Connecting Operational Technology to AWS Using the EXOR eXware707T Field Gateway
23 May 2019
Advances in the Industrial Internet of Things (IIoT) have made smart factories a reality through the application of artificial intelligence (AI) and cloud computing technologies that improve machine operation and manufacturing processes. To do this, purpose-built hardware at the edge needs to interface with operational technology (OT), which includes industrial machines and assets that often speak their own proprietary fieldbus protocols.
The edge hardware translates the fieldbus protocols, processes that data, and connects to Amazon Web Services (AWS). In this post, we will explore how EXOR Internationalâs systems-on-module (SOM) and edge gateways, powered by Intelâs Cyclone V FPGA, allow system integrators and application builders to deliver AWS-based IIoT solutions with faster time-to-market, lower total cost of ownership (TCO), and reduced development efforts. Intel is an AWS Partner Network (APN) Advanced Technology Partner with the AWS IoT Competency.
Industrial Internet of Things
Cloud computing, machine intelligence, and internet ubiquity have ushered in the IIoT era, manifested through the rise of smart factories and smart manufacturing. There are IIoT applications for different use cases. Some applications serve to boost efficiency and reduce operational expenses, while other applications target optimizing production processes and improving product quality. Implementation of these applications requires a scalable and robust cloud infrastructure for data aggregation, storage, analytics, and visualization. Devices are required at the edge to support data collection, machine control, and cloud connection. Some use cases involve execution of data analytics and AI models directly on edge devices. In Figure 1 below, you can see the use of edge devices to acquire and process data from motors, programmable logic controller (PLC), and other operational technologies. The illustration shows a typical split between OT and IT, which includes on-premises network infrastructure and AWS. đˇ Figure 1 â Using edge devices to interface with operational technology.
AWS for Industrial IoT
For real-world IIoT implementations, cloud infrastructure must meet the following criteria:
- Functionality: OT creates a large amount of unstructured data. IIoT analytics require aggregation, transformation, storage, and analysis of this data. AWS provides building blocks for these functions, and enables ease of creating data pipelines out of the building blocks.
- Scalability: A typical IIoT deployment may have many endpoints across multiple locations, with each endpoint continuously generating data. As such, AWS is able to handle large volumes of data at-rest or in-transit without compromising performance and responsiveness.
- Reliability: When industrial processes depend on functions running on the cloud, the infrastructure must be reliable and highly available (HA) because disruptions can result in financial losses.
- Security: Machine data often contains private and sensitive information. This data must be protected when transiting from edge to cloud and safeguarded when stored on AWS. Furthermore, the cloud itself must be fortified with secure access control to prevent intrusion on the factoryâs operation.
AWS infrastructure is robust, secure, and highly elastic. The AWS portfolio of services provide the building blocks for data pipelines to support many different IIoT use cases. Many AWS services are suitable to process and present machine data:Â AWS Lambda for data transformation, Amazon Simple Storage Service (Amazon S3) for storage, Amazon Aurora for database support, Amazon EMR for big data analytics, Amazon QuickSight for data visualization, and Amazon API Gateway for secure sharing of data. AWS also offers several IoT-specific services that enable and simplify end-to-end implementation.
- AWS IoT Core: This is a message broker that can handle thousands of connections to devices on a manufacturing floor across multiple sites. AWS IoT Core follows a publish and subscribe model, allowing for edge devices and services running on AWS to consume data, act on that data, and communicate to other subscribed endpoints.
- AWS IoT Greengrass: This software is for use on edge devices and enables deployment, execution, and management of local compute workloads. In the context of IIoT, itâs used to support workloads performing signal processing, protocol conversion, local analytics, and AI inference.
- AWS IoT Analytics: This is a managed data pipeline service for the collection, transformation, and analysis of machine data in bulk. Itâs suitable for use cases that require mining of historical data to derive insights, such as AI-powered predicative maintenance.
- Amazon Kinesis Data Analytics: This supports querying, processing, and analysis of streaming data in-flight. Amazon Kinesis is suitable for consuming large amounts of data and processing that data in real-time, enabling use cases such as real-time monitoring and fault detection.
- Amazon SageMaker: This is a machine learning (ML) service that simplifies and accelerates the workflow for building, training, optimizing, and packaging AI models. It can be used to develop IIoT applications that rely on edge inference or machine vision, and enables use cases such as defect detection and safety surveillance. Trained models from Amazon SageMaker can be deployed onto edge devices with AWS IoT Greengrass.
Industrial Edge Devices
Broadly speaking, there are two types of IIoT edge devices: Field Gateway (FGW) and Industrial PC (IPC). A field gateway acquires data from operational technology endpoints and routes this data to AWS. The FGW must accept command and control messages from AWS and route these messages to the endpoint. The OT endpoints can be sensors, programmable logic controllers (PLCs), and other controllers of machinery.
Communication with these endpoints often takes place over a fieldbus. The FGW has to support the specific protocol (e.g. handshake scheme, message format, data rate) the fieldbus operates with.
