Edge Computing technology has emerged as one of the hottest new technology proposals; data does not need to be fully centralised, instead some of it can be processed on distributed computers called Edge Nodes, that is, in the same place where the data is being generated.
The world’s first IoT device was created in the early 1980s, when students at Carnegie Mellon University in Pennsylvania found a way for a Coca-Cola vending machine to report its stock level over the campus computer network, avoiding unnecessary trips to the machine to check.
Today, over half of all electronic devices manufactured in the world are Internet of Things (IoT) devices, i.e. they are capable of communicating data over computer networks; and this will increase exponentially by 2025. The total installed base of IoT connected devices worldwide is projected to reach to 30.9 billion units by 2025, according to Statista.
It is fair to say that virtually any data needed for a company to strengthen or optimise its decision making or operational processes, is available on its computer networks at one level or another.
In 2011, the industry coined the term “data lake” to define those company