An interview with Thomas Kallstenius, Director, Research & Innovation Strategy at iMinds.
Thomas Kallstenius: Today about 15 billion objects are connected. That means about 2 objects per individual on our planet. It’s predicted that this number will rise to 50 billion objects by 2025, with a ratio of 6 connected devices per person. That’s when the industry expansion will truly happen.
The challenge will be to keep all of the connected devices up to date and secured. Because the smaller and less sophisticated they are, the easier it is to hack them. But adding security functionalities normally means you will need more memory, and more battery. Just imagine what encrypting a pacemaker will do to its battery: an average pacemaker would run out after a year. That’s simply not an option. So we will need to find solutions like lightweight encryption. Power consumption is an important IoT issue that tends to be overlooked a lot.
Which industry will feel the most impact of the IoT?
Thomas Kallstenius: Manufacturing will certainly be one of the most prominent early adopters. There are several reasons for this. To begin with, most machines are not connected today. Just linking them will increase flexibility exponentially and open up entirely new business models, like using connected devices as a service. But the step before all that is obviously that you need to connect them first. Leveraging their knowledge can only come after that.
Secondly, the decision-making processes in manufacturing tend to be rather straightforward, so this will stimulate adoption as well. Typically, there is only one decision maker who makes the choice to implement IoT. Smart city applications have as much potential but the complex decision environment of governments can in some cases slow adoption down.
Last, but not least, data security and privacy are some of the biggest hurdles towards IoT adoption in every industry and that is no different in manufacturing. The difference is that manufacturing deals more with machines than with human beings, with somewhat fewer consequences for privacy. So it is less of a hurdle there.
Which new industries will be created by the IoT?
Thomas Kallstenius: There is a huge market potential for any type of automation of work. I’m curious to find out what will happen when we start combining IoT with artificial intelligence and deep learning. This, is of course, all very futuristic, but it’s something which we should start thinking about now because we will see some very interesting things develop in that area in the coming years.
How will ‘everything connected’ affect cyber security? It almost seems scary when you think of what might be possible.
Thomas Kallstenius: That’s a sound reaction. We should be cautious. That’s why IoT security is one of the focus areas of iMinds. We believe that it is good practice in security to store and process data as much as possible locally, instead of sending everything to the cloud, though this obviously clashes with the economic point of view. You have to find a tradeoff between technology and security economics. At iMinds, we are investigating how we can efficiently send essential information to the cloud while keeping the confidential information local.
On top of that, there are interesting new security solutions for the cloud, like homomorphic encryption which allows one to perform calculations and searches on encrypted information without decrypting it first. There is also differential privacy which allows the accuracy of queries from statistical databases to be maximized while, at the same time, minimizing the odds of identifying records. At iMinds we are actively looking at these new kinds of technology that could really unlock this issue.
Which are the main obstacles for IoT adoption, according to you?
Thomas Kallstenius: Apart from security and privacy, interoperability is a major obstacle: the ability of things to talk to other things. It is not so much a matter of introducing new standards, as some suggest. There are probably too many competing standards already today, which actually defeats their purpose. You just need to find those standards which are already widely adopted. This interoperability is not a new problem. We used to have the same issue with the internet, as you may recall. There are lessons to be learned from that, seeing that the IoT is actually an extension of the internet.
iMinds is fundamentally in favor of interoperability and open platforms. We want to avoid the situation where customers are locked into certain vendors. Open source software is also preferable from a security point of view since it allows auditing the software any time by any independent party. Auditing of proprietary software, in contrast, requires the owner of the software to allow access to the source code.
IoT, just like any other technological innovation, changes the work environment. What do you think will be the most important job skills of the future?
Thomas Kallstenius: Though there is still a lot that humans do that machines can’t, it remains a fact that some functions are quite easy to automate, even knowledge jobs if they are characterized by a high percentage of routine. For these types of employment, we should think about alternatives in the future. Though some see this as a threat, I prefer to perceive this positively: history tells us that technology does not take jobs away, it creates better jobs in the long run.
Not so long ago, it was possible for people to have the same job during their entire career but this is no longer an option. Lifelong learning has a crucial role to play in this, continuously educating people to adopt new technologies and capabilities. That is why one of the key future job skills will be adaptability.
Predictive analytics will be one of the most important drivers of the IoT. What will be the biggest challenges for organizations in that area?
Thomas Kallstenius: It depends on which kind of predictive analytics we are talking about: one involves a massive amount of data while, in the other, data is scarce and expensive. When it comes to Big Data, the greatest challenges consist of picking the right data to analyze and making sure that the different kinds of formats are interoperable. But with small data – when you only have a few useful data points – you have to be absolutely sure of the quality of your data and use it as wisely as you can. That’s why you need better meta-models or surrogate-based models based on machine learning.
IoT is not just about Big Data. Sometimes you are dealing with just a few readings from a sensor somewhere and you need to use them wisely.