Data-Driven Generation of Services for IoT-Based Online Activity Detection

Business process management (BPM) technologies are increasingly adopted in the Internet of Things (IoT) to analyze processes executed in the physical world. Process mining is a mature discipline for analyzing business process executions from digital traces recorded by information systems. In typical IoT environments there is no central information system available to create homogeneous execution traces. Instead, many distributed devices including sensors and actuators produce low-level IoT data related to their operations, interactions and surroundings. We leverage this data to monitor the execution of activities and to create events suitable for process mining. We propose a framework to generate activity detection services from IoT data and a software architecture to execute these services. Our proof-of-concept implementation is based on an extensible complex event processing platform enabling the online detection of activities from IoT data. We use a running example from smart manufacturing to showcase the framework.