Evaluate Weigh the pros and cons of technologies, products and projects you are considering.

A look back on eight IoT predictions

Back in early 2015, I made a series of predictions about the trajectory of IoT at my company’s user’s conference. The talk took place five years ago, but things move fast in IoT and so I looked back to see how I fared with my predictions.

1. An explosion of devices

Prediction: 50 billion connected devices by 2020

Reality: Technically wrong, but right in the larger scheme of things. The 50 billion devices prediction began in the early part of the decade, promoted by Cisco and Ericsson. Cisco later upped their forecast to 500 billion by 2020. Gartner predicted 20.4 billion, 451 Research 8.8 billion and IDC 41 billion, but these figures didn’t include PCs, IT equipment and phones. However, 451 Research did place a stricter definition.

The predictions were a bit optimistic, but here’s the interesting part: The growth rate is faster than expected. In 2016, the world had 17.6 billion IoT devices and would have 30.7 billion by 2020, according to IHS Markit. In 2018, IoT devices were grazing the 31 billion mark, according to IHS said. The forecast for 2030? 125 billion.

If you look at some successful categories, such as WiFi, smartphones and LCD monitors, you see a pattern of forecasts being regularly upgraded. Thus, the growth rate points to positive outcomes, though I am now questioning the real value drivers behind this uptake.

2. Data variety will blossom

Prediction: We’ll see greater variety of data types, more complex data types and unusual combinations of data. For example, sensor or device specific context, such as GPS and location data, will be combined with different sources to produce unexpected insights.

Reality: Yes, there is a proliferation of new cost-effective sensing technologies and sensors that are leveraging a broader range of human senses, such as acoustic, vision and smell. These technologies are also sensing phenomena way outside the bounds of human capability, driving incredible improvements in physical and situational awareness.

But what’s equally interesting is the unusual combinations and use cases. For example, insurance companies have begun to experiment with IoT devices to monitor the health of pipes as a way to reduce water-related claims. A mining company in Canada, Syncrude, uncovered a driver safety problem while mining oil temperature data.

What’s the fastest growing market in IoT? Fast Food, said Charlie Wu, Product Manager at Advantech.

3. Big data will get bigger

Prediction: Cheap sensors and proliferating techniques for extracting value of data will mean that the conventional idea of a refinery having 100,000 or even 500,000 sensors gets “blown away”.

Reality: Correct, but I missed a bigger issue. The total amount of data continues to double every two years. Real-time data has also become one of the fastest growing categories of information and will double in its demographic heft to 30% of the data population by 2025, according to IDC.

Here’s the twist: In industry, we’re not seeing new sensors as much as companies taking advantage of data they already collect. For example, a graduate student working for Lonza — which makes specialty ingredients for food and pharmaceutical companies — figured out a way to potentially increase capacity by 15 to 20% using data Lonza already had.

4. Devices get dynamic

Prediction: Companies will shift from fixed-location and fixed-function sensors, shifting instead to dynamic sensors and devices that can be configured for different applications and locations.

Reality: I was expecting a move from specialty appliances and sensors to agile and dynamic solutions, equivalent to the “Software Defined X” phenomena, just as I am expecting this to take root in the processes and operations of traditional manufacturing. Efforts we see in the Open Process Automation hint at this agile approach.

The dominant paradigm is still mirrored in companies such as Petasense, who have successfully brought devices like this to market. The declining cost of hardware might also result in fewer dynamic devices and more rip-and-replace devices. For example, EDJX is coming out with nano servers made from inexpensive or used parts that can be run to failure.

I am now more conservative when I expect to see this phenomenon reach a critical impact. Which paradigm wins remains to be seen, but I don’t expect it to be either or.

5. Protocols gone wild.

Predictions: Vendors will develop their own protocols to meet their specific needs, whether driven by the nature of the information, the overall architecture or evolutions in technology that are cutting edge. Unfortunately, that will create confusion and incompatibility.

Reality: Unfortunately, this is true. While industry standards such as OPC UA have achieved broad adoption, each vendor tends to implement the standard differently, resulting in incompatibilities. If history is any guide, many of these problems will work themselves out as time goes by. In contrast, networking is seeing the evolution of different standards for different tasks. For example, 5G devices that require constant, fast and high-volume bandwidth with LoRa targeting applications that generate less and less urgent data. Over time, expect a few to emerge as favorites for particular tasks and use case scenarios.

6. Community and data sharing.

Prediction: Data sharing between organizations present tremendous opportunities for efficiencies, but questions about security and privacy will invariably make acceptance gradual.

Reality: True, but there’s more to it; some sharing is occurring. For example, Eli Lilly monitors contract manufacturers in real time to help ensure its product quality. Similarly, YPF uses it to monitor its third-party wind providers and their performance.

We see significant advances in supply chains. And there are massive needs for community and data sharing if we are going to solve the future energy generation and distribution challenges as renewables such as distributed energy and microgrids continue to disrupt the centralized approach to energy. But still, widespread communities are not here yet.

One big thing I missed here was blockchain, which could accelerate communities by enabling decentralized solutions.

7. Computing will become geo diverse

Prediction: While a building or a factory is rooted in one spot, the computing resources to better optimize them could be anywhere, and increasingly will be exploited by a wider variety of people. Public clouds will be necessary to efficiently manage this intelligence.

Reality: Partly right. Public clouds play a critical role in helping companies analyze large data sets or handling certain types of applications. But it’s becoming increasingly clear that a significant percentage of compute tasks and data storage will continue to take place locally, either inside facilities or devices. Concerns about latency, bandwidth availability and bandwidth cost are driving a shift to hybrid computing architectures.

To put it another way, we regularly swing from eras of centralized to distributed computing architectures. IoT is pushing the pendulum toward distributed.

8. Data silos will raise their ugly head

Prediction: Controlling every aspect of an IoT application, such as device analytics and data, might be good for vendors, but it’s terrible for customers.

Reality: Luckily, it’s not happening as much it could have as consolidation and exits in this market have helped. In addition, the approach is evolving: IoT platforms will increasingly not be offered as monolithic services. Rather, companies will focus on a few core areas of expertise, such as analytics, device management and security, and customers will weave together platforms from these offerings.

Final thoughts

I think I generally predicted the direction of the market. What’s going to be most interesting is to watch the development of data communities. Technology can be easy to predict because hardware gets better and cheaper while algorithms become more accurate. Human behavior and cooperation are more of a wild card.

All IoT Agenda network contributors are responsible for the content and accuracy of their posts. Opinions are of the writers and do not necessarily convey the thoughts of IoT Agenda.

Data Center
Data Management