Fog-based IoT Framework for Large-Scale Data Classification

FOG-BASED IOT FRAMEWORK FOR LARGE-SCALE DATA CLASSIFICATION Marc Jayson Baucas University of Guelph, 2019 Advisor: Professor Petros Spachos Internet of Things (IoT) is a network that is used to develop large-scale data sensing applications. As networks grew, servers could no longer handle the processing constraints. As a result, processes were reallocated to the end devices, which led to high power consumption. Then, fog computing was incorporated to offload processes between the cloud and the end device into an intermediary device. However, an unoptimized fog-based IoT network can still run into the same issues. Therefore, this thesis aims to propose an architectural template or framework that can be used to balance the fog-based IoT network for large-scale data sensing. For feasibility testing, Urban sound classification was the selected application. Therefore, reallocation and active power states were included as techniques in configuring the framework. The results optimized the framework and created a reasonable balance between cloud and fog-based networks that demonstrated low power consumption and lesser server strain.