The Internet of Things is probably on the agenda of all suppliers and users of automation technology. Regardless of whether the term IoT, Industry 4.0 or simply digital transformation is used. And experts in all companies have probably already been working on connecting their own products to IoT or cloud platforms.

From a purely technical perspective, an IoT integration task can often be solved very quickly, as highly developed interfaces, starter kits, seminars and sufficient examples are now available. However, if you search the market for successful (I)IoT solutions that deliver sustainable and valuable user benefits as a result, very few examples can be identified. One reason for this is certainly that the capabilities of an 'XYZ IoT solution version 1.0' are not sufficient to offer potential customers a measurable return on the required additional investment in IoT technology plus the monthly operating costs - such as cloud usage fees and communication costs. Experience shows that very few users are willing to pay a machine manufacturer or control system provider money to be able to view a few sensor readings or operating data in real time at any time using a smartphone app. Conversely, most providers have not yet understood the benefits of simply offering customers such a solution free of charge.

Four subject areas with concrete benefits

Overall, the potential quantifiable benefits of (I)IoT activities in mechanical and plant engineering can be divided into four categories based on the current state of the art, which include both the supplier and the user:

Optimizing the operation of delivered machines and systems: Using sensor data, for example, anomaly detection via machine learning (ML) can be realized. For this purpose, data describing the good condition of the machine or system is recorded during system commissioning. This data can then be used to train a suitable machine learning classification algorithm. In subsequent continuous operation, the real-time data stream is evaluated accordingly by an ML inference engine, for example using binary classification. This allows the following question to be answered: Is the current sensor data in the good range or should it be classified as an anomaly? If an anomaly is detected, both the system operator and the responsible service company receive a notification. If there is secure remote access to the relevant control system, it may be possible to eliminate the causes of an anomaly by remotely updating the setpoint values.

Improve support for the delivered systems: The data received in the cloud could be merged and stored with data from other sources - for example, the latest weather data - as part of a sensor data fusion. The resulting overall data could be used by the system manufacturer itself or an external service provider to develop system optimization strategies that enable the system operator to make quantifiable energy cost savings, for example. The remuneration for such an after-sales service could be based on a percentage of the costs saved.

Improving an existing machine or system: With our PCs and smartphones, we have become accustomed to the fact that the functions of individual applications are constantly being improved via software updates. IT security gaps that are subsequently identified are also closed with the help of updates. In this respect, the internet interface of a component, machine or system (IoT device interface) should definitely also be used for software updates. Of course, the corresponding safety requirements (safety, IT security) must be observed.

Develop even better machines with the knowledge gained from operation: Sensor data acquisition creates a fairly detailed collection of data on the use of tools, machines and systems, which, with the right data quality, can provide very deep insights into user behavior and environmental conditions. If this data and the knowledge it contains is fed back to the responsible development or design department with the help of a feedback loop, the next generation of machines and systems can usually be significantly improved. Furthermore, the machine and system manufacturer often has access to data that can be very valuable in the guarantee and warranty phase.

Why do IoT projects fail?

Security alone requires a public key infrastructure (PKI) at the current state of the art in order to at least guarantee the authenticity and integrity of the update.

© SSV Software Systems

Among IoT marketing experts, a missing or flawed business model is currently the most common reason for the failure of an IoT product. In many cases, the use case and the criteria that are actually relevant to success are not defined or worked out precisely enough before the start of development - in addition to a viable business model, sometimes the problem to be solved is actually missing! In order to avoid the known mistakes, a multidisciplinary team of experts is required in any case, but also the appropriate methods to manage such a team.

If, for example, a machine or system is to be permanently improved by automatic software updates via the internet, it is technically sufficient for the pragmatic first solution approach to transfer a file with new software code or configuration data to a cloud storage service. The relevant control units can check there from time to time and - if available - download and install the update from the cloud. In addition to this easy-to-implement cloud-based file transfer, however, there are other aspects in the areas of IT security, machine and plant safety as well as one or two legal issues. You therefore not only need a software developer for connectivity and file transfer, but also experts in IT security, cryptography, plant security (a process failure mode and effects analysis is useful here) plus an experienced lawyer for the team of specialists. With the current state of technology, security alone requires a public key infrastructure (PKI) for digital signatures with private and public keys, certificates and revocation lists in order to at least guarantee the authenticity and integrity of the update. In many cases, encryption is also added to ensure confidentiality outside of TLS/SSL connections. This alone requires a great deal of expert knowledge.

The development of IoT-based services is much more complicated - after-sales services would still be the simplest option here. IoT service development through to a completely new business model is currently the 'supreme discipline' of the industry. Many companies, including in automation, have already launched projects in this area and - as Airbnb, Facebook, Google and Uber prove - have completely changed entire industries. But many have also failed. There is currently a high failure rate in this segment. In the case of success, above-average benefits for users and providers are possible. A component or machine manufacturer could theoretically even position itself with an ingenious (I)IoT service in a market environment without serious competition. However, however carefully put together a team of experts, which should also include customers, is probably not enough for such a successful service development. It will also require special methods and additional expertise.

Author:
Klaus-Dieter Walter is a member of the management board at SSV Software Systems.