How does the Industrial Internet relate to your work at Vanderbilt’s Institute for Software Integrated Systems (ISIS)?
My primary research interest is the model-integrated design of cyber-physical systems (CPS). CPS are engineered systems created as networks of interacting physical and computational processes. Most modern products in major industrial sectors, such as automotive, avionics, medical devices, or energy production and distribution rapidly becoming – or already are – CPS driven by new requirements and competitive pressures. Science and technology advancements in the 20th century have produced methods and tools for designing computational and physical systems in isolation. However, these methods have proven to be inadequate in a large range of CPS applications, where computational and physical processes are so tightly integrated that it is not possible to identify whether behavioral characteristics are the result of computations (computer programs), physical laws, or both working together.
The Industrial Internet represents the networking aspect of CPS; therefore, it is a fundamental enabler and crucial technology component for important categories of CPS applications. There are unique challenges for networking in CPS that must influence Industrial Internet design. (1) Physical time is essential in all CPS applications. Networks are frequently used as virtual implementation media for physical interactions; therefore, networking uncertainties (delays, pocket drops) directly influence physical system dynamics and essential properties such as stability and safety. (2) Dependability needs to be explicitly addressed and provisioned in CPS network connections. Determined by the level of criticality of CPS functions, network errors may introduce significant safety hazards in operation. (3) Networks may be the source of unacceptable security vulnerabilities in CPS applications. Industrial Internet protocols need to be designed by keeping security as a first-order design objective. A well-designed Industrial Internet infrastructure will tremendously help the establishment of a correct-by-construction design process for CPS.
How does the Industrial Internet differ from the legacy Internet?
There is an explosion of interest in CPS design technologies because of the huge industrial stake involved. It is a disruptive technology that will change the status quo by creating new industries and eliminating others. Those who are able not only to understand but to create the new science and technology foundations will determine the future trajectories of whole industrial sectors. The Industrial Internet is a fundamental component of the emerging new science and technology infrastructure.
The term captures well that there is a key difference between the legacy and the Industrial Internet. The Industrial Internet is a networking concept that is designed for network-based interaction among physical machines. Much of our problems with the industrial applications of the legacy Internet emerge from the fact that it was not designed for industrial use cases requiring precise timing, high dependability, and high security. Still, economic factors, competition, and functional benefits drove industry to build ever more complex networked CPS applications on an inadequate network infrastructure leading to massive problems such as pervasive security risks in SCADA systems. The Industrial Internet concept is proactive: it prepares for the upcoming wave of fine-grain networking applications, for extensive use of cloud-based deployment of safety critical IT systems, and for the dominance of CPS in all engineered systems.
What advances do you expect to see in the Industrial Internet in the next decade?
My perspective reflects the needs of correct-by-construction design of complex CPS applications. Correct-by-construction design means that essential properties of the implemented/manufactured CPS system can be established with the required level of accuracy during the design process. This goal is important because in CPS systems, testing-based property verification is all but impossible, and correction of identified design errors is expensive or infeasible. I expect to see in the Industrial Internet the following advancements during the next decade.
1. Predictable timing - Many CPS applications include control loops closed via networks. In these applications, predictability of timing properties of the networks is essential for verifiability of basic dynamic properties of the integrated systems. Since increasing timing accuracy is expensive, applications should be able to request timing accuracy according to their actual needs.
2. Timing consistency - Time is a fundamental independent variable in physical systems and interactions. Many CPS applications from small to global scale require the establishment of consistent temporal relations and distances among events. It means that the Industrial Internet needs to address the issue of global time base.
3. Support for heterogeneous models of computation - Heterogeneity of CPS applications requires the use of a range of models of computation from reactive, event-triggered to synchronous, time-triggered interaction models. The Industrial Internet infrastructure should support these different models simultaneously with adjustable timing accuracy.
4. Many CPS functionalities require event processing and reasoning along causal chains. A necessary condition for this functionality is determinism in the basic mechanisms used for event detection and transmission.
5. Safety and Security - There are a large number of planned applications, such as car-to-car or car-to-infrastructure communication that are not only timing, but simultaneously safety and security critical. The Industrial Internet needs to deliver safety and security in a flexible, demand-driven manner. It needs to be designed such that applications can define the required level of safety and security guarantees.
6. Naming - The vision is of a world with a trillion sensors from which data can be harvested, and equally huge number of smart objects where data can be sent to demand a naming architecture that enables searching and identification of objects with variable resolution under timing constraints. The Industrial Internet will include a new naming architecture that provides search and access services with the required level of security and privacy restriction. This list is far from being complete. Synchrony, dependability, and security are challenging topics, and the development of a new networking infrastructure for CPS would represent a huge step ahead.
Dr. Janos Sztipanovits holds the E. Bronson Ingram Chair in Engineering at Vanderbilt University School of Engineering. He is a Professor of Electrical Engineering, Professor of Computer Engineering and Director of the Institute for Software Integrated Systems.
Dr. Sztipanovits researches the model-integrated design of cyber-physical systems.