“Simple is Hard”, so the saying goes. Reducing the essence of a complex idea or system into something that is readily understood and easy to use takes time, effort, and creativity. Never has this been truer than today’s push toward Industrial Internet of Things (IIoT). While much progress has been made with cloud analytics and backend support, the sensor domain has remained largely impervious to change. With multitudes of different sensors, interfaces, and applications, standardization has been a necessary but missing ingredient. Simplifying this problem is hard. In 2019, the PICMG® standards organization, in collaboration with the DMTF®, launched two new industry specifications targeted at bringing plug-and-play to the sensor domain of IIoT. The first of these specifications focuses on a small hardware module that today’s sensor vendors can use to create smart sensor nodes. The second specification defines a network architecture and data model that ensures uniformity at the software level. Our aim is to enable and accelerate industrial smart sensor market by making their creation and deployment simple.
To date, PICMG made great strides toward the standardization of the sensor domain for Industrial IoT. The hardware technical subcommittee has defined the physical signals, communication interface, environmental conditions and started tackling the configuration and connector needs. The Network Architecture and Data Model technical subcommittee, on the other hand, has focused more on the overall network architecture, to enable sensor plug-and-play. We have created a framework for the architecture and behavior of the various elements and selected communications protocols.
Industrial Internet of Things (IIoT) is the fastest growing segment of Internet of Things with annual revenues expected to be larger than the commercial segment of Internet of Things through the first half of next decade. Within this segment are many traditional embedded markets including: defense and aerospace, transportation, energy and industrial automation. Among these markets, Smart Factory (industrial automation) is the fastest growing vertical (24% CAGR) as factory automation seeks to deploy intelligent factory equipment, tightly couple factory operations and backend processes, leverage IT skillsets, and improve overall efficiencies through analytics.
At the 2019 Embedded Technology Conference, I had the
opportunity to present to a room of long-time industry professionals about the
state of Industrial Internet of Things.
In part of the discussion I told the fictional story of how Mary, a director
of operations at a smart factory, was faced with the monumental task of bringing
up a new factory line while upgrading the factory equipment; however, she
needed to leave the existing software infrastructure intact. To do this, Mary’s new hardware would need to
plug-and-play with existing infrastructure and interoperate with disparate
equipment at levels so far unachievable in industry. Mary moved forward with her plans, confident
in positive results. Her upgrade was a
huge success. Mary got a large promotion, and everyone lived happily ever
after. The question I posed then is just as relevant today: “Is this a fairy
tale, or could this really be possible?”
At PICMG, we believe the answer is: “Not only is this possible, it is achievable today.” With collaboration between our member companies and other industry consortia, we are targeting open specifications to address the need for interoperable industrial computing solutions at the sensor plane. This paper outlines the first two of these specifications which were launched earlier this year and are expected to be released in the first half of 2020.
Sensor Data Model and Network Architecture
If sensors in an IIoT deployment are to plug-and-play with the rest of the automation infrastructure, there must be agreement upon how the devices communicate with the network, how they report their features, and how they can be interacted with from higher layers of the network. With IIoT, the first of these challenges, how to communicate, is largely taken care of by internet technologies: Ethernet, HTTP, JSON and the like. The second problem – how the sensors report themselves, and how they interact with the rest of the network requires the standardization of data models and network architectures.
IIoT specification underway by PICMG addresses just this need. Key elements of the architecture under
discussion today are a low-level binary data model definition that enables
lightweight sensor nodes, a gateway architecture (for converting binary coded
data models to DMTF Redfish), methods of synchronization of multiple endpoints,
and security recommendations.
specification will enable an ecosystem of new smart sensor vendors to create
sensors that interoperate seamlessly within the Redfish/PICMG sensor-domain
network architecture. Existing vendors of Com Express and CompactPCI
Serial will also benefit as potential gateway suppliers into the network.
This proposal is expected to have two primary outputs. A DMTF Redfish-compatible data model, and a specification that documents the system architecture of the sensor-domain network.
The second specification currently
underway proposes a new microcontroller-agnostic ultra-small form-factor module
for the enablement of smart sensors. This module, which is expected to be
no more than 30mm squared, will provide a hardware platform for traditional
sensor vendors wishing to quickly create smart sensors. When combined
with the PICMG sensor domain network architecture and data model, sensors will
seamlessly integrate into the network with plug-and-play interoperability.
We envision that this specification
will benefit the industry in three specific ways. First, it will enable sensor vendors to
create smart sensors without having to manufacture the control circuitry and/or
software by purchasing these components from PICMG-compliant suppliers. Second, it will enable controller suppliers
who wish to create smart sensors or smart-sensor components to do so in a way
that is interoperable with other suppliers. And lastly, it will accelerate the
uptake of smart-sensor technology through open-specifications and
Joining the efforts
At PICMG, we are excited about these two new contributions aimed at accelerating the adoption of standards-based IIoT. Together we are working on moving “plug-and-play” at the sensor domain from fantasy to reality.
At PICMG, we have kicked off a new focus on the requirements for Industrial IoT (IIoT). From there, our efforts can expand out to other IoT market requirements. In IIoT, hardware and software interoperability tends to be more important than household/consumer applications as sensors, actuators, and controllers from multiple vendors must work together seamlessly. But, standardization has not yet materialized.
IIoT, is different than traditional industrial automation in the fact that it combines ubiquitous sensing, advanced analytics, and IT technology. Going beyond traditional automation control functions, IIoT includes sensors and actuators for facility operations, machine health, ambient conditions, quality, and a variety of other functions. Advanced analytics enables the IIoT system to realize higher levels of operational efficiency by extracting meaning from the potential data available from a vast array of deployed sensors. Similar to cloud data centers, where sensors data is used to optimize virtually every aspect of operational efficiency, smart factories and other IIoT applications utilize analytics to improve up-time, optimize asset utilization, and reduce overhead costs. Migration to IT technology enables the IIoT operator(s) to deploy, monitor, and optimize their IIoT application. Standardization around IT practices helps to eliminate islands of proprietary equipment within the installation and provide tighter integration between the control domain and the operations domain. Adoption of IT methodologies enables IIoT companies to leverage the large existing base of IT hardware and software solutions when appropriate. Each of these benefits offers significant potential for capital and operational savings.
Standardization of the upstream interfaces for controller devices and meta-data models for sensors can help solve hardware and software interoperability and ease-of-use issues. Standardized interfaces would allow dissimilar pieces of hardware to communicate with the IIoT command center in a uniform fashion and eliminate isolated islands within the installment. Likewise, an extensible standardized meta-data model for sensors would allow for systematic detection and control of sensors and control points without extensive code re-writes. From a hardware standpoint, the IIoT marketplace would also benefit from greater standardization around communications interfaces, power, and environmental requirements.
Large industrial automation suppliers are not incentivized to embark on open standardization because it loosens the customer’s dependence upon their proprietary solutions. Smaller automation suppliers lack the industry clout or size to take on such an ambitious undertaking. This is a task best suited for an industry standards organization, and one which PICMG is well equipped to handle.
COM Express is one logical starting point to build upon because it has the small form factor, processing performance, and flexible I/O configuration to make it a natural fit for small gateways and control functions in small to medium installations, with distributed controllers for larger deployments. In larger installations, CompactPCI Serial or MicroTCA have been adapted for railway control and other rugged applications and may also serve as a flexible gateway/controller.