This chapter discusses the transformative nature of cyber-physical systems (CPS), their importance to industry, and the associated workforce needs. It looks at broad indicators of the economic importance of CPS applications as well as testimony presented to the committee about the CPS skills sought by industry.
The engineered world has seen a major transformation during the last few decades. Elements that previously existed in purely mechanical or electrical (i.e., physical) form, and in particular those elements describing logic, control, and decision-making, increasingly take the form of embedded systems and software (i.e., cyber elements). The acronym CPS is often used to describe “engineered systems that are built from, and depend upon, the seamless integration of computational algorithms and physical components.”1 In this definition, “cyber” refers to the computers, software, data structures, and networks that support decision-making within the system, and “physical” denotes not only the parts of the physical systems (e.g., the mechanical and electrical components of an automated vehicle) but also the physical world in which the system interacts (e.g.,
1 Definition from National Science Foundation (NSF), 2016, “Cyber-Physical Systems,” program solicitation 16-549, NSF document number nsf16549, March 4, https://www.nsf.gov/publications/pub_summ.jsp?ods_key=nsf16549.
roads and pedestrians). CPS is closely related to terms in common use today, such as Internet of Things (IoT), the Industrial Internet, and smart cities, and to the fields of robotics and systems engineering (Box 1.1).
Several emerging technology trends support the increased deployment of CPS:
- Communication networks, databases, and distributed systems allow control and decision-making on physical systems to be done remotely, collaboratively, and in a distributed manner, which is enabling functionality impossible a few years ago.
- The developments that have given rise to the field of data science make it possible to collect, store, analyze, and act on large amount of real-world data.
- Decreasing costs of components and systems have allowed the use of CPS within everyday devices such as home thermostats and automobile brakes. For example, lower cost sensors are being deployed across the board, from the use of sensor nets to detect approaching natural disasters such as flooding and earthquakes to those that support safer car travel.
- Wide deployment and increased reliability of high-speed wireless networks support devices that rely on a continuous connection to the Internet.
CPS can be small and self-contained, such as an artificial pancreas, or very large and complex, such as a regional energy grid. They are increasingly used to provide economically or societally important capabilities, many with critical infrastructure or life-safety implications (Box 1.2). CPS can provide extraordinary flexibility by allowing unprecedented growth in economy, functionality, safety, performance, and accuracy of control and operational decision-making. Indeed, virtually all industries have embraced CPS. A recent McKinsey Global Institute report on the IoT, for which CPS provides the technical foundation, captured some of the economic importance of CPS applications succinctly by stating, “the hype has been great—the value may be greater.”2 The McKinsey report estimates a potential worldwide economic impact of as much as “$11.1 trillion per year in 2025 for IoT applications in nine settings”—devices attached to or inside the human body, homes, retail environments, offices, factories, custom production environments, vehicles, cities, and other outside settings.3 Gartner recently forecast a 30 percent increase in the number of
2 McKinsey Global Institute, 2015, The Internet of Things: Mapping the Value Beyond the Hype, June, http://www.mckinsey.com/business-functions/business-technology/our-insights/the-internet-of-things-the-value-of-digitizing-the-physical-world.
3 Ibid, p. 2-3.
“connected things” from 2015 to 2016 and a threefold increase to over 20 billion devices in 2020.4 A related concept is the Industrial Internet, which combines IoT and big data analytics for industrial applications. A 2015 report from GE and the consulting firm Accenture cites projections
that worldwide Industrial Internet spending could reach $500 million by 2020 and be responsible for as much as $15 trillion of the global economy by 2030.5 At the same time, firms in the information technology sector are increasingly investing in CPS areas such as self-driving cars (e.g., Google and Uber) and the IoT (e.g., IBM).
Speaking to the potential of CPS and the technical challenges of realizing that potential, in testimony to the House Committee on Science and Technology in 2008, Don Winter, vice president for engineering and information technology at Boeing Phantom Works, observed the following:
Cyber-physical systems are pervasive at Boeing, and in the aerospace industry at large. They are becoming increasingly prevalent in other sectors, notably automotive and energy management. Their importance to our products is huge and their complexity is growing at an exponential rate.6
The contribution of CPS to aerospace systems has grown dramatically, noted Winter, having risen from less than 10 percent of the design, development, validation, and certification cost for transport aircraft in the 1970s to about 50 percent by the 2000s.
