The internet of things (IoT) has become such a buzzword and so over-used that it feels like it might be way too complicated to deploy. But in simple terms, it’s about sensing and measuring parameters specific to a process, doing something with the data, and then providing a feedback and control mechanism to enable corrective action.
This may sound simple enough. However, many end-users are not so well versed in systems design, networks, data analytics and security, so they often have little idea where to start. Chip manufacturers as well as embedded boards and modules providers have recognized this, and have bene building their portfolios of ready-made ideas which can be implemented or deployed quickly.
One company providing embedded modules and solutions for the IoT is Advantech, so we caught up with Jash Bansidhar, managing director of Advantech Europe, to understand its’ approach for providing solution-ready products. Central to the company’s strategy is what it calls “co-creation”, which involves working with independent systems integrators to provide application-ready system-level solutions for IoT implementation in industrial automation, healthcare and smart cities.
Bansidhar emphasized the need for presenting pre-packaged ideas to end-users, and the need to do so with partners who have domain-specific knowledge for the many different potential applications of IoT. He remarked that decision-makers for industrial automation applications or smart cities are not going to implement IoT based on some ideas or PowerPoint slides, and that they are less interested in the actual technology stacks below it. “They are interested in knowing how it will look in their environment. So, proof of concepts can only be made once we understand the domain we are working in and the pain points of that domain. And then you need partnerships with ecosystem players who are working in those domains.”
Hence the company’s co-creation strategy aims to address this domain specific expertise by collaborating with system integrators and solution partners in both information technology (IT) and operational technology (OT) to build proofs of concept.
Addressing the trade-offs between edge and cloud
A key area of debate in IoT is around edge intelligence, so what are trade-offs users need to think about in the context of getting the right balance between power consumption, performance, latency and even security?
Bansidhar explained, “Today, IoT is all about data acquisition (transforming physical information into data), communication, computing and the software gluing all those things together. Computing can be done on different levels: on the control level, on the edge level, in the cloud level (on a private cloud or global cloud). All decisions are made in different layers of computing. With our huge portfolio of hardware and software stacks running on the hardware, we can cover from data acquisition to the edge and also the cloud.”
“In the cloud, there are many major players who have the cloud infrastructure already. We are talking about Microsoft Azure or Amazon Web Services. Big cloud systems which are used to, let’s say, make your dashboards on a country scale or a global scale. That is their strength. But the problem is that the cost for having those cloud services are very high once you started to communicate data from the edge and bringing back. So, the necessity to make IoT cost effective was to make decisions on the edge.”
This is now possible because of a portfolio of boards and modules to address different performance levels. “Once you run the edge algorithms, you need high performance and the Nvidia Jetson CPU or Microsoft-Intel CPUs are able to provide that.”
Covid-19 accelerated adoption of AI platforms for medical
Bansidhar said that he didn’t expect such rapid adoption of AI platforms for IoT, but Covid-19 has certainly accelerated the use of AI in medical applications. “We now have applications running where AI algorithms running on our computing platforms are able to visualize potential Covid-19 contamination in real-time in a patient entering a space. Without having this technology, it would have taken a lot of testing, such as taking blood samples and so on. But now, it’s just done in real-time, based on the availability of high-performance computing at a reasonable price – because that is also of course important to be able to scale the implementation of AI algorithms.”
This is where a bundled solution comprising the technology stack and solutions brought in from specialist systems integrators with domain expertise and AI algorithms come in, as part of the company’s co-creation strategy, to deliver a very quick proof-of-concept. The benefit, Bansidhar said, is that hospitals can then decide very quickly to implement the solution. “With these solutions, decisions can be made locally, and information shared securely and locally in real-time. But if you want to run 40 hospitals, or 40 factories across the globe, or in a country like the UK, that’s where you also need cloud services.”
He added, “A combination of the knowledge of computing technology, low power consumption CPUs with real-time capability, and with communication protocols like OPC UA, you are able to run a real-time connection platform connecting all the devices necessary to process the KPIs [key performance indicators] within the algorithms in real-time. All these competencies come together in a platform from a supplier like Advantech, including the AI algorithms, because of our co-creation strategy.”
