An Architecture for Intelligent Data Processing on IoT Edge Devices

As the Internet of Things edges closer to mainstream adoption, with it comes an exponential rise in data transmission across the current Internet architecture. Capturing and analyzing this data will lead to a wealth of opportunities. However, this ungoverned, unstructured data has the potential to exhaust the resources of an already strained infrastructure. Analyzing data as close to the sources as possible would greatly enhance the success of the IoT. This paper proposes a distributed data processing architecture for edge devices in an IoT environment. Our approach focuses on a vehicular trucking use case. The goal is to recreate the traditionally centralized Storm processes on the edge devices using a combination of Apache MiNiFi and the user's custom-built programs. Our approach is shown to preserve computational accuracy while reducing by upwards of 90 percent the volume of data transferred from edge devices for centralized processing.