Skip to content

Latest commit

 

History

History
151 lines (129 loc) · 8.58 KB

system-decomposition.md

File metadata and controls

151 lines (129 loc) · 8.58 KB
Raw

System decomposition

Breaking it down into sub-systems.

Functional decomposition

Here's the breakdown of the Kitt4Sme platform into sub-systems and layers.

TODO: the text below is a straight C&P from the proposal and needs some massaging!

With the aim of building an open source platform for uptake of AI solutions in manufacturing SMEs, KITT4SME selected the currently best-in-class open source platform for IoT: FIWARE. Figure 6 illustrates the different platform layers (components in the mid-blue boxes are provided by FIWARE). FIWARE is a scalable open source component framework that can be composed to integrate, handle and manage context information from different data sources, distribute that data and stream it into involved external components for persistence as well as for AI-based processing aims. Computed results can then be integrated back enabling actuation and the enrichment of the current context.

Starting from the lowest layer, the KITT4SME architecture covers data gathering from different devices, deployed in the factory, feeding the knowledge-base of the system with raw and pre-processed data:

  • Wearable Sensors, worn by the worker, are meant to help monitor psychophysical parameters related to health and well-being. According to the “human in the loop” principle, those sensors can be also involved within the decision-making loop.
  • Cyber-Physical Systems, especially those typically involved within the process (machining equipment, 3D printers, collaborative robots, automated logistics, etc.).
  • Environmental Sensors, scattered in the factory and aimed to collect information on the environmental conditions (i.e. air pollution, temperature, noise level, etc.).

At the next layer, the FIWARE framework provides a set of Generic Enablers interfacing devices for the purpose of gathering context information or triggering actuations in response to context updates:

  • The IDAS Generic Enabler which offers a wide range of IoT Agents meant to interface devices using the most widely used IoT protocols (LWM2M over CoaP, JSON and UltraLight2.0 over HTTP/MQTT/AMQP, OneM2M or OPC-UA).
  • The FAST RTPS Generic Enabler that adopts ROS2 middleware, which helps to interface with robotics systems.
  • The Generic Enabler Kurento that enables real-time processing of media streams to support the transformation of video cameras into sensors as well as the incorporation of advanced application functions (integrated audiovisual communications, augmented reality, flexible media playing and recording, etc.)

The next upper layer shows the FIWARE Context Broker, the core and mandatory component for a Powered by FIWARE solution. It enables to manage context information in a decentralized and large-scale way. It provides a RESTful API enabling to perform updates, queries or subscribe to changes on context information. The Context Broker holds information about current context data, i.e. the current and most recent update status of all the devices, components and processes that are engaged in a processing industry factory. However, context information evolves over time, creating a context history that is fundamental for analytical and machine learning tasks that base their algorithms on dedicated time windows of factory context data. FIWARE provides own Generic Enabler to persist Time Series historical data, QuantumLeap and STH-Comet, but can also forward the data to other third-party databases. Furthermore, data can be streamed to distributed processing engines like the Big Data and AI-based Detections Tools to enable analysis on historic data context.

Layers

The next upper layer shows those Big Data, AI-based detection and optimization tools components. They process context history data in order to detect patterns and correlations, anomalies, malfunctions, unexpected behaviours, outliers by means of Big Data analytics, stream processing and to provide value-added functionalities like process optimization or quality improvements. The results of their processing are injected back into the FIWARE Context Broker to activate the decision-making processes and actuations. The detection and optimization tools can be continuously created and edited by using integrated functional models editors such (like JupiterLab) and added as new FIWARE components made available to the open source community by means of a Functional Model Catalog.

At the same layer, the KITT4SME architecture foresees four further components:

  • Human description Database, deployed by means of Cygnus, in charge of persisting data related to the physiological parameters, information about workers, machine parameters and environmental data that contribute to create a complete representation of the worker in activity at the shop floor.
  • The BI platform and suite Knowage: a FIWARE Generic Enabler to perform traditional business analytics over classical data sources, databases and big data systems. It enables manual ad-hoc OLAP analytics, dashboarding as well as publishing reports about historical data and KPIs to extend the automatic anomalies detection analytics capabilities.
  • A Mashup Platform as KPI-driven process governance dashboard: it helps the different operators within the factory to have immediate and easy access at all the relevant context information that represent the latest status of the processes. The dashboard is contextual: i.e. the interface is automatically customized to the operator and to his location within the factory. Localization is based on different technologies depending on the concrete scenarios developed in the use cases. The RAMP platform, here integrated and described afterwards, provides a set of factory dashboards, including one integrated with this Mashup Platform provided by FIWARE.
  • External IDS Connector: this connector represents a component of the IDSA reference architecture. It works as a trustful interface between internal data sources and external data consumers to enable secure and interoperable data exchange. This connector makes the KITT4SME architecture ready to be integrated with external value- added services where the data exchange is regulated by the IDS policies.

As anticipated before, the results of the Big Data and (X)AI-based detections and optimization solutions activate the decision-making mechanisms whose logics can be modeled and then managed during execution by MPMS. Tasks belonging to the activation process can foresee, on the one hand, a user interaction (for example for inputting some information or simply for supervising some decisions by means of a wearable devices) and, on the other hand, a behavioral update for a group of involved CPS. Basically, the decision maker, supported by the Intervention Manager, is meant to react in order to both optimize process performances and workers’ well-being. The interventions are launched at different levels (single workstation, work cell, production system), and orchestrated so as to exploit the flexibility of the production system to support operators whenever their behavior deviates from optimal and/or safe performance.

The triggered actuations need to be identified by IDAS IoT Agents (again through the FIWARE Context Broker) in order to finalize the retroactions towards CPS on the basis on the decisions made.

Within the KITT4SME project, RAMP (Robotics and Automation MarketPlace) enables the Software-as-a-Service provision of the above-mentioned components. By enabling FIWARE-compatible equipment (robots, machines, sensors, mobile devices etc.) in the production floor, it will be possible to directly use the different tools offered by KITT4SME, without the need of timely software deployments and IT expertise, allowing SMEs to focus on the value-adding activities of their core business. On top of that, the distributed capabilities of the architecture allow the tools and production data to be used collaboratively between manufacturing SMEs and technology providers, providing a better overview of the production status and facilitating the online co-creation, which in turn minimises the need of continuous on-site inspections and system installations. The KITT4SME components include different Dashboards (Process monitoring, FIWARE Mashup, Data visualisation), as well as the Process Design sub-module of the MPMS.

The KITT4SME Platform is supported by an IDM (IDentity Management & Access Control) Generic Enabler that brings support to secure and private OAuth2-based authentication of users and devices, user profile management, privacy-preserving disposition of personal data, Single Sign-On (SSO) and Identity Federation across multiple administration domains.