Automation in manufacturing means:
data, lots of data and even more data!
A lot of car manufacturing is automated. Every machine provides status updates on its progress in production. There are all kinds of sensors checking whether quality is up to scratch and if the corresponding progress has been correctly updated by the system. Therefore, there is a lot of data being recorded and processed. If no tyres are mounted at the end of the construction phase, it should have already been recognised by the system.
Thus, there are huge amounts of data involved here. If this considerable amount of data needs to be transferred to the cloud, then it first has to be uploaded, then processed and analysed there, and the result then downloaded again. That not only requires Internet access, but also sufficient bandwidth. It may also be the case that video data needs to be analysed, to see if four tyres really have been mounted, for example. Now imagine that all the data, as in every bit of it, is uploaded.
- Connected cars that access navigation, communication and other personal services via the Internet. In future, this application will become the prerequisite for autonomous vehicles.
- Smart-grid electricity networks with decentralised monitoring and remote management can considerably improve efficiency, reduce loss and improve reliability. That reduces consumption and saves money.
- Smart cities whose services inherently connect people, places and things via complex networks that are sophisticatedly decentralised. The areas of application are highly diverse.
Time is an important factor
The entire production line would be at a standstill until the data had been uploaded and analysed. That means: in reality, that all has to happen fast. That is why edge computing would be a good choice in this case. Of course, a mix of edge and cloud would also be an option, whereby data would be prepared for transfer to cloud intelligence that would then decide on the basis of all the data whether the wheel has been properly mounted.
What does that mean? The later the data is evaluated and information is converted, the more expensive it gets, as production is at a standstill for that period.
And once we are talking about multiple sets of video data to be processed, we soon come up against technical limitations. With this in mind, systems are put into operation that process the data there directly.
Thus, companies need an edge platform to adjust the digitalisation process to the technological conditions.
Acquiring an edge platform for such purposes is not simply a case of putting it on the shopping list for the electronics supermarket around the corner and inserting it like a battery in a torch. The sensitive process of weighing up what solution is suitable and when requires advice from experts. It can’t be about simply saying yes or no to the cloud. A cloud has limits, and companies will automate and digitise more and more processes in the course of digitalisation. For many companies, integrating edge platforms to support such endeavours will be unavoidable. Edge IoT platforms are platforms that effectively perform a hybrid function. Data is collected and analysed where it arises. Due to the amount of data, it is processed first so that only the most important data is transferred to a cloud service, where the decisions are then made. Administrating and updating the corresponding systems centrally is recommended.
Does everyone now need an edge IoT platform? The answer is a resounding maybe.
As part of digitalisation, it is necessary to work out how much data will arise and where that will happen. Many companies will then find themselves faced by the challenge that they potentially collect more data than they can transfer to the cloud.
They therefore need support in the form of a platform that meets their demand and displays the processes as required. Whether the processing then takes place in the cloud or where the data is generated, or perhaps both – i.e. hybrid – is down to highly diverse factors. Companies will then select the provider who offers them a platform that can provide them with flexibility now and in the future. Let’s look again at the car manufacturing plant: the production location could change, factories may be merged or a new factory may be opened elsewhere. If this cannot be implemented with existing technologies, that means higher costs and, of course, greater complexity. The production processes usually stay the same, after all, and the technologies should do so, too. However, it is also possible that other conditions exist that require a different approach.
As an entrepreneur, you therefore need a platform that allows you to process data as you require. As the customer of the provider of such a platform, you want to make use of a selection of additional functions. We are all familiar with this from our smartphones. In that case, we simply download the additional functions from an app store. Shouldn’t a business IoT platform work in a similar way? That is a case of distributing as many applications or additional functions across different devices and software systems as efficiently as possible. The benefit for the customer is very clear: the same mechanism is used across a broad range of applications or devices.
The nightmare scenario: production standstill
In the business environment it would be fatal to simply update an application. That update could bring the digitalisation process chain to a standstill because a device is out of operation while the update is ongoing.
Using the example of our car manufacturing plant, the entire production line would stop if a device were currently out of action due to an update. Companies therefore need to be able to plan and schedule updates for all their IoT devices and applications. A business IoT platform is thus device- and software-independent.
If the car then rolls down the road as a driverless vehicle, we have a textbook example of edge devices. A driverless car is a data centre on wheels. It is essential that it can also function without cloud support. Why?
Usability is the number-one topic for SMEs when it comes to digitalisation, because IT can become very complex very quickly. #jobwizards https://km.social/35KkqoL
An autonomous vehicle that relied on cloud data would ignore traffic lights and cause accidents
The latency period for the data transfer is simply too long. Edge computing is therefore essential to process the data on-site. A centralised cloud is also not a smart solution in many machine-learning scenarios, as direct processing for quick decision making is needed.
Data processing can however take place at different locations, for example a central server in a networked production hall. Detached stations, such as the water pump of a freshwater reservoir, meanwhile, do not usually have this option. However, since a pump like that still needs a communication connection, it makes sense to place the processing power right there.
Edge computing becomes dangerous when an excess of data needs to be processed or stored or a scalable solution is required. Aside from the initial investment, the running costs and effort can get out of hand if appropriate dimensioning was not carried out in advance or the computing and storage requirements are very irregular.
For SMEs, usability is a particularly important criterion
IT can become very complex very quickly. An IoT platform attempts to simply process all that complexity. At the touch of a button, you can perform updates, install software, add functionalities. SMEs in particular can benefit from putting their faith in IoT platforms. This gives them a completely secured system, so they do not have to worry about how authentication on device A or B works or how to most efficiently ensure who has access to what. Unfortunately, in many countries the Internet connection is problematic, with the Scandinavian countries leading the way as a positive example.
The sensible balance
In summary: edge computing is the relocation of computing power, applications, data and services directly to the logical ‘edge’ of a network. A system like this must be easy to install, use and manage and be remotely accessible, as there is usually no specialist IT personnel available on-site.
The company network is at far lower security risk in terms of manipulation or data theft in comparison to purely cloud computing.
No SME will be able to avoid a decision on when to use edge computing and when to use cloud computing
Fundamentally, the application is decisive for the selection of the location and the scope of the data processing or storage. There are processes that must take place on-site, and some for which a wide variety of data from different sources, some distributed locally, is required. And for some processes, both variants work. The decision therefore depends on the possibilities offered by edge and cloud computing in each case.
Weigh up priorities!
In the same way that the cloud has developed in comparison to on-premises storage, the hybrid approach will also influence the development of IoT infrastructures.
The criteria for the selection of the processing location are as diverse as the applications. Security provisions, real-time requirements, bandwidth, processing capacity, access to cognitive functions – the weighting of the priorities determines what data is processed on what level: ‘at the edge’, in the cloud or on intermediate levels.