“I think that’s a unique proposition where we say, if you want to speed up, if you want to be flexible and agile – and what we have also learned from Covid-19 that companies who are not able to adjust quickly, will be in trouble – then as a result of the fact that we were already along this path of co-creation and had these platforms, we were able to make this Covid-19 AI IoT application in only a few weeks.”
Verticalization in a domain is vital
“We have similar examples in manufacturing, in transportation onboard trains on rolling stock, and we will see more and more in other vertical applications. One thing is clear and that verticalization in a domain that you want to be the best in, is crucial, otherwise you will stay a generalist and AI is really a domain focused orientation.”
“You cannot create general AI algorithms to take care of everything in each application. There are some common things you can reuse, like basic algorithms, but you need to have specific KPI calculations in the vertical domain. You cannot be specialist in everything and that’s why we have these three sectors: industry, embedded and smart cities. With these three sectors we can offer a lot of intelligent applications with their own ecosystem and with their own specific hardware. That’s because you can imagine that in factories you need different communication protocols than in smart grids or rolling stock, because the certifications are also different, the environment is different. So, having platforms with vertical AI algorithms and the communications stacks, I think that is really the future.”
Security depends on the market vertical
Security is always a big topic in IoT, so what about this critical aspect of the system? Bansidhar said, “Of course, we have the standard security connections that are being used like VPN and SDL type communication. But that is not enough. If you look to how companies are preparing for their security platforms, they are still fairly uncertain about placing critical data outside their firewall. It’s important you can give the customer solutions before the firewall, which means on premises. That is where edge comes in, because you can run a lot of decisions before your firewall on the edge, and only release critical data if you want to go outside the firewall into the cloud.”
“It depends on the vertical market and the application that the customer has. Our own software has all the security features and we also have a cloud platform which is communicating into the edge with all the security stacks available. There are different ways to secure your connection. You can, for example, have databases in between creating authentication possibilities first before you can go to another layer. You can have the standard McAfee security stacks. But we also bring the competency of companies already working for decades in security with communication protocols to add that as part of our security stack.”
“You can imagine that companies who are printing money have a different security requirement to companies managing water installations or automotive. If you want to bring in vertical, end-user ready solutions, it is natural for a company like Advantech to work with solution integrators with domain specific core competence in transforming processes into software, processes that will bring different revenue streams. However, these companies don’t have the competence of 35 years of manufacturing and the product technology stack. We have more than 10,000 SKUs in our portfolio.”
Creating vertical IoT proof of concept that is 70-80% ready
“With the diversity of our portfolio, if we are able to bundle these into applications like energy management effectiveness, for example, or equipment efficiency for machines or factories, we can create instant out-of-the-box solutions for the end-user. Our idea is to have a proof of concept which is 70% to 80% ready for the end user to start visualizing data, for different scenarios, because each factory or each smart city has different needs, but with a common basic foundation.”
“Giving them [the end-user] this part number as a bundle which contains the data acquisition products, communication products, edge computing products and edge analytics software, plus communication to the cloud, allows you to show the customer you can implement this application in just one day. This can give an instant view of what benefits it can bring, without the need to spend a lot of resources and time to do the integration, which is the traditional way of doing automation.”
One of the reasons Bansidhar emphasized the delivery of a bundled package is that an end-user doesn’t often know what he or she wants; hence the pre-packaged solutions ideas showing how their operations could be optimized using IoT technology. The company offers a number of solution-ready products (SRPs), he said. “If you look at the Advantech web site, you will find dozens of SRPs today because of our strategy launched some years ago, to team up with domain-focused software integrators and to co-create these SRPs so they trigger ideas for scenarios.”
“Three years ago, IoT was really like a buzzword. Now we can show it is happening with many applications already. So, we were able to find these domain-focused system integrators across the globe.”
“The last point I want to make is the global footprint of Advantech helps these co-creation partners, these system integrators, to globalize. That’s because innovation is often done by startups and scaleups, so these companies are looking for a platform to exploit opportunity, not only in their city or the country, but across Europe or across the globe. From this pool of startups and scaleups, if you are able to create sustainable solutions with this co-creation approach, then you have the benefit that your hardware is tested and standardized, with the specific software and installation of the solution-ready package. So it’s a worry free IoT application.”