DEV Community: Christoph Burgdorfer

Is Edge Computing just another Buzzword, or is there more to it?

Christoph Burgdorfer — Thu, 16 Apr 2020 09:57:35 +0000

In contrary to Cloud Computing¹, Edge Computing refers to connected computing in a decentralised fashion at the edge of the network. The idea is to move applications, data and services away from a centralised data center, and shift them to the outer boundaries of a network (hence “edge computing”). In other words, data streams and resources are at least partially processed on-site where they are being created or collected, while still benefiting from the advantages of the cloud.

This approach requires the application of resources that are not necessarily connected to the network in a permanent fashion, such as sensors, controllers, smart-phones or notebooks.

Edge computing also includes numerous technologies such as sensor networks, data creation through mobile devices, signal processing, peer-to-peer systems as well as ad-hoc networking.

This architecture makes Edge Computing particularly interesting for Internet of Things (IoT) applications.

Arguably, the history of computing has always been oscillating between edge and cloud computing. The very first computers were neither edge nor cloud computers, because they were not interconnected. As the computers started to become interconnected, users could use "terminals" to log into those mainframes, which arguably were the first cloud computers.
When the Personal Computer (PC) was invented however, the computing power would again shift to the edge, until the Internet would re-centralise computing power to servers and cloud machines. It's only the fact that computers are becoming smaller, more efficient and more inter-connectable again, which moves the computing power back towards the edge.

Advantages and Disadvantages of Edge Computing

Services of edge based applications significantly reduce the amount of Internet traffic and therefore costs and waiting times. There is no need for a centralised data center, which reduces the risk of data bottlenecks and single points of failure. Furthermore there is no issue with data privacy and data ownership as the data physically never leaves the space where it is generated.

There is also an advantage of security as the edge system is confined to the local, “physical” network.

Disadvantages of edge systems are that they take more effort to maintain, deploy and the data may be more distributed. This may also impact scalability.

Applications

Edge computing applications are becoming increasingly popular in any industry that is “up-smarting” physical space, whether it’s on the move or not. This can include making trains and planes more interconnected, autonomous driving, or shipping and logistics can include certain on-site computation. Stationery applications include production factories, smart buildings, public spaces, traffic systems or monitoring of agricultural infrastructure.

Especially big corporations that handle sensitive data are increasingly reluctant towards sending data to connected cloud system. They prefer if the “smartness” develops on-site and the building or office space can make “decisions” such as switching off unused appliances to save energy without sending any information other than analytics and control data to a centralised hub.

This strategy also increases the level of security as the outside is not involved in triggering any control-actions on-site.

Outlook

In the future we can expect that various edge computing systems can interact between each other through an agreed standard. I mentioned an example of this in my other blogpost, where a traffic light system at a cross road is an edge system, and so are the self driving cars that approach the traffic lights. Not only would the self-driving cars from various manufacturers be able to interchange data between each other through a local wireless connection without the internet, the cars would also be able to connect to the traffic light system during the time they are in its proximity.

Such systems can be imagined in many different contexts, including office buildings, retail, public space or even at home.

The future remains interesting.

Photo by Harrison Broadbent on Unsplash

Notes

"Cloud computing - Wikipedia." https://en.wikipedia.org/wiki/Cloud_computing. Accessed 16 Apr. 2020. ↩

What is IoT and why is it important?

Christoph Burgdorfer — Wed, 08 Apr 2020 06:34:29 +0000

The abbreviation IoT has been popping up increasingly frequently in the past few months. Partly also due to 5G technology¹ which seems to catch the attention of the media. But what is IoT? And why is it important? And how is it related to 5G?

IoT stands for the "Internet of Things". "Things" in this context relates to anything that can be connected. It doesn't even have to be connected to the internet like some sensors are just connected between each other, but the information they gather can then be connected to the Internet and therefore useful applications.

We are now living in a world, that is increasingly connected. There a number of factors that allow the world to become more interconnected:

Devices are becoming cheaper to produce: because devices can be and are now produced in larger quantities, the production has become more efficient.
Devices are becoming smaller and at the same time more powerful: American businessman Gordon Moore² described in "Moore's Law³" that the number of transistors in a densely integrated circuit doubles about every two years." This was true for the last 60 years.
The effect of Moore's Law is that the devices also require less energy to operate.
Emerging standards are increasingly allowing devices of different manufacturers and origins to communicate with each other, making them more valuable and usable.

Image Source: Wikimedia Commons

https://upload.wikimedia.org/wikipedia/commons/thumb/8/8b/Moore%27s_Law_Transistor_Count_1971-2018.png/1280px-Moore%27s_Law_Transistor_Count_1971-2018.png

We are on the verge of understanding that the more interconnected things, people, machines and data are, the more value can be extracted out of an ecosystem. A typical example is a building automation system. You surely have come across toilets that automatically switch on and off the lights when people are entering/exiting. These are old technologies such as so-called PIR (passive infrared sensors) and a light switch. But now imagine you could connect the switch on/off event to an analytics program, you may also be able to get some statistics, how many times a toilet has been used, which you could connect to a notification system to the cleaning crew, that goes and cleans after 10 sessions.

