides a balanced assignment from the service providers' (SPs) viewpoint. This paper presents a model called LETO that reduces the total offloading delay for real-time tasks and achieves a balanced assignment across FNs. The overall problem is modeled as a one-to-many matching game with maximum and minimum quotas. Owing to the deferred acceptance algorithm (DAA) inapplicability, we use a proficient version of the DAA called multi-stage deferred acceptance algorithm (MSDA) to obtain a fair and Pareto-optimal assignment of tasks to FNs. Extensive simulations confirm that LETO can achieve a more balanced assignment compared to the baseline algorithms.