Dracena: A Real-Time IoT Service Platform Based on Flexible Composition of Data Streams

摘要

How to process the increasing amount of IoT data in real-time collected from a huge number of devices in the real world into a cloud data center with high frequency is a challenging problem. Furthermore, to share the data among multiple data processing pipelines each of which is to provide an IoT service and to allow service developers agilely develop IoT services through a trial-and-error process, a new IoT platform which enables flexible composition of data streams is required. We developed the data processing platform “Dracena” which enables the flexible composition of real-time IoT services. Every real-world device can synchronize its state to the corresponding virtual object on Dracena through sending the state data to the platform. The behavior of each virtual object, that is data processing on the object, can be agilely defined through Dracena's plug-in mechanism. In this paper, we introduce the architecture of Dracena, show a trial application built on it for mobility services designed to handle a large number of connected vehicles, and experimentally evaluate its performance. The results of the evaluation demonstrate that Dracena has the ability to process frequently generated data in real time sent from hundreds of thousands of devices, while allowing the data produced by real-world people or things to be used across multiple services.