On the opening day of Industrial Internet of Things (IIoT) platform provider Advantech’s online conference, company representatives and other industry experts gathered to discuss new developments on the horizon for IIoT, artificial intelligence (AI), and industrial networking. In particular, many sessions focused on the hurdles that still remain if IIoT and associated Industry 4.0 technologies are to see ubiquitous adoption in the future.
Perhaps the greatest take-away from the first day of the event was that, while the real bedrock of value provided by IIoT is to be found in the data it generates, nothing can be attained from it unless that data is effectively gathered, communicated, and analyzed. As such, several speakers spotlighted burgeoning technologies such as 5G wireless connectivity, intelligent sensors, and AI as the most consequential industry trends going forward. Through the improvements these technologies enable in data gathering, transmission, and analytics, Advantech envisions industry moving beyond IIoT and toward an Artificial Intelligence of Things (AIoT) that allows cloud-delivered applications to make real-time, autonomous decisions at the device level. Within this framework, cloud-based AI trained on large amounts of data can provide industry operators a means of more easily extracting value from their IIoT infrastructure in exchange for furnishing AIoT platforms with the datasets necessary to continue expanding their capabilities.
Allan Yang, chief technology officer at Advantech, stressed the need for a platform approach if AIoT is to be realized in a timely and cost-effective manner. “AIoT is cross-disciplinary. It requires edge computing, cloud platforms, data know-how, and domain expertise in many specific areas. No one company can do this alone successfully. However, we have seen many companies that are still trying to build their essential technology modules in-house, rather than adopting a platform approach,” he said. “This takes a lot of time and involves a lot of trial and error. We strongly encourage all companies, regardless of their size, to evaluate the possibility of collaborating or engaging in a partnership to speed up adoption.”
The future of IIoT
The Advantech event also explored why IIoT adoption rates have not yet met projected expectations, with Dirk Finstel, deputy managing director at Advantech Europe, noting that although 50 billion IIoT devices were expected to be in operation by 2020, only 8.5 billion have been deployed in reality. According to Finstel, much of this can be attributed to shortcomings in the associated infrastructure needed to make large-scale IIoT a reality. He believes that the high speed and bandwidth capacity of 5G networking will improve the feasibility of many IIoT technologies that rely on cloud computing in the near future.
Advances in edge computing are also expected to play a larger role in IIoT deployments by easing the burden of sending large quantities of data in and of out of plants via cloud computing applications, said Jerry O’Gorman, associate vice president at Advantech North America. Not only does O’Gorman see edge computing reducing costs and accelerating adoption, but by extending cloud-native software to the edge, latency can be reduced and less bandwidth will be required for data transmission. In fact, he estimated that by extending cloud-native software to the edge, up to 75% of data generated may never need to be sent to the cloud.
He also noted how software-as-a-service (SaaS) models are likely to grow in prominence as 5G allows complex applications to be rapidly delivered to the edge. O’Gorman perceives that this could greatly reduce costs for end-users, making increasingly sophisticated AIoT applications easily accessible even to small-and-medium sized enterprises.
Though AI promises to offer impressive new functionalities, end-users shouldn’t expect it to solve all issues surrounding IIoT deployment and integration, said William Webb, author of “The Internet of Things Myth,” during his presentation at the Advantech event.
“There’s a number of promising new developments in this field, but they need to be treated with caution and used in the right way. AI only works when you’ve got the data in the first place, and that means it can only enhance an IIoT system that’s already there and working well,” Webb said. “Until you’ve got an IIoT system in place delivering all of the data, you can’t really use AI to make sense of that data.”
According to Webb, approaching IIoT projects with an eye toward harmoniously adjusting overall business processes may be the best way to ensure success. In numerous early IIoT technology deployments, it was not uncommon for operators to put new systems in place without fully realizing the degree to which they would need to alter their overall operations to efficiently act on insights derived from their data, Webb noted. For example, even when equipment had been outfitted with IIoT technology to allow failures to be predicted in advance, this information could only be used to yield productivity gains once new processes were designed to efficiently allocate labor to maintenance on machines that needed it and redirect it to other valuable activities when they didn’t. So, while predictive maintenance is more efficient in theory, without proper systems support, fixed and regular maintenance schedules are more simplistic and easier to keep to in practice.
Of course, operators are shaking out these kinks, and predictive maintenance is now one of the most common applications for IIoT technology. Still, Webb stressed that it is challenges like these that highlight the importance of viewing IIoT projects not only as technological installations, but initiatives that also require cultural, workforce, and business-oriented changes within an organization.