Software Open Access

Automated Fusion System Design and Adaptation Implementation

Fritze, Alexander; Holst, Christoph-Alexander

This published prototype is a JAVA-based implementation of the automated fusion system design proposed in [FMH+17]. The implementation orchestrates a distributed information fusion system, i.e., it identifies features and attribute supported by the system. The automated orchestration is carried out at a central device called system manager. Basic elements of the fusion system are intelligent sensors. Intelligent sensors monitor a system using elementary sensors (e.g., temperature sensors or acoustic sensors) [MDL+16, FML16]. The sensor signals of all intelligent sensors are gathered and fused to evaluate the condition (i.e., health) of the monitored system. An intelligent sensor is additionally equipped with processor units, memory, and communication interfaces. It is self-adaptable and self-aware. An intelligent sensor hosts a semantic self-description stating available elementary sensors and algorithms. Algorithms are used to extract certain features from sensor signals. This implementation uses the Raspberry Pi 3B as platform for intelligent sensors. The Raspberry Pis 3B supports several interfaces to read multiple elementary sensor signals. This implementation reads sensor signals via the Serial Peripheral Interface (SPI). Communication between intelligent sensors uses the Raspberry Pi’s Ethernet interface. All communication for the organisation and configuration of the fusion system uses TCP/IP. Process data (sensor signals and features) are communicated via an Industrial Ethernet in real-time. The process data communication is not part of this publication. The automated fusion system design is structured into the following four phases:

  1. Discovery: The system manager searches for available intelligent sensors. The discovery phase is carried out continuously independent of the other three phases. If a new intelligent sensor is discovered, the knowledge building phase is triggered.
  2. Knowledge Building: Semantic information (self-description of intelligent sensors) is transferred to a knowledge base at the system manager.
  3. Orchestration: The system manager carries out the fusion system configuration automatically.
  4. Operation: All intelligent sensors periodically send their sensor signals and features to the system manager using a real-time Ethernet protocol.

Discovery of intelligent sensors and transfer of semantic information is implemented using the Open Platform Communication Unified Architecture (OPC UA). OPC UA offers a Local Discovery Server (LDS), which exposes available OPC UA servers in a local network. As soon as the system manager has discovered an intelligent sensor, the semantic self-description is collected and stored in the system manager’s knowledge base. Then, the fusion system is orchestrated using a rule-based system. The orchestration engine identifies based on available sensors and algorithms features and different kinds of attributes (physical, module, functional,quality). For details about the orchestration process and the rule-based system the reader is referred to the corresponding journal article [FMH+17]. The last step in the orchestration phase is the creation of an configuration file for the real-time communication. This configuration file is used to determine the layout of the real-time Ethernet communication network.

The source code is included in the ZIP file of this upload. The accompanying PDF contains a description on how to compile and execute the implementation.

[FMH+17] FRITZE, Alexander ; MÖNKS, Uwe ; HOLST, Christoph-Alexander ; LOHWEG, Volker: An Approach to Automated Fusion System Design and Adaptation. In: Sensors 17 (2017), Nr. 3, 601. – DOI 10.3390/s17030601

[FML16] FRITZE, Alexander ; MÖNKS, Uwe ; LOHWEG, Volker: A Support System for Sensor and Information Fusion System Design. In: 3rd International Conference on System-Integrated Intelligence - New Challenges for Product and Production Engineering, Paderborn, Germany, 2016

[MDL+16] MÖNKS, Uwe ; DÖRKSEN, Helene ; LOHWEG, Volker ; HÜBNER, Michael: Information Fusion of Conflicting Input Data. In: Sensors (Basel, Switzerland) 16 (2016), Nr. 11. – DOI 10.3390/s16111798. – ISSN 1424–8220

This work was partly funded by the German Federal Ministry of Education and Research (BMBF) within the Leading-Edge Cluster ''Intelligent Technical Systems OstWestfalenLippe'' (it's OWL) (Grant No. 02PQ1020).
Files (43.6 MB)
Name Size
1.3 MB Download
42.3 MB Download


Cite as