There are a large number of fieldbus types that can be used on factory floors, each with its own protocol. Examples are Ethercat, Profinet, Ethernet IP, Modbus TCP, Profibus DP, CAN OPEN, Modbus RTU, to name a few. An Industrial PC performs everything that a field gateway does, plus more. An IPC can also support advanced workloads at the edge, such as data analytics and AI inference based on deep learning algorithms.
Supporting time-sensitive networking (TSN) is a common requirement for both types of edge devices. TSN enables deterministic latency in local area network communication. With TSN, the edge device can cooperate with other devices on the network to perform real-time tasks. For example, an IPC can be used to synchronize multiple robot arms on a TSN, ensuring timely delivery of control messages from the IPC to the robot arms.
For use in industrial environments, edge devices must be able to operate 24/7 over a wide range of temperatures. FGWs, in particular, are often required to fit into tight spaces around machinery, and typical requirements include a compact form factor.
Building Industrial Edge Devices with Systems-on-Modules
Edge devices can be built with system-on-modules (SOMs) that provide all of the required compute power and interfaces on a single module. SOMs come in an ultra-compact single-board computer form factor that meets the requirements of most IIoT devices. Original equipment manufacturers (OEMs) can build FGWs and IPCs using SOMs, and only need to focus on integrating additional peripherals, such as wireless connectivity or physical connectors for fieldbus protocols. EXOR International builds industrial-grade SOM products targeting IIoT use cases. The EXOR uS05 microSOM fulfills all system requirements for AWS IoT Greengrass. In the table below, you can see the hardware specification for the uS05 SOM.
The uS05 microSOM is a low-power platform based on the Intel Cyclone V FPGA System on Chip, which is made up of a FPGA fabric and an Arm-based host processor. It supports several types of interfaces along with two Ethernet ports and a TSN switch. Capable of high data rate operation, the FPGA fabric can be configured to support fieldbus connections via one or more interfaces. The host processor is used for fieldbus protocol processing using EXORâs JMobile software that has support for 200+ fieldbus protocols. The uS05 is ideal for use in field gateways to connect industrial endpoints to the cloud. The eXware 707T gateway, powered by the uS05 microSOM, is AWS IoT Greengrass-qualified and listed in the Partner Device Catalog. The gateway integrates all of the benefits of the uS05 microSOM with Ethernet ports, expansion ports, and storage in an industrial housing that can be deployed as a standalone field gateway in an industrial environment. AWS IoT Greengrass-qualified hardware has been tested and proven to integrate all of the necessary software and hardware requirements to use AWS IoT Greengrass and connect to AWS.
Asset Condition Monitoring
In this section, weâll examine how the eXware 707T gateway with AWS IoT Greengrass can collect data to capture the state of an industrial asset. Based on the health state of the asset, a maintenance engineer can monitor the asset or be alerted of required maintenance. Motor failure is a common cause of downtime on production lines. The vibration characteristics of the motor during operation can tell us the health of the motor. Untimely failures may be prevented via proactive servicing guided by continuous monitoring and assessment of motor vibrations. An end-to-end implementation of a motor condition monitoring based on this concept is illustrated in Figure 3Â below. It makes use of the eXware 707T gateway to connect to vibration sensors mounted on the motors.
On AWS, frequency data arriving at AWS IoT is routed to an AWS IoT Analytics pipeline with an IoT Rule. AWS IoT Analytics cleans, transforms, and stores the data. Time-series analysis is periodically invoked on the dataset to detect anomalies and results are displayed to a maintenance technician in Amazon QuickSight. The data is also sent to AWS IoT Events, where a complex event processing engine determines if the motor is operating normally. In the event that maintenance on the motor is needed, an AWS Lambda function is triggered to alert the maintenance technician with an Amazon Simple Notification Service (SNS) notification. In the next illustration, you will see the software and hardware architecture of the eXware 707T gateway itself. Figure 4Â demonstrates how data is collected, processed, and transformed by the eXware 707T gateway with AWS IoT Greengrass and JMobile software installed.
Communication between the gateway and vibration sensor happens over CANbus, one of the protocols supported by EXORâs JMobile software. The gateway executes a Lambda function in AWS IoT Greengrass to continuously process vibration data received from the sensor by interfacing with the JMobile software APIs. It computes the frequency spectrum of the vibrations using Fast Fourier transform, and sends this frequency data to AWS IoT.
Industrial IoT (IIoT) solutions on AWS are harnessing edge intelligence along with cloud computing to derive value from operational technology. AWS IoT services such as AWS IoT Analytics allow you to create data pipelines using data collected from edge devices to gain insights such as machine health. AWS depends on APN Partners such as Intel and EXOR International to create edge solutions that can collect this data, perform edge processing, and move this data to AWS IoT Core. The eXware707T gateway, powered by the us05 microSOM, can connect to over 200 different fieldbus protocols to interface with OT and industrial machinery by using the EXOR JMobile software.
Together with AWS IoT Greengrass, the gateway provides a complete intelligent field gateway solution with connection to AWS IoT Core.
For further ideas on how AWS services can be used for IIoT, see the case studies on the AWS Industrial IoT website.