iThermoFog: IoT-Fog based automatic thermal profile creation for cloud data centers using artificial intelligence techniques

Preventing failures in Cloud Data Centers (CDCs) due to high temperatures is a key challenge. Such centers have so many servers that it is very difficult to efficiently keep their temperature under control. To help address this issue, we propose an artificial intelligence (AI) based automatic scheduling method that creates a thermal profile of CDC nodes using an integrated Internet of Things (IoT) and Fog computing environment callediThermoFog. We use a Gaussian Mixture Model to approximate the thermal characteristics of the servers which are used to predict and schedule tasks to minimize the average CDC temperature. Through empirical evaluation on an iFogSim and ThermoSim based testbed and IoT based smart home application, we show that iThermoFog outperforms the current state-of-the-art thermal-aware scheduling method. Specifically, iThermoFog reduces mean square temperatures by 13.5%, while simultaneously improving energy consumption, execution time, scheduling time and bandwidth usage.