2 Outstanding Edge Computing Technologies That Will Make Your Factory Smart

Edge computing technologies may have become a common offering by various vendors in the industry today. But there are some that seem poised to dominate the factory floor. The distinguishing factor being interoperability and user interaction, I'd like to believe. For instance, in a market survey report by ARC Advisory, the majority of respondents stated that they expect simplified edge infrastructures that can be easily managed. And I agree - connecting your factory processes and assets to an edge infrastructure doesn't have to be difficult and messy. Let's break down two outstanding edge computing technologies that will make you factory smart.

Azure IoT Edge

Simply stated, there are many manufacturing and industrial experts who are skeptical about Microsoft's "involvement" in making factories smart. They believe that, as a software company, it has no solid foundation in the sector. Unfortunately, the most important work in the commercialisation of IIoT is clustered around companies like Microsoft, at least as far as I'm concerned. Hence their influence lingers. No doubt about it, its release of Azure IoT Edge is an exhibition of a deep understanding on what's needed to ensure a robust and capable edge processing to real-time decision making. 

Azure IoT Edge is a cross-platform edge runtime that runs on Windows and Linux environments. It runs even on devices that are resource constrained, around 128MB and a single processor should suffice. What's more, you can use it to remotely deploy, manage and monitor software components on your IoT Edge device. Meanwhile, data from your device goes through an Azure Endpoint, called the IoT hub. This is a component of Azure IoT Suite that is deployed in the cloud and plays a key part in enabling reliable and secure bi-directional communication between the cloud and Azure IoT edge. Let's take a look at the elements that make Azure IoT Edge stand out.

1. IoT Edge Modules

An Azure IoT Edge application consists of modules. These are unique software entities that have a specific task to accomplish. For example, one module could read data from a Modbus device and another one may convert the data into a format that is easy to ingest into a third module, a machine learning one perhaps. 

Best of all, custom modules can be written in the language of your choice. And in a smart factory scenario, these could be protocol converters for industry standard protocols such as utility metering protocols(gas, electric), Profibus, Modbus or some special protocol to collect data and control machines. 

In order to keep track of the health of the software modules, Azure IoT edge runtime contains a module management agent responsible for managing the lifetime of the software components running on the edge device and communicates that back to the IoT Hub. Even better, if a module sends a message to the cloud and there is no connection, the runtime will persist that data until the connection is restored and then send it.

And it doesn't stop there, detection of critical equipment malfunction doesn't have to rely on the cognitive services in the cloud as you are able to take services that normally run on the cloud, package them and push them to the edge device. In fact, Azure IoT Suite components such as Azure Stream Analytics and Azure Machine Learning can easily be pushed to your edge device to work on the data right at its source.

In my experience that is remarkable. Because it means that if you already have logic running in the cloud you are able to move it out to the edge. Effectively cutting out the embedded, security and gateway development skills that would otherwise be required. 

Overall, Microsoft's intention to dominate the factory floor is clearly seen in their massive support of OPC UA. And their pre-built OPC UA publisher and Proxy modules really push the envelope. The publisher module can be deployed onto the Azure IoT Edge runtime to handle on-premise device connectivity. And the proxy module to handle communication and authentication between your cloud-based OPC UA client and devices on the factory floor. Anyone who's ever tried to securely implement command and control from a client int the cloud to a device on the factory floor will tell you how difficult that is. This really sets the tone for a seamlessly connected factory as it allows you to focus on information instead of protocols.

2. Device Management

So here's the thing, Edge Computing is not just about getting intelligence to the edge, once its out there you have to manage it, update it and configure it e.t.c. Azure IoT edge is cloud managed, thereby enabling rich management from Azure IoT Suite. 

At the end of 2016, Microsoft introduced the Device Twin concept, some sort of state synchronisation mechanism between devices and the cloud.The device twin has a set of properties. There are DESIRED PROPERTIES that are owned by the cloud and replicated on the device, they cannot be changed from the device. This is a way of sending updates from the cloud to the device. And then there are REPORTED PROPERTIES which are generated from the device and replicated on the cloud where they cannot be changed. This makes it super easy to communicate back and forth between devices and coordinate changes such as initiating firmware download and so forth. 

But wait there's more, the Device Twin has tags used to group edge devices, for example, into devices with the same OS, or devices at a certain section of the plant, whatever is meaningful to you, so that you can batch some updates and configurations. 

