How does the Industrial Internet relate to your work at Vanderbilt’s Institute for Software Integrated Systems (ISIS)?
The emergence of the Industrial Internet as a viable business and technology paradigm demonstrates the maturation of the types of work that researchers atISIS and elsewhere have been conducting over the past several decades to (1) connect machines embedded with sensors and sophisticated software to other machines (and end users), (2) enable access and control of mechanical devices in unprecedented ways, (3) extract data from these devices, make sense of it, and deliver the right information to the right people at the right time (and in real-time), and (4) derive some form of value in terms of improved utility and/or cost savings. At the core of this work are advances in systems science and engineering technologies, including quality-of-service (QoS)-enabled middleware, end-systems, and networks; methods and tools for assured integration of components in heterogeneous systems, time-bounded machine learning; and model-integrated computing.
ISIS researchers have developed and applied open-source QoS-enabled middleware frameworks and model-integrated computing tools that are used in thousands of projects around the world. Some of the most vexing research and development challenges we’ve encountered are those associated with creating, validating, deploying, and sustaining mission- and safety-critical enterprise distributed real-time and embedded (DRE) systems, such as air trafﬁc management, supervisory control and data acquisition (SCADA) systems (e.g., for managing smart power grids),integrated health care, integrated air and missile defense, and electronic trading systems. The Industrial Internet is an example of the broader shift towards enterprise DRE systems, where the right answer delivered too late becomes the wrong answer.
Enterprise DRE systems have become more dynamic; larger in topology, scope, and data volume; and more sensitive to end-to-end latencies than traditional real-time and embedded systems. The key challenges we face when ﬁelding these systems stem from having to (1) organize the information so producers and consumers can address and locate each item of interest, (2) distribute a high volume of messages per second while dealing with requirements for scalability and low/predictable latency, (2) maintain availability during bandwidth ﬂuctuations and service disruptions, and (4) analyze telemetry data to self-diagnose and self-correct at runtime. Moreover, assuring end-to-end quality-of-service (QoS) properties (such as dependability and security) is hard because end-system QoS mechanisms must work across different access points and within—and across—network domains.
Why did you choose Industrial Internet as one of the Top Ten Tech Terms for 2014?
I chose to focus on the Industrial Internet because many people are only familiar with conventional information technology (IT) products and tools like
- personal computing devices, such as smart phones, tablets, or laptops,
- office productivity tools, such as word processors or spreadsheets, and
- infrastructure services, such as routers and servers.
Increasingly, however, our safety, health, and privacy in the air and in our cars, in healthcare delivery and record-keeping, in the water and food supply, and in energy grids and other utilities, depend on complex and interconnected machines that perform their tasks relatively autonomously. In particular, over 90% of all processors are now used to control physical, chemical, or biological processes and devices in real-time. As these processors are connected together to span multiple computing, communication and physical sub-systems, the resulting enterprise DRE systems must be developed, validated, and managed rigorously, despite the complexity inherent in those ever-larger combinations.
When this capability is harnessed properly, it creates knowledge that enables us to act quickly, save money, and produce better outcomes. When poorly understood or improperly applied, on the other hand, the consequences can be catastrophic. An educated workforce and user base is therefore essential to devise, test, and operate/supervise machines that can sense conditions, communicate dependably, and become instruments of effective decision making and economic value.
For example, an intelligent urban transportation system may automatically provision trains and buses based on regular usage patterns (such as the time of day), knowledge of special of events, (such as concerts or sports event), and real-time feedback (actual loads experienced on the trains and buses). If this system functions as expected the riders will gain significant utility. If it fails, however, there will be chaos, with many people missing their appointments, excess fuel consumption and carbon emissions due to traffic congestion, etc.
What advances do you expect to see in the Industrial Internet in the next decade?
Depending on the availability of sustained funding, effective collaborations between researchers and practitioners, and visionary technical and business leadership in government and industry, I see the following advances occurring in the Industrial Internet—and more broadly in the domain of enterprise DRE systems—in the next decade:
- Precise auto-scaling and management of resources with a dynamic and system-wide focus. Auto-scaling in the Industrial Internet requires applications that indicate their resource needs a priori so providers can effectively scale the resources up or down. For example, when load increases, services are provisioned with higher demand on existing resources and potentially given access to new ones, while resources also can be de-provisioned or loaded less heavily when load reduces. The Industrial Internet will be composed of interacting services, so they will require dynamic resource management and auto-scaling algorithms that operate at the level of service groups working together in end-to-end task chains, while ensuring that end-to-end QoS requirements are met.
- Flexible optimization algorithms to balance real-time constraints with cost and other goals. Since applications and services in the Industrial Internet will be realized as end-to-end real-time task chains, their deployment on computing and networking resources must be schedulable to ensure real-time response times, while optimizing desired objective functions, such as minimizing operational costs. These requirements must be met in the context of the auto-scaling algorithms mentioned above. Due to different criticalities of task chains that could be deployed on the resources, principled means for co-scheduling or performing admission control and/or eviction of mixed-criticality task sets will be needed.
- Data provisioning and load balancing algorithms that rely on physical properties of computations. The Industrial Internet will generate load on a computing environment due to a combination of cyber and physical stimuli, such as traffic, power grid fluctuations, human movement and changing weather patterns. To build the most scalable and high-performance systems, algorithms and techniques are needed to exploit physical characteristics of data and computation to improve the distribution of work in the Industrial Internet. For example, data may need to be clustered onto nodes based on geographic associations, social network linkages, or other physical world aspects. Understanding the relationship between physical world aspects and cyber optimizations to improve scalability and response time of cloud systems is critical to support the Industrial Internet.
- Embodiments of these technology advances in the form of standards-based commercial-off-the-shelf middleware. Enterprise DRE systems have traditionally been developed in a stove-piped manner that locks customers and users into a limited number of system integrators, each designing their solutions using proprietary approaches to system architecture and integration. These stovepipes have yielded many redundant “point-solutions” that are prohibitively expensive to develop, certify, and sustain. Advances in standards-based, commercial-off-the-shelf middleware (such as the Object Management Group’s Data Distribution Service) will enable application developers and systems integrators of the Industrial Internet to achieve seamless interoperability between hardware and software via common programming interfaces, communication protocols and data models.
I mention these examples since they relate to existing and emerging priorities that we are working on at ISIS. Our strengths in this area stem from our ability to produce not only solid theories and fundamentals, but also methods and high-quality open-source platforms and tools that enable researchers and practitioners to access and apply our results to foster rapid transition of enterprise DRE technology to academia and industry in a range of mission- and safety-critical domains, including the Industrial Internet.
Naturally, there will be other advances, as well. For example, the Industrial Internet offers a vision that can potentially be extended through engagement and enrichment from other stakeholders. For example, researchers funded by theNational Science Foundation and other agencies are beginning to examine how precise sensing, control, and actuation from cyber-physical systems (which feature a tight combination of—and coordination between—a system’s computational and physical elements) can be integrated with scalable and dependable internetworking and cloud computing approaches. The goal is to enable a new generation of safe, secure, reliable, resilient, enterprise DRE systems in areas of vital national interest, including energy, transportation, communication, defense, and food and water distribution.
Dr. Douglas C. Schmidt is a Professor of Computer Science and Associate Chair of the Computer Science and Engineering Program at Vanderbilt University.