Mercury: A modeling, simulation, and optimization framework for data stream-oriented IoT applications

Abstract The Internet of Things is transforming our society by monitoring users and infrastructures’ behavior to enable new services that will improve life quality and resource management. These applications require a vast amount of localized information to be processed in real-time so, the deployment of new fog computing infrastructures that bring computing closer to the data sources is a major concern. In this context, we present Mercury, a Modeling, Simulation, and Optimization (M&S&O) framework to analyze the dimensioning and the dynamic operation of real-time fog computing scenarios. Our research proposes a location-aware solution that supports data stream analytics applications including FaaS-based computation offloading. Mercury implements a detailed structural and behavioral simulation model, providing fine-grained simulation outputs, and is described using the Discrete Event System Specification (DEVS) mathematical formalism, helping to validate the model’s implementation. Finally, we present a case study using real traces from a driver assistance scenario, offering a detailed comparison with other state-of-the-art simulators.