There are many functions, more than we can cover in this article, that make Azure IoT Edge a compelling technology. But for now let's turn our focus to Edge X, another technology that promises to become the de-facto standard as an Industrial IoT Edge framework.

Edge X Foundry

In one of my previous articles, I highlighted that open source and collaborative efforts offer a greater chance of accelerating market adoption for Industrial IoT based systems. Personally, I believe it has become the only way organisations can engage with communities of developers and integrators to create and evolve innovative technologies. The Edge X Foundry is one such project that promises to dominate the factory floor, ahead of Azure IoT I suppose. This is mainly because it is an open source edge platform supported by a rich partner ecosystem. In fact, at the time of its launch in April 2017, it had close to 50 partners.

The intention with EdgeX is to "build a common framework for Industrial IoT edge computing". It has components that can quickly deliver interoperability between machines and devices in a factory. And like Azure IoT Edge, it consists of small software components that perform specific tasks. In EdgeX, these small pieces of a larger application are called Microservices.

EdgeX is platform independent and is capable of being distributed across runtime environments. The idea behind EdgeX is to make it easy to plug in your own edge computing modules such as protocol translators onto a readily built IoT Edge platform. This is made possible by its loosely coupled Microservices architecture.

As it turns out, the fundamental idea common between Edge X Foundry and Azure IoT Edge is that edge applications become easier to build and maintain when they are broken down into smaller pieces which work together and can be reused.

Now let's take a look at elements that gives Edge X an edge (excuse the pun) over other technologies in turning your factory into a connected enterprise.

1. Micro-services

In EdgeX, the microservices are categorised into 4 Service Layers and 2 system layers. The 4 Service Layers include: 

Core Services Layer - includes the provision of micro-service configuration properties, core data management service, meta-data management and facilitation of command and control of actuation requests.

Support Services Layer - supports a wide range of microservices that provide edge analytics and intelligence. And also provides functions for logging, scheduling, alerting and notifications among many others. 

Export Services Layer - is responsible for transporting to the cloud the data that is created and collected at the edge. It also enables Edge X to operate and sustain itself over long periods of time without connection to the cloud. 

Device Services Layer - device services are responsible for interacting with sensors and devices. The micro-services at this layer communicate with the "things" through protocols native to that thing, the data gets converted into a common Edge X Foundry data structure for sending to other micro-services.

On the other hand, the 2 system layers System Infrastructure and System management protect the data facilitate the installation and upgrade of microservices.

The loosely coupled architecture allows for independent microservices to be distributed across discrete hardware devices, written in different languages. And they communicate with 'things' and services via REST API and other standard IoT protocols such as MQTT. 

2. Smart Factory Project

Edge X Foundry recently approved the Smart Factory Project as part of its Vertical Solutions Working Group. The project is concentrated on developing functions that will be delivered in the form of microservices in order to enable a Smart Factory with EdgeX common features.

Not surprisingly, OPC UA is also at the core of data transmission for the EdgeX Smart Factory Project. From sensors and machines to edge devices. The OPC UA microservice is currently being developed. The smart factory project also involves the development of device and microservice management components as a lot of edge devices might be required to process data in a smart factory. The two components are the services deployment manager and the service deployment agent.

The role of the service deployment manager is to deploy, update and monitor the microservices in the edge devices in the factory while the service deployment agent controls the installation and removal of microservices in each edge device. The service deployment manager is capable of administering the work of multiple service deployment agents. 

Publish-Subscribe patterns are always crucial when it comes to moving data between different types of machines in a factory so as to abstract the underlying protocol implementations. In that respect, EdgeX smart factory project uses a messaging scheme based on the principle.

Conclusion

Like much of Industrial IoT and Industry 4.0 technologies, Edge Computing is still in its infancy. Standards are being set and more technologies being developed. But one thing is certain, for effective smart manufacturing implementations, there is no more room for closed systems.  

What other Edge Computing technologies are worth mentioning, I'd like to hear any thoughts at all on this topic. Please share in the comments section below.

Want the latest best and latest practices delivered straight to your inbox? Please sign up here

Disclaimer: I am in no way affiliated with, authorized, sponsored or endorsed by Microsoft or EdgeX Foundry. I claim no credit for the images used this article.