Dynamic Session Layer Selection in IoT Actuation

Abstract

While most Internet of Things (IoT) solutions involve sensing, some of them also introduce actuation mechanisms. Specifically, devices interact with assets in the environment and transmit sensor readouts to applications that perform analytics. These applications typically reside on the network core and, in turn, process the readouts that trigger the transmission of actuation commands to the device. One important issue in these schemes is the nature of the communication channels. Most devices are wireless and therefore they are affected by the effects of signal multipath fading that results in application loss. More importantly, these impairments may cause actuation commands to be lost or to be critically delayed. In this context, several standard mechanisms have been proposed for the transmission of traffic from the application to the devices. They fall under two main architectural categories: (1) Representational State Transfer (REST) and (2) Event Driven Architecture (EDA). In this paper, we analyze two protocols associated with each of these two architectures by comparing them in order to assess their efficiency in IoT actuation solutions. This analysis leads to the development of a mathematical model that enables the dynamic selection of the right technology based on network impairments.