It is worth observing that even as it offers enormous safety benefits, the adoption of CPS also introduces new risks. For example, although it is also susceptible to failure, a purely mechanical linkage may be less dangerous than separate sensors and actuators that could lead to failure and injury as a result of a software mistake, hardware malfunction, or cybersecurity attack. These risks magnify the need for a highly skilled workforce.
Foundational advances resulting from academic research will support the next generation of CPS that can be designed, implemented, deployed, and maintained to meet requirements using emerging functional and nonfunctional properties. Advances in achieving functional properties allow new solutions to be realized; for example, tomorrow’s solutions will allow micro-electric grid transactions for higher energy efficiency and disease prevention (not just maintenance). Advances in achieving nonfunctional properties (i.e., security, safety, reliability, and dependability) will enable future systems to operate with increased confidence in the presence of risk—for example, realizing confidence in city-scale autonomous transportation systems.
5 General Electric and Accenture, 2014, Industrial Internet Insights Report for 2015, http://www.ge.com/digital/sites/default/files/industrial-internet-insights-report.pdf, accessed November 1, 2016.
6 Don C. Winter, 2008, Testimony at a hearing on the Networking and Information Technology Research and Development (NITRD) Program, Committee on Science and Technology, U.S. House of Representatives, July 31.
The National Science Foundation (NSF) has an ongoing CPS research program that was given additional impetus by recommendations of August 2007 and December 2010 reports of the President’s Council of Advisors on Science and Technology.7 Reflective of both the diverse applications of CPS and its importance for progress in many sectors, the NSF program works with a wide array of federal mission agencies: the U.S. Department of Homeland Security’s Science and Technology Directorate; the U.S. Department of Transportation’s Federal Highway Administration and Intelligent Transportation Systems Joint Program Office; the National Aeronautics and Space Administration’s Aeronautics Research Mission Directorate (ARMD); several institutes and centers of the National Institutes of Health; and the U.S. Department of Agriculture’s National Institute of Food and Agriculture.8
The National Institute of Standards and Technology has established a Cyber-Physical Systems and Smart Grid Program Office pursuing research and the development of architectures, frameworks, and standards for CPS and CPS applications.9 Other federal CPS research initiatives include the Defense Advanced Research Projects Agency’s Adaptive Vehicle Make and High-Assurance Cyber Military Systems programs and the Department of Transportation’s Connected Vehicle and Intelligent Transportation Systems program. CPS research initiatives can also be found in many other countries (Box 1.3).
It proved difficult for the committee to obtain comprehensive data on demand for CPS skills and knowledge. It is especially challenging to gather systematic information of the sort requested for an emerging, highly interdisciplinary field like CPS. It is likewise difficult to gather even anecdotal information from smaller firms because they tend not to have readily identifiable points of contact on these issues. No surveys appear to have been conducted on industrial demand for skills or of CPS-related university programs in the United States. Nor do current
7 From the President’s Council of Advisors on Science and Technology reports Leadership Under Challenge: Information Technology R&D in a Competitive World: An Assessment of the Federal Networking and Information Technology R&D Program, August 2007, and Designing a Digital Future: Federally Funded Research and Development in Networking and Information Technology, December 2010, https://www.whitehouse.gov/administration/eop/ostp/pcast/docsreports; and NSF, 2016, “Cyber-Physical Systems,” program solicitation 16-549.
8 NSF, “Cyber-Physical Systems (CPS),” https://www.nsf.gov/funding/pgm_summ.jsp?pims_id=503286, accessed November 1, 2016.
government statistics provide sufficient granularity to separate out CPS positions from other computing or engineering jobs. The committee was not in a position to commission systematic surveys of either industry or academia to collect such information itself.
Lacking comprehensive data about workforce needs in CPS, the committee relied on the perspectives of industry experts who participated in
the two workshops convened during its study as well as a set of briefings. A list of all workshop speakers or briefers to the committee, which included a number from industry, can be found in Appendix B.
Workshop speakers representing a wide array of industry sectors—automotive, agriculture, medical devices, and space, along with a large industrial conglomerate and a vendor of CPS engineering software tools, discussed the changing nature of their products, the array of new skills needed in their engineering workforce, and the challenges they face in developing the necessary talent. A summary of some of their observations is provided in Box 1.4. People from diverse industry sectors reported that they needed people with CPS engineering skills. In some cases, products
were not being developed because there were not enough people available with the CPS skills necessary to do the job. In other cases, people from industry noted that their workforce would be restructured if more CPS-educated individuals were available.
