Revolutionize Industrial Data Integration: The UNS Approach
I4.0 Insight
3 minutes
Feb 7, 2024
The shift from traditional Purdue models to tierless/layerless models is a well-worn topic in the industrial digitalization field. Virtually every vendor in this space uses a diagram similar to the one below:
We dream of a day when any level of systems, equipment, and personnel within an enterprise can converse with a Single Source of Truth, achieving unparalleled data fluidity and interoperability.
Yet, with the conventional IT tools like RESTful API, cluttered OPC stacks, cumbersome ETL processes, and SQL-based solutions prevalent in today's commercial software, achieving Single Source of Truth remains elusive.
（from OPC Foundation Official）
Around 2015, a North American IIoT enthusiast named Walker Raynolds began using a Message Broker as the core for building a three-tier architecture. He propagated his practices within the community, and this concise method surprisingly proved highly efficient. It quickly gained popularity in the community and was adopted by some cutting-edge enterprises.
The core of building a UNS
In practice, the core of UNS is MQTT Broker. Simply put, UNS enables bidirectional subscription of data sources and consumers to a centralized MQTT Broker.
How to create a UNS？
1. Define data modeling frameworks and MQTT topics
ISA95 Part 2 and SparkPlugB are popular schemas for defining the MQTT topic format. The former is used for enterprise-level data integration, while the latter is used for device-level data integration. Methodology and research are crucial in this process.
2. Data processing/Contextualization
For instance, when connecting a thermocouple via a gateway box, it sends a payload defined by the box itself, which includes an integer value of 00355. This value, 00355, corresponds to 35.5 degrees Celsius. To handle this, you have to use a tool to convert 00355 to 35.5, reformat the JSON payload accordingly, and then publish it to the MQTT Broker. This process is always the job of IIoT platforms or software like NodeRED.
3. Establish connections- Subscribe & Publish
Define the topic Site1/Unit1/ThermoCouple1 in the MQTT Broker and publish the thermocouple data from the previous step to this topic using NodeRED. Any system that wants real-time values of this thermocouple only needs to subscribe to this topic.
Completing these three steps establishes the simplest UNS. This method enables real-time data exchange across software layers.
Features of UNS
Simple and lightweight
This is the most crucial feature of UNS.
UNS can be comprehensively explained in a thousand-word article. In contrast, elaborating on a companion specification in OPC UA would necessitate ten thousand words, and for an IIoT platform, you may still be reading the preface at that time.
Deploying and connecting an MQTT Broker is much simpler than using Kepware, which is why Mosquito is widely used in home automation.
Event-driven, Edge-driven
UNS based on an MQTT Broker is a typical event-driven architecture. By leveraging the concept of triggering actions upon event occurrence, we can make it easy to define and develop industrial applications.
MQTT payloads are accepted by all, allowing publishers to customize and modify their messages and logic at any time. This decentralizes initiative and data processing to the devices and nodes accessing UNS. Combined with edge-level data tools like NodeRED, enterprises can scale UNS significantly without compromising security, performance, or fault tolerance.
Report by Exception
Many data in control systems and software systems remain unchanged for long periods. Using traditional methods of periodic polling would lead to bandwidth wastage. In large corporate enterprises, such waste can escalate and prevent the establishment of a single unified data source. However, UNS enables a mechanism of "Report by Exception", making our lives much easier.
Highly scalable
EMQX and HiveMQ, two commercial brokers, have refined their products to cloud-native standards. When we connect to an MQTT address, it could be backed by a highly available cluster. This aligns with the trend towards layered, high availability, and scalability.
Real-time snapshot
As mentioned earlier, UNS only includes real-time data. Industrial data integration naturally involves OLTP and OLAP for utilizing historical data.
Next steps
In the next article, We will discuss how open-source time-series databases and NodeRED are revolutionarily addressing this issue, along with Balena's "MING Stack".
Nowadays, large institutions and suppliers, such as Gartner, SAP, ARC, Wipro, etc., are beginning to discuss UNS (Unified Namespace). However, startups are truly implementing or productizing this method.
With its agility and simplicity, UNS will eventually triumph over the cumbersome and aging OPC. Small teams can surpass large companies that only care about weaving concepts by relying on passion and perseverance.
This marks a new trend in the digital age.
