Industrial IoT (IIoT) refers to the amalgamation of disparate technologies and platforms like Robotic machines, Cloud Computing, Analytics and Machine operating personnel into one cohesive platform where they all work in tandem to improve the overall performance and productivity of industrial processes. IIoT starts with digitizing the Industrial processes into workflows which are then automated, transformed and analyzed while decreasing waste and thereby improving efficiency. A typical Industrial facility or a Shop floor comprises thousands of sensors, operating units, various inter-communicating hardware. All these components generate a large amount of data every minute and it can quickly become complex trying to process such large data sets in a timely bound manner to derive meaningful insights and to automate industrial processes by identifying events in the data set.
Building Industrial IoT Platform using AWS IoT Services
AWS IoT platform offers an array of services that can be used to build a tailored IIoT environment for our specific Shop floor in the areas of Device management, Data gathering, Cloud Computing, Security and Analytics. In this blog, let’s look at different services offered by the AWS IoT Platform and how we can build a self-sustaining IIoT environment for a typical manufacturing unit.
- Data Ingestion: AWS offers multiple ways to ingest data seamlessly from the Manufacturing plant to AWS Cloud. MQTT is an industry-standard Messaging protocol used to send sensor data and is platform agnostic. We need to register an IoT Device in AWS IoT Core and setup the communication between the sensor and AWS Cloud by using Certificates.
AWS IMC Kit is physical hardware that supports OPC UA architecture which is another Industry-standard protocol used by Ignition and Kepware software which directly integrates with on the floor SCADA systems and PLCs and share data to AWS Cloud. SCADA and PLC systems are low-level devices that can generate rudimentary data like State LOW or State HIGH for different Switches and sensors. Therefore, we need to use Integration applications like Kepware and Ignition which map the output from such PLC systems into a Digital Asset which can be easily visualized.
- Data Processing: The ingested data can be in the form of Messages (units of data) or streaming data (continuous). Messages can be processed and transformed using AWS Serverless Lambda service. Amazon Kinesis is a powerful, scalable stream service that can process continuous data with sub-millisecond latencies.
Not all sensor data are ready to be processed. In some scenarios, we need to transform the raw sensor data into desired output data like getting current and voltage from a machine and we need to calculate Power and Energy on-the-fly using formulas. There are use cases where we would need to aggregate the data over a period of time to calculate the average & sum of raw sensor data. AWS Offers many services for such on-the-fly operations which can scale from hundreds to thousands of messages. We can also utilize AWS Simple Queue Service to queue messages or ingest all messages into AWS Elastic Map Reduce (EMR) for Big Data processing.
- Data Storage: After the data is processed, filtered, transformed and analyzed, it needs to be stored for long term usage and to generate offline/real-time dashboards. Amazon offers RDS for relational data storage in the form of MySQL, PostgreSQL databases and Dynamo DB for document and NoSQL data storage. We have to design the data storage depending on the data being ingested, the volume of data over time and whether the data will be queried frequently or stored for long term usage.
IoT Sitewise is a newly introduced service where we can model digital representations of the physical asset and store time-series sensor data against the digital asset model instead of a relational or non-relational data storage. Sitewise does not require any maintenance like storage space allocation, scalability. Once the digital assets are created AWS handles the maintenance and storage data scalability. Sitewise also offers easy to create Dashboards for personnel operating on the Shop floor to get quick diagnostics of the operations.
- Client Application & API’s: Analytical capabilities form the crux of Industry 4.0 and IIoT revolution. Therefore, the stored data need to be queried or fetched for various Business requirements. Amazon offers Beanstalk (PaaS), EC2 (Server) and API gateway (Serverless Microservices) to build our Client-side application which can query the Database based on business use cases. We can use Beanstalk to deploy our Java / .Net Application or build an array of Microservices which are serverless by integrating Lambda & API Gateway which can then be consumed by Third Party Dashboard applications.
Amazon QuickSight is a cloud-native scalable, serverless, Business Intelligence (BI) tool with inbuilt Machine Learning (ML) capabilities. It seamlessly integrates with data storage services of AWS like Dynamo DB, RDS and Sitewise. We can quickly create highly interactive dashboards by dragging and dropping different Chart components like Pie Chart, Bar Chart, Grid Table and selecting a relevant data source to pull the source data from. We can also perform Queries using Natural Language queries which are then converted to low-level technical queries by the QuickSight ML engine and data is fetched and presented to us. The complexity of building / coding a client-side application can be avoided by using QuickSight as an alternative ready to build visual Dashboard Modeler.