Digitalization offers a wide range of benefits, including predictive maintenance to reduce downtime by creating digital twins, enhanced quality control, demand-driven production, inventory optimization, lower energy and material costs, and improved safety and environmental performance.

Many forecasts attempt to quantify the value proposition. McKinsey, the consultancy, says the economic impact of the Internet of things could be between $1.20 and $3.7tn by 2025. A recent Commerce Department survey of U.S. manufacturers and smart manufacturers showed annual cost reductions of $57 billion.

Of course, there is one problem, actually several. Investment cycles in manufacturing are long, powerful processes and equipment do not appear overnight and, crucially, the technologies needed, such as artificial intelligence, have not yet been fully developed.

Artificial intelligence (AI) as a catalyst

Smart factories leverage the convergence of the Industrial Internet of Things (IIoT), big data and advanced analytics, as well as information technology (IT) and operational technology (OT). In addition, devices that communicate with each other lead to real-time decisions that optimize value creation.

It happens both inside the factory and throughout the value chain, from raw material procurement to order delivery and customer service.

The potential catalyst for this shift is artificial intelligence (AI). Much of the current interest in AI has to do with machine learning -- a set of techniques that combine real-world data and experience with statistical analysis to draw conclusions and predict outcomes.

Machine learning is not a new field of AI, but the growth of the Internet, the proliferation of large amounts of data, and the increasing processing power of computers have greatly increased the depth, breadth, and accuracy of its predictive power.

While AI is clearly improving, it has its limits. The underlying algorithm is tricky to design, which can lead to bugs and unexpected deviations; Training steps often require a very large amount of data and practical experience that may be difficult to obtain; Neural networks usually take a long time to train. When a decision to enable artificial intelligence (AI) goes wrong, it is often difficult to determine why, which is a major problem in safety-critical systems.

Why is artificial intelligence now being used in factory environments? Technology is, of course, a driving factor: the availability of large amounts of data, the growth of machine learning, the emergence of cloud computing (for network-wide monitoring and optimization) and edge computing (to provide machine learning for real-time decision making), and the integration of information technology (IT) systems with operational technology (OT) systems.

But current social trends are also important, including the increasing complexity of global supply chains and the continuing challenges in attracting skilled production workers. In other words, the emergence of smart factories is the result of technology drive and market pull.

If all the AI problems can be solved -- and eventually will be. However, without the best information governance, the smart factory will still not develop rapidly.

Three such governance issues include technology standards, network security/privacy, and spectrum allocation.

The technical standards

Smart factories rely on information flow and system responsiveness and cannot be achieved without standards -- basically specifications or requirements related to technical systems.

Hundreds or even thousands of standards are used in the manufacturing process, and many new standards are needed to realize the smart factory. According to a February 2016 report from the National Institute of Standards and Technology (NIST), the smart manufacturing ecosystem can be viewed as a pyramid composed of four progressive levels: the device level, supervisory control and data acquisition (SCADA) level, manufacturing operations management (MOM) level, and enterprise level. Information must flow within and between each level, and dozens of standards have been developed or are being developed to speed up this collaboration.

According to NIST, "Within the manufacturing pyramid, communication standards are established, but interoperability between systems is limited, meaning that manufacturers are often locked into a single vendor solution. There are several well-established standards throughout the business cycle, however, the extent to which information can be interlinked with production systems is very limited."

In addition to developing standards to fill these gaps, the report also identifies two other standards-related barriers facing smart factories:

(1) Lack of tracking of standards and their adoption;

(2) overlap and redundancy between standards.

In order to remove these obstacles, coordination and cooperation among the various organizations are necessary, and some of these are under way.

Standards are also being developed to facilitate the adoption of blockchain technology. Blockchain is a digital ledger that records transactions in a verifiable and secure manner. The Department of Homeland Security (DHS) is piloting blockchain with industry to see if the technology can stop counterfeiting and intellectual property theft. Security and defined interoperability standards will be needed to facilitate the adoption of the technology.

Network security/privacy

Smart factories require interconnection between devices and devices both within the factory and throughout the value chain. Such connections increase the risk to manufacturers of cyber attacks, espionage and data theft.

These are not hypothetical problems, such as the one that disrupted a German steel mill in 2014 after hackers gained access through phishing emails. A recent survey in the UK found that 50 per cent of manufacturers admitted to being hacked and half of those affected suffered losses as a result. Manufacturers are prime targets for cyber attacks on critical infrastructure, according to the Department of Homeland Security.

As smart factories increasingly target traditional factories, security issues become more important. Security objectives include maintaining production (without downtime or delay), preventing system failures that result in property or personal injury/death, preventing espionage, and protecting the privacy of customers and employees.

Achieving these goals will be neither simple nor easy. To protect a smart factory, a variety of approaches and systems are required, including a secure architecture of the network's physical systems, verification of software integrity through certification (the process of being able to detect malicious software or unexpected code), and secure device management.

Suppliers of intelligent manufacturing equipment and services are clearly involved in these security developments, as are governments. The U.S. government has worked with industry to develop a risk-based and voluntary cybersecurity framework for critical infrastructure that is widely applicable to a range of businesses, including manufacturers. NIST has also released a Smart City Framework related to Smart Factories.

Another growing problem concerns the privacy of personal information. The EU's General Data Protection Regulation (GDPR) is a legal framework that sets out guidelines for the collection and use of personal information. The new law also has implications for smart factories, where, for example, technology that measures production from a production line may collect data on individual workers. Manufacturers need to ensure that they are transparent about the personal information they collect using these technologies by updating privacy statements and ensuring that these statements comply with GDPR requirements.

Finally, the smart factory will drive change in the insurance industry, which will face the need to build solutions to manage risk changes.

Spectrum allocation

The number of devices needed to fulfill the promise of a smart factory is an important consideration in information governance. The devices are expected to operate via wireless communication. There are currently billions of wireless devices, and this number is expected to grow exponentially thanks to the Internet of Things and the Industrial Internet of Things (IIoT).

All this demand for wireless communications requires spectrum, a scarce public resource. For smart factories to succeed, governments must allocate enough spectrum to meet this growth in demand. In the United States, the Federal Communications Commission (FCC) allocates spectrum for consumer and commercial use.

Last year, the US Government Accountability Office (GAO) investigated the issue. According to the GAO report, the FCC believes that the spectrum currently available is sufficient to meet the growth of the Internet of Things in the near future, unless devices that use a lot of spectrum proliferate. Gao also noted that "as the number of wireless devices grows, managing interference becomes increasingly challenging, especially in frequency bands where wireless licenses are not required." The GAO recommended that the FCC start tracking the growth of the Internet of Things to make sure there is enough spectrum available.


If additional spectrum is needed to support a smart factory, then is it the licensed spectrum? Unlicensed spectrum or shared spectrum? The FCC will decide how to allocate the available spectrum between each type and within which bands.



These government decisions will affect the availability and quality of spectrum for smart factories in the United States. Other countries are also grappling with how to allocate spectrum for industrial uses. The GAO report notes that each country is taking a different approach, with at least one, South Korea, dedicating spectrum to industrial use.