IoT Tech Expo Europe 2018 – Event Highlights

Nikita and Yuliya at IoT Tech Expo Europe 2018

Yuliya:

This year IoT Tech Expo co-hosted blockchain and AI sections, and half of all content was about blockchain. Other topics covered data analytics for AI and IoT, developing for the IoT, connected industry, smart energy and cities, connected transportation and more. Both of the event days have incorporated the IoT, blockchain and AI agendas.

The event attracted 8000 attendees from all over Europe, including more than 300 exhibitors and the same number of speakers. They shared their experience on how they apply the above mentioned technologies. We listened to the speakers, visited other companies’ stands, did some networking, and now can share what we have taken from the event.

Nikita:

We have paid the most attention to the exhibitors and speakers whose thoughts and projects could help in our current work. In terms of several of our projects, we use out-of-the-box Kontakt.io BLE devices. In case you have access to the Internet it is not a problem, you may manage all your BLE devices via the Kontakt.io dashboard. But, if you want to implement a solution with BLE devices in a closed private network without Internet access, you need a solution that will manage all your devices. In this case, an Alcadis solution may be helpful. It allows to organize and collect all your BLE devices and control them. Thanks to the Alcadis IoT controller, IoT endpoints can be managed, coordinated and connected to IoT cloud services as a part of a single, converged IoT access network.

Another example – when we run an IoT project with a thorough description and requirements, sometimes we have no hardware to create soft for, so we need to wait for it. With the help of the solutions I have described, we can start the development soon after we have discussed the requirements – read about Xinabox.cc, for example. No time lost on waiting for the hardware to be delivered.

Part of our projects contains custom mathematical calculations and logic. During project development, the logic may be extended or modified by customer requests. In this case, the final result will be different from the initial one. And we need to be sure every change and modification affects the particular part of the application only, as planned.

We found the solution that helps us to define mathematical logic in an easy way and get real-time calculations – Streamsheets by Cedalo. In this case, we may define the initial project calculation logic, change it anytime in parallel with the application changes, and compare the results. Also, the solution may be configured as a part of continuous integration with automated tests and control layers. It will save time for developers and QAs and make our solutions qualitative and more controlled.

Yuliya:

Another interesting project was presented by a food company HelloFresh – they deliver home cooking kits with detailed recipes. In the speech, the company’s Head of Data Science Adrian Foltyn showed how they analyze their business. They decomposed the meals into single ingredients and analyzed them via neural networks. The results were such insights as shares of recipes during a particular week, that show the user demand. The company can benefit from such analysis and create new meal recipes that their customers will totally like, and gain other insights that can’t be seen with the naked eye. The bottom line is that such technological approaches work even for companies you do not expect to see in such a list.

HelloFresh presentation on recipe analysis

Nikita:

Speaking about statistics once again, developers sometimes face the following problem. When the project is finished and the solution is delivered to the customer, it is important to track how everything works. There is a huge difference between these two issues: when the solution includes 100,000 devices, and 10% of them do not work at all, is different from 100,000 devices having 90% of functionality working properly. The statistical data for both of them will look similar, this is why it is important to examine the solution in detail. The data developers get need to be thoroughly analyzed, and when it is clear that 10% of devices do not work at all it is easier to fix the issue – the solution is more likely to be unambiguous. Here the team can determine that you need to take 2-4 extra actions to solve the problem, but in the other case, there are number of options. The developers and QA specialists need to spend more time deciding on what action to take.

Yuliya:

In terms of the topic Data Analytics for AI and IoT, the representative of AkzoNobel Pascha Iljin brought a case study “Predictive analytics within a corporate world: Black box or transparent box?” He described the idea of deep-dive asking “Why?” five times before analyzing data. He suggested that asking such questions as what is the purpose I gather data for, what insights I want to get, establish the concept of what you need to analyze the data for. In particular situations, you will reach the conclusion that it is better to use small data models. When you collect data it is important to understand that you can not collect every piece of it. Storing huge piles of data is hard, so determine the parameters you want to learn, build a small model, and determine what is the outcome of the analysis.

The trick is that AkzoNobel is a company that produces paints and performance coatings – what could this company have to do with predictive maintenance? One of the company’s customers are businesses that rent vehicles. AkzoNobel helped them to determine the trends in car accidents, and when the car is damaged it needs repair, including a new painting. They have evaluated the car accident probability and thanks to this collaboration they have sold more paint. For the proper evaluation, they have selected 15 parameters that commonly affect accidents, such as new cars bought, which probably means more unpracticed drivers on the roads, holidays, and so on.

Pascha Iljin from AkzoNobel with a case study “Predictive analytics within a corporate world: Black box or transparent box?”

Pascha also quoted Jim Barksdale, the former Netscape CEO. Jim said, “If you have data, let’s look at it, but if you have an opinion, I prefer mine”. The deep-dive asking “Why?” approach and this sentence about data were quoted throughout the day by other speakers. The idea is simple but clear – analyze the data if you want to understand what you do and whether you do it right. As you can see, even a food delivery company uses neural networks to analyze its business performance. Breaking the business down into a number of components and analyzing them leads to an infinite number of conclusions, and businesses of any type can do so.

Nikita:

We have seen a stand where IoTify.io presented their platform. It is capable of simulating the entire IoT network: connectivity, a gateway, hardware elements. You can configure literally everything in the network without any device in possession: operation time, data transmission and sync rules, and so on. Using such a solution we can start the development process quickly, have the possibility of flexible configuration, amend the network parameters. We do not have to wait for the hardware to be shipped from the customer, we can deliver a working solution that just needs to be tested on real hardware on the client’s side.

One more interesting participant is Microsoft Azure Cloud, it is a cloud solution that allows the gateway to store a part of data and architecture logic. This solution helps to eliminate the risks of the lost Internet connection or synchronization interruption. With the help of Azure, we can build local IoT networks with a part of its logic in the gateway. It is a perfect approach for projects where the connection to the Internet is unavailable. Sometimes the solutions we use, such as iBeacon, demand a direct Internet connection and we need to get around such a requirement, and Azure helps. This is similar to edge computing and is very convenient for a significant number of our current projects.

Yuliya:

Now there is a trend called industry 4.0 – whereas previously only 33% of manufacturers had a high level of digitization, in 5 years we will see more innovative producers and 72% of them will be highly digitized. Manufacturers realize that it is not enough to just produce hardware, they need to make it smart and provide additional services to keep ahead of the competition. The speech was brought by an ICT representative John Koot, who also noted that predictive maintenance based on data analysis is beneficial.

ICT presentation on predictive maintenance

To sum up, it was a great event that gathered bright minds from all over the industry and various locations. Thanks to knowledge sharing and networking we brought a number of fresh ideas and valuable connections from the expo. See you at IoT Tech Expo next year!