A Novel Matching Theory-Based Data Offloading Framework for a Fog Network With Selfish and Rational Nodes

Abstract

Fog Computing (FC) is an emerging paradigm that brings the processing capability of the cloud closer to the end devices. Each processing entity, known as the fog node, is owned by a Fog Service Provider (FSP) and has a set of IoT users subscribed. A fog node can be under-utilized or over-utilized based on the processing resources and the number of subscribed IoT users. The data packets of the over-utilized fog nodes can be offloaded to the under-utilized fog nodes for processing. This is beneficial for both the source and destination fog nodes. While the destination fog node has monetary advantages due to the leasing of its processing resources, the offloading fog node has benefits in terms of latency satisfaction. Our objective is to maximize the monetization of the processing resources of the fog nodes without affecting the Quality of Service (QoS) constraints of the subscribed IoT users. To achieve our objective, we propose a Matching Theory-based offloading (MAT) framework based on many-to-many matching without externalities. Additionally, we prove the convergence and stability of our framework, and also demonstrate its effectiveness. We further show the efficacy of our approach over existing models through extensive simulations.