FogR: A highly reliable and intelligent computation offloading on the Internet of Things

Abstract

Cloud Computing is inevitably going to meet the challenges of an unprecedented scale of Internet of Things. Fog Computing was introduced as a consequential and partial augmentation of Cloud Computing where the latter, by design, struggles to deal with the demand for, among many others, lower latency, decentralized management, and resiliency, especially in the case of unreliable connectivity to the Cloud. Fog Computing takes place at the edge of the network around a resource-poor Fog Node, and the network of user devices, sensors, and actuators connected to it, which may support Fog-aware applications hosted on them, referred to as Droplets. In search of a language independent Remote Procedure Call middleware between Fog Node and the Droplets, we found that besides executing tasks, Droplets can be built context-aware and participatory to the Fog Node's decision-making process while offloading computation tasks, observing the premises of Fog as stated above. We developed FogR, a lightweight and reliable architecture that can construct a more intelligent Fog network for computation offloading. We also show how enabling intelligence distribution of such nature can be applied in the case of Smart Traffic System of the future to respond reliably to an emergency.