A Model for Power-Performance Optimization in Fog-Cloud Environment by Task Off-Loading of IoT Applications

Abstract

Task offloading is a solution to compensate for resource constraints on the Internet of Things (IoT). Deciding on the location of offloading is very important. The IoT systems provide a three-tier (IoT-fog-cloud) architecture and use the locations of cloud and fog for task offloading. Fog is a more suitable location for task offloading than cloud in terms of energy consumption and response time, and this paper aims to optimize these criteria in IoT systems. In this paper, fog is modeled by queuing theory, and the minimum number of its servers is determined based on its availability by the binary search algorithm and reinforcement learning policy iteration algorithm. Different scenarios are considered for evaluating the impact of different parameters on the cost of the fog. The proposed dispatch policy improves the results by 31% compared to the policies of Slowest Server First, Fastest Server First, and Randomly Chosen Server.