Suddenly the toilet cleaning process changes from a traditional "rota" which may incorporate a lot of "wastage" or inefficiencies, such as cleaning toilet cubicles that haven't even been used, to a highly efficient system, that not only works dynamically on-demand but also, if connected cleverly to machine learning, may autonomously improve its efficiencies.

This is just one example. Of course, we can go arbitrarily complex. Here is another example how connectedness of devices can create value, this time at the other end of the spectrum of complexity: autonomous cars. A modern autonomous car has three types of sensors: cameras⁴, radars⁵ and lidars⁶. Those are mounted in large arrays around the chassis and have shown that they can produce easily up to 11 Terabytes per day⁷. This means, that in the unfortunate event of an accident, there will be a substantial amount of sensor data available to analyse the event and the causes for it. A thorough analysis therefore must also result in conclusions, how such an event can be avoided in the future. Now imagine all this information to be aggregated and re-distributed across all cars over time, we can assume that no two identical (or even similar) accidents should ever happen in the future. One “device” has learned “for all devices”.

Image Source: Wikimedia Commons

https://upload.wikimedia.org/wikipedia/commons/thumb/d/d3/Waymo_Chrysler_Pacifica_in_Los_Altos%2C_2017.jpg/1280px-Waymo_Chrysler_Pacifica_in_Los_Altos%2C_2017.jpg

Edge and Cloud IoT Systems

Now let’s imagine a future, where everything that runs off electricity has the capability to connect everything else that runs off electricity. We can now distinguish between two different architectures: On the one hand side, there is the cloud based system⁸, which means that data travels through the internet to a cloud hosting infrastructure where it is processed and stored. On the other side, we have the “Edge” based system (sometimes referred to as Edge Computing⁹), where the majority of the processing and data consolidation happens decentrally at the edge of the network. Both systems have their advantages and disadvantages depending on how they are being applied.

A cloud based system is great for “remote control” or “remote surveillance”. But it requires permanent and robust internet connectivity and security. An edge based system is better for autonomous systems that can do processing on-site and do not rely on internet connectivity. This has certain advantages in terms of efficiency, security, resilience and potentially data ownership and privacy. But the magic really unlocks, once different edge systems can talk to each other.

Imagine that a car is an edge based computer system on wheels. The car takes to a different brand car behind, and another brand car in front. They exchange data and share learnings in real-time. But the data doesn’t go beyond their physical space. Then the cars arrive at a traffic light system, yet another self-learning edge computing system. The cars connect to the traffic light system, and bilaterally coordinate the most efficient way to cross the crossing. Meanwhile the systems are (machine-)learning.

These kinds of systems are the future. And thinking of it, isn’t this all too different from just humans interacting with each other?

Image Source:
https://media.giphy.com/media/Vy2PVgMNlCHdK/giphy.gif

In the next article, I am planning to publish, I will talk about getting started with your own little edge computing project. For this, you can use a platform called Gravio¹⁰.

Notes

"5G - Wikipedia." https://en.wikipedia.org/wiki/5G. Accessed 8 Apr. 2020. ↩
"Gordon Moore - Wikipedia." https://en.wikipedia.org/wiki/Gordon_Moore. Accessed 8 Apr. 2020. ↩
"Moore's law - Wikipedia." https://en.wikipedia.org/wiki/Moore%27s_law. Accessed 8 Apr. 2020. ↩
"Image sensor - Wikipedia." https://en.wikipedia.org/wiki/Image_sensor. Accessed 8 Apr. 2020. ↩
"Functionality and technology of radar sensors | Baumer." https://www.baumer.com/gb/en/service-support/know-how/function-principle/functionality-and-technology-of-radar-sensors/a/Know-how_Function_Radar-sensors. Accessed 8 Apr. 2020. ↩
"Lidar - Wikipedia." https://en.wikipedia.org/wiki/Lidar. Accessed 8 Apr. 2020. ↩
"Autonomous and ADAS test cars produce over 11 TB of data ...." https://www.tuxera.com/blog/autonomous-and-adas-test-cars-produce-over-11-tb-of-data-per-day/. Accessed 8 Apr. 2020. ↩
"Cloud computing - Wikipedia." https://en.wikipedia.org/wiki/Cloud_computing. Accessed 8 Apr. 2020. ↩
"Edge computing - Wikipedia." https://en.wikipedia.org/wiki/Edge_computing. Accessed 8 Apr. 2020. ↩
"Gravio." https://www.gravio.com/. Accessed 8 Apr. 2020. ↩