Speaking to the demand for CPS skills, Joseph Salvo, director at GE Global Research observed that “going forward . . . almost all of our employees are going to be touched by this.” Asked how many CPS engineers Ford Motor Company needed, Craig Stephens, from Ford’s Research and Advanced Engineering organization, responded “[the] short answer is, more than we can get.”
Given the prevalence of CPS throughout industry, the work of many
engineers revolves around CPS, whether they consider themselves experts in this area or not. Many have not received formal education or training in key CPS topics, such as formal methods, verification, or security, and may not fully understand the challenges of designing the software or physical systems for life-critical systems.
Ad hoc CPS system design and implementation runs the risk of not supporting the scalability, security, and design flexibility required to meet today’s and tomorrow’s needs. This is of particular concern given the role CPS plays in mission- and safety-critical systems, and the cybersecurity challenges faced with all computer systems. Better education and training and the development of a CPS discipline is therefore a priority, since many if not most of the systems that society relies on will be CPS.
Developing effective CPS solutions requires a workforce that has the right mix training and skills. This workforce will include skill levels ranging from those who can help develop sophisticated capabilities to those who can help deploy and maintain CPS solutions over long periods of time. Engineering projects are by nature collaborative, and engineering teams involve a range of expertise, including CPS.
Accordingly, a variety of educational and training regimes will be needed. The multidisciplinary skills required will build on existing workforce capabilities in areas of engineering, computer science, and information technology. To that end, part of the effort will need to focus on supplementing the skills of the existing workforce, while another part will need to focus on a future workforce that has all prerequisite skills built in from their education.
FINDING 1.1: CPS are emerging as an area of engineering with significant economical and societal implications. Major industrial sectors such as transportation, medicine, energy, defense, and information technology increasingly need a workforce capable of designing and engineering products and services that intimately combine cyber elements (computing hardware and software) and physical components and manage their interactions and impact on the physical environment). Although it is difficult to quantify the demand, a likely implication is that more CPS-capable engineers will be needed.
The emergence of a new field such as CPS from preexisting domains of knowledge is not a new occurrence. In fact, analogies can be drawn to the history of computer and software engineering. Electrical engineers in the 1940s could not have conceived of computers as commodities. Then, a computer was a very large room packed with rack after rack of hot
vacuum tube assemblies, relays, huge power supplies, and the unenviable punched card reader and line printer. The field of computer engineering slowly emerged as a separate discipline and practice. The separate discipline and practice of software engineering later answered the need for people to more easily and effectively program the computers. It should come as no surprise that, much the same way that an army of electrical engineers is no longer required to build a computer, there is no longer the need for armies of varied engineering disciplines required to build, program, and employ small processors with sensors and controllers (either attached or built in) as components in other systems—or, for that matter, as systems themselves. However, although the components and tools for designing small, embedded systems are accessible to a hobbyist, the skills and knowledge necessary to develop a large system with verifiable reliability and safety requirements are considerable.
Following a similar pattern, CPS incorporate components of disciplines such as embedded systems, software engineering, control systems, networking, and systems engineering. In fact, domains such as aerospace and mechanical engineering and related fields such as robotics have incorporated many CPS principles for some time. The experts in this nascent field will be experts on this intersection of disciplines.
FINDING 1.2: The future CPS workforce is likely to include a combination of (1) engineers trained in foundational fields (such as electrical and computing engineering, mechanical engineering, systems engineering, and computer science); (2) engineers trained in specific applied engineering fields (such as aerospace and civil engineering); and (3) CPS engineers, who focus on the knowledge and skills spanning cyber technology and physical systems that operate in the physical world.
FINDING 1.3: Given that most entry-level engineering and computer science positions are filled by undergraduates, it is important to incorporate CPS into the undergraduate engineering and computer science curricula.
RECOMMENDATION 1.1: The National Science Foundation, together with universities, should support the creation and evolution of undergraduate education courses, programs, and pathways so that engineering and computer science graduates have more opportunities to gain the knowledge and skills required to engineer cyber-physical systems. The efforts should be complemented by initiatives to augment the skills of the existing workforce through continuing education and master’s degree programs.