Three Challenges in Industrial Analytics

We described the growth potential of IIoT market in a previous post. In that same post, we also suggested a framework to grasp Industrial Analytics. In this article we will use that framework to describe the most critical challenges in industrial analytics.

Industrial customers have stringent requirements such as the need for reliable security and control, unfailing connectivity despite harsh environments, sub-millisecond response time and ability to operate without humans. Further, billions of industrial devices were built to operate for decades so are prohibitively expensive to replace. These are just some of the many challenges faced by IIoT solution providers. We interviewed hundreds of industry experts and determined the 3 most critical challenges in industrial analytics.

I. Inadequate infrastructure to aggregate data from disparate sources

Industrial machines generate massive amounts of streaming data. Most of this data is dissimilar and unstructured, without any standards in formatting or versioning. This is problematic because current data infrastructure products were designed to process human-generated, standardized data — low in velocity and variability. Further, most IIoT analytics applications require both historical and real-time data to contextualize and normalize the data, before generating insights or highlighting anomalies. Again, the current data warehouses are unable to aggregate data from disparate sources. As a result, only the most critical and expensive machines are monitored, and only by a handful of companies that can afford expensive solutions. There is an opportunity to build data integration platforms that can clean, aggregate and normalize both real-time data streams and legacy data. For example, such a solution would allow maintenance crew in the aviation industry to reduce air travel delays by determining optimal maintenance schedules from correlations of engine performance and historical engine performance data

II. Strict network constraints

Industrial analytics require an intense, two-way flow of information in order to manage a huge conglomerate of producers and consumers of data. This can consume a lot of bandwidth and decisions should be made as close as possible to the source of data in order to reduce this cost.

There are two potential approaches to address this challenge.

a. Technologies that allow processing the data close to the generation of this data (‘edge devices’) that can either respond via pre-planned actions or send an alert. For example, oil well site valves that transmit only vital data like instantaneous high and low pressure will greatly reduce the bandwidth requirements. This would allow for actions to be quick and autonomous, and will lower the communication costs.

b. Industrial operations that deploy mobile devices can improve overall responsiveness. As an example, responses by remote expert workers for various alerts can be recorded via their mobile devices, and can be subsequently harnessed for autonomous and immediate responses by an industrial analytics system. Such a solution would significantly reduce delays caused due to alert fatigue*.

III. Security of IIoT Data

The massive, yet highly critical IIoT data needs protection from hackers, device breaches, DOS attacks and viruses. For example, the need for security on the IIoT smart utility grid increases as the grid coverage increases because of the extent of damage a single security breach can inflict. Currently, on the event of a data breach, the measures taken by public utilities commissions are outdated and sluggish. Solutions that can provide impermeable security of IIoT data while helping regulators to adopt pertinent standards to keep up with the advances technology can have a significant impact on the whole utilities industry.

Our research suggests that developing scalable solutions to aggregate IIoT data, learning from this data as close as possible to source, applying these learnings via a collaborative mobile platform, and keeping it all secure is critical to advancing IIoT.

Foot Notes:

* Alarm fatigue or alert fatigue occurs when one is exposed to a large number of frequent alerts and consequently becomes desensitized to them. Desensitization can lead to longer response times or to missing important alarms.