Cloud vs Fog: assessment of alternative deployments for a latency-sensitive IoT application
Marcus N. Gomes Jr., Miguel L. Pardal · Procedia Computer Science · 2018
Internet of Things (IoT) smart places are systems composed of sensors, actuators and computing infrastructure that acquires data about the surrounding environment and uses that data to improve the user experience of the smart place. For instance, RFID readers can detect a tag approaching and, after the event is processed in a dedicated server, open a door automatically. Many IoT applications are latency-sensitive because actions need to be done in a timely manner. To meet this requirement these applications are usually provisioned close to the physical place, which represents an infrastructure burden because it is not always practical to deploy a physical server at a location. Utility computing in the Cloud can solve this issue but the latency requirements must be carefully assessed. Fog computing is a concept that brings the cloud close to devices at the edge of the network, aiming to provide low latency communication for applications and services. The present work implemented a provisioning mechanism to deploy a “smart warehouse” IoT application according in utility computing platforms: Cloud and Fog. We compared the event latency performance of both approaches and the results show that a fog deployment is more adequate for the considered IoT application.