By Thomas Rosen
The Industry IoT Community Council is an industrial-focused executive community comprised of business leaders, technical experts, and academics representing a diverse array of global industries. The group convenes periodically to share insights regarding the impact and potential benefits of incorporating new and emerging technologies in day-to-day business operations. Following is a recap of an Industry IoT Community Council session focusing on the concept of A New Intelligent Reality.
Making the move to an intelligent systems business model
Industrial organizations are increasingly exploring the potential benefits to be had by integrating intelligent systems into business processes and workflows. Intelligent insights derived by leveraging Big Data, artificial intelligence, machine learning, and intelligent edge systems and made accessible in near real time via 5G and the cloud, have given rise to A New Intelligent Reality. The technology framework is well in place. But are industrial organizations ready to take advantage of the opportunities?
Intelligent systems + cloud = greater operational efficiencies + valuable business insights
In the course of our discussion, the IoT Industry Community Council panelists observed that leveraging AI and machine learning for generating business intelligence and automating industrial applications has great potential for making the workplace more efficient and their jobs more productive — today, and especially over the course of the next five years. That potential is clearly reflected in the numbers: According to Gartner, enterprise use of AI grew 270% between 2014 and 2019,(1) while Forbes noted that its market size was valued at $27+ billion in 2019, projected to reach nearly $267B by 2027. (2)
The use cases cited by Council members ranged from applications in traditionally low-tech industry sectors to those on the cutting edge. For instance:
● Livestock management — In Sweden, farmers are using IoT devices and AI to keep tabs on the cattle under their watch. With sensors humanely affixed to a cow’s tail and geolocation data transmitted wirelessly to AR headsets, farmers can easily find and round up any strays. Low-tech has suddenly become high-tech, and it’s changing the industry for the better.
● Manufacturing — AI, machine learning, and smart algorithms are playing a growing role in factory environments. Electronics manufacturers are leveraging real-time digital and data feedback loops to train algorithms to inspect machinery, gauge equipment utilization, devise proactive maintenance strategies, and conduct QA oversight on product output. With no need to take production offline, intelligent systems and predictive analytics are helping manufacturers rethink factory operations.
● Inventory control — In a variety of supply chain settings, smart, connected pallets are collecting data to help companies analyze loss due to damaged goods and improve overall inventory management. In one instance, over the course of a year, some 63M pieces of data from 25 manufacturing sites were crunched to help identify when damage to goods was most likely to occur — the night shift!
● Assembly lines — While AI and machine learning in manufacturing environments are beginning to make headway in predictive use cases, the technologies are still predominantly used for driving manual operations. In one case, pattern recognition of repetitive tasks is enabling industrial robots to assist in the assembly of aerospace turbines.
● Safety-critical processes — Digital twins are being use to model, emulate, and simulate plant operations in order to improve control systems and automate safety-critical processes at nuclear power plants.
A few bumps on the road to digital transformation
As noted, the digital framework for embedding intelligent systems in industrial environments is fairly well established. However, navigating the road forward does requires overcoming a number of obstacles. What are some of the challenges our panelists are facing and what are their main concerns?
Security: IoT devices are simple devices that were not necessarily designed with security in mind, making them vulnerable. They are highly susceptible to cyber threats, and because they can be compromised, some operators of industrial applications feel that they are losing control of their business. Plus, ingesting data from diverse systems and multiple vendors, and the lack of standard methodologies for connecting to devices further heightens security concerns. Some numbers tell the story:
● In 2020, IoT devices were responsible for nearly 33% of infections in wireless networks, more than twice that in 2019. Industries with higher dependency on software and intelligence-based systems now need to ensure that security is addressed at every step in an intelligent system’s lifecycle. (3)
● Two thirds of IoT developers use open source and one-third are using cloud-native architectures. Their top three concerns in digital transformation are security, connectivity, and data collection and analytics. (4)
In today’s enterprise environment, the network security perimeter traditionally guarded by firewalls and endpoint protection has all but vanished. User and device identity are now the key factors that need to be verified before access is granted to system resources. With remote, work-from-home access increasingly replacing on-premise, inside the LAN communications and a growing variety of cloud, SaaS, and dashboard apps, security risks are on the rise. These security developments are alerting developers of the need to address security holistically, not in a bit-and-piece fashion, and with an emphasis on encryption and multi-factor authentication. Some see this as a key impediment to deploying intelligent systems and are pointing to the need to design from the bottom up a new, so-called Zero Trust security architecture with an eye towards achieving regulatory compliance.
Industry IoT Community Council Survey says:
What is your top concern as your organization undergoes
digital transformation and introduces more intelligent systems?
● Security - 66%
● Data Collection & Analysis - 17%
● Performance - 17%
Data Issues: Successfully realizing the advantages of A New Intelligent Reality depends on the ability to effectively manage massive amounts of data. According to IDC, edge devices will create more than 60,000 petabytes of data by 2023, an amount that is expected to grow six-times faster than data from other systems.(5) And since there’s no storage on the edge, this predicament raises several questions. First off, how to decide what’s reliable and relevant — what’s worth keeping and what to let go?
To determine the data-mining value of a given dataset, data parameters need to be set. What is the organization trying to accomplish, what’s the hypothetical outcome, and how will the premise be validated? Then, if a smart decision about what data to keep can be made, the issue of storage comes to front. A whole new data storage strategy is called for. Does the organization have enough compute on edge devices to analyze the data’s worthiness before deciding to keep it or jettison it? If it’s worth saving, where to — cloud or on-premise?
Algorithm Training: There’s no out-of-the-box perfection. Algorithms need to be fine-tuned frequently to deliver valuable results. A common practice is to take existing algorithms and train them offline in a containerized digital twin for new pattern recognition use cases. Doing so allows developers to develop once and build a library of algorithms that can be trained for deployment in different scenarios. It takes time, but it’s a DevOps process that resonates with many companies looking to add intelligence to their systems.
Skill Gaps: The old guard is moving on and it can be difficult to find people with the right background and skillsets. Fortunately, the academy is working hard to fill skill gaps in modeling and analytical technologies.
What’s down the road?
A few observations from the IoT Industry Community Council on where A New Intelligent Reality is today and where it might be tomorrow:
● Investments in intelligent systems will continue to grow. In the wake of COVID, with continued remote work and increasing demand for healthcare and diagnostic applications, money will flow into intelligent system developments. The current chip shortage will also drive spending in the semiconductor manufacturing sector. Companies seeking a competitive advantage via intelligent systems must first have a solid business strategy and a clear understanding of the potential benefits — and hurdles — before making any investment decisions.
● The demand for better logistical decision-making on the supply chain will spur innovation. Companies cannot abide by cash tied up by inventories that can’t ship. The digital feedback loop is critical for well-informed decision-making.
● AI, intelligent sensors, Big Data, and blockchain will all become interoperable. It’s a journey of alliances and collaborations across industries.
● Every enterprise needs to adopt a cloud-native way of working. Including development and the entire life-cycle of the intelligent system.
● Self-improvement capabilities are critical to the success of systems driven by machine learning; federated AI enables distributing compute power and preserving privacy, which is critical considering use-cases and regulations.
● Finding a smoother path forward; It’s not just about the technology, it’s also a people process. Technology won’t replace people, it will just complement human intelligence. It’s a strive to achieve improvements and increase effectiveness in our daily lives. It’s about our mission and objectives.
(2) Forbes Business Insights, July 2020
(4) Eclipse 2020 IOT Developer
(5) Embedded Intelligence: Innovative Outcomes with Edge Cloud, IDC InfoBrief, May 2020