How does the Industrial Internet relate to your work at Vanderbilt’s Institute for Software Integrated Systems (ISIS)?

At ISIS, we have been working with embedded systems and networked systems of systems for many years. These are exemplified by multiple compute devices performing local, isolated tasks and working together to accomplish a larger goal. Examples of this include:

  • Automated factories, where local controllers operate machines but work together to build a whole vehicle.
  • Military command and control, where distributed units sense and move and combine this information into a global situational awareness and control.

While any single device is relatively simple, the combination of multiple devices and their interactions can become very complex. On the other hand, the systems and subsystems are designed to work together. Simplifying assumptions can be made about individual units, and tight specifications can be asserted to control how the subsystems interact. Finally, the system configuration can be relatively static, or at least change in a controlled, managed way.

My research at ISIS has addressed these distributed systems issues using models, or abstracted views, of the problem. Models capture specifications and behavior of the subsystems, and the interconnection between systems. Designers can reason about the properties of the composite system via analysis and simulation. There are many interesting properties, including performance, safety, security, power use, etc. The Industrial Internet presents new twists to the distributed systems problem:

1. Devices are designed independently and without knowledge of the ways they will be used in a system.

2. Systems are not static, predefined sets of devices, but an ever-expanding combination of diverse, multi-sourced subsystems. Dealing with these dynamic configurations, and determining the properties in advance of their occurrence represents a challenging yet very important task.

Why did you see the Industrial Internet as one of the major technology trends of 2014?

The adoption of devices with processing and communications capability is rapidly expanding, approaching an exponential pace:

  • This technology is essential as competitive forces push for more efficiency in almost every area.
  • Cost reduction in embedded systems will enable virtually everything to be interconnected. Lower cost IC’s enable this functionality to be pushed into lower cost products.

With this capability comes significant risk. As systems become more connected and gain more physical control, improper operation can cause actual physical damage. This can result from incorrect design, due to unexpected interactions between components, or from intentional sources, entering thru insecure systems.

Because of the growth in coverage, and the potential for positive and negative impact, I believe that 2014 will see the Industrial Internet as a hot topic.

What advances do you expect to see in the Industrial Internet in the next decade?

The pervasiveness of the devices will increase, propagating deep into the value chain of product lines. Tools and infrastructure will evolve to maximize their benefits and minimize risks. For tools, I expect to see systems that effectively protect themselves from malfunction and intrusion. These tools must assess and predict the behavior of Industrial Internet systems, as capabilities and components are added. Specifically, these tools should rapidly assess the impact of a new device or capability on security, performance, safety, lifespan, etc. This capability will require:

  • Devices models and metadata that specify their behavior, protocols, modes, etc.
  • Network models that specify the interaction methods and architecture for communication between devices.
  • Application models that specify the requirements for system behavior and interaction.

These models must be used to ensure that system requirements are satisfied dynamically as networks expand and contract and include new devices. Properties that are at risk, such as security or timeliness, will be evaluated. Changes that violate properties can be rejected by excluding devices, or disallowing changes. The tools and devices will combine to produce a reliable, resilient, and secure Industrial Internet.

About the author

Renee Ferguson