An Orchestrator Framework for IoT-based Disaster Prevention Simulation

Human casualties from flooding are caused by various factors, such as damage caused by flood water, delays in evacuation due to road flooding or communication network failures, or accidents during evacuation. However, conventional simulations (e.g., river flooding, road flooding, and communication failures) are developed as proprietary systems, and it is difficult to use them to estimate and predict phenomena or casualties related to flood damage. This paper proposes an orchestrator framework for IoT-based disaster prevention simulation, which simulates urban flooding, numbers of affected people, and network failures in an integrated manner. We develop an orchestration scheme that coordinates several independent simulators into one simulation to observe mutually related phenomena (e.g., evacuation behavior in a flood simulation). Our orchestration scheme deploys the simulators, data exchange system, and virtual clock system on a Kubernetes cluster and controls the synchronization and execution order of these simulators. We describe our orchestration scheme architecture, automatic determination of the execution order, and detection of misconfigurations for the continuous operation of multiple simulators.