IoT and the Time-Critical Edge

July 07, 2020


Challenges faced by IoT deployments related to latency, network bandwidth, reliability, and security cannot be addressed by cloud-only models, so the focus of IoT is moving towards the edge.

Challenges faced by IoT deployments related to latency, network bandwidth, reliability, and security cannot be addressed by cloud-only models, so the focus of IoT is moving towards the edge.

The forecast is that in just two years’ time, 45% of all data created by industrial devices will be stored, processed, analyzed, and acted upon close to or at the edge of the network.

High performance, secure, open, and interoperable edge software platforms are becoming vital to support this new computing paradigm. The current generation of edge IoT platforms have been designed to run on general-purpose hardware, such as IoT gateways and on-premise servers. These are environments that support data processing rates measured in hundreds of milliseconds to seconds and typically have at least 1GB of available memory. For many use cases ranging from building management to equipment monitoring and predictive maintenance, response times in excess of 100ms are usually quite acceptable.

However, there are a significant number of important industrial IoT use cases, such as closed-loop process control, real-time signal processing or high-frequency analytics, for example, where cycle times are measured in microseconds and predictable real-time execution is required. This specialized subset of IoT Edge Computing is referred to as the time-critical edge.

IoT Momentum is Driving Change

The time-critical edge is not new, but like other forms of edge computing, it most certainly is evolving as IoT adoption gains momentum.

Industrial control systems using dedicated PLCs have been around for decades. PLC and PACs offer limited software programmability and configurability so other custom or bespoke software-oriented solutions have also been used.

With the emergence of Industrial IoT these traditional approaches do not adequately support the convergence between the operation technology (OT) and information technology (IT) worlds.  This is limiting the ability to leverage the latest advances in artificial intelligence (AI) and machine learning (ML).

The process automation world in particular has recognized the need for their edge systems to be much more software friendly, hence, the creation and growing adoption of standards such as the IEC6113 and, more lately, IEC 61499 for the development of the next generation of distributed control systems (DCS).  In turn, this is enabling both the vendors and users to leverage COTS hardware hosting either a specialized real-time operating systems (RTOS) or Linux with real-time kernel extensions.

The trend to move from dedicated point solutions to support time-critical edge use cases to software defined multi-mode, configurable IPC-based solutions will most certainly accelerate. Other technologies that are already becoming common in more general-purpose IoT edge systems such as containerization and virtualization will also be used more widely with time-critical systems. They offer capabilities that can help address application deployment and orchestration, concerns that become incredibly important as a system scales and evolves.  

Open Edge Software Platforms

A new generation of open software platforms are starting to emerge and will play an increasingly important role in enabling IoT edge systems, including those with time-critical requirements.

They provide the central integration point between OT and IT environments, enabling OT connectivity and dealing with data acquisition from many different southbound protocols. They can also abstract applications from the underlying hardware and operating system details, enabling portability and software reuse. Importantly, they enable integration with higher-level systems and IT endpoints, allowing data to be filtered and transformed before being sent northbound. Fundamentally, edge software platforms enable low-latency local decision-making and analytics independent of whether cloud connectivity is available.   

To support time-critical use cases, it is important these edge software platforms provide a set of core functions such as:

•       Data aggregation from southbound OT equipment

•       Support for edge analytics/rules/AI

•       The ability to issue commands/actuation requests to the connected equipment

•       The ability to filter and transform data

•       Export to northbound IT endpoints

•       Store and forward for cases when northbound connectivity is intermittent 

•       Logging and alerts for manageability

•       Secure data communications and access to platform resources

•       Support for modern software approaches to deployment, such as containerization and virtualization   

These represent the basic functions on which to build an edge system. However, a time-critical edge platform must provide additional capabilities that enable:

•       High-frequency data acquisition and ultra-low latency response times (< 50ms to microseconds)

•       Real-time deterministic data processing

•       The ability to be deployed in environments with limited CPU power and memory or legacy environments

•       Compatible with emerging industry standards such as IEC6113 and 61499

Edge platform technology that embraces openness is key to addressing the heterogeneity inherent in industrial edge systems. It’s crucial that the choice of edge platform solution provides users with choice and flexibility at multiple levels, independent of silicon, hardware, operating system OT technology and cloud connectivity. Well-defined APIs enable interoperability and application “plug and play.”

Industry has recognized this need. Open-source edge computing initiatives such as the Linux Foundation’s EdgeX Foundry™ project are starting to gain significant momentum supported by a growing ecosystem of companies across industrial sectors. Collectively, the member organizations of the project are focused on creating a common software foundation for low-latency industrial edge systems. Through community extensions, such as IOTech’s Edge XRT, systems with the most demanding time-critical requirements can now benefit from a common open framework on which to support a broad range of industrial use cases. This includes support for low-latency “hard” real-time processing requirements and deployment on specialized (e.g., IPC running an RTOS or a Linux kernel with real-time extensions) or resource-challenged edge environments such as a microcontroller.  

In conclusion, a new generation of open edge software platforms is poised to have a significant impact on the deployment of high-performance edge systems. They will enable solutions that are far more software defined, multi-mode, evolvable and flexible than has previously been possible. System developers will be shielded from many of the underlying data connectivity, networking, application integration and hardware concerns. This has the effect of simplifying development, reducing integration costs and time-to-market for new edge solutions, even those with the most demanding time-critical requirements.

About the Author

Andrew Foster is a Product Director at IOTech and has over 25 years of experience developing IoT and other Distributed Real-time and Embedded (DRE) software products and services. He has held senior management positions in Product Delivery, Product Management and Product Marketing. Prior to joining IOTech he was the Product Marketing Manager at PrismTech and lead a team responsible for helping to grow PrismTech’s global market presence. Andrew holds a M.S. in Computer-Based Plant and Process Control and a B. Eng in Digital Systems Engineering.