Sequential Randomization load balancing for Fog Computing

Fog Computing is considered a key enabler for meeting the computation requirements of the billions of objects or Things expected to be deployed in the near future. Fog nodes can be viewed as mini-clouds deployed close to the end users, that complement the current big but far cloud paradigm. Although load balancing among fog nodes is a poorly addressed topic, it may improve the capacity of fog nodes to deliver computation service. In this paper we study load balancing among fog nodes, addressing the specific problems arising from the fog model. In particular, we study how to exploit randomized based load balancing protocols leveraging the power-of-random choice property. We propose sequential probing in contrast to the classical randomization protocols based on parallel probing. We show by a mathematical analysis and simulations the superiority of the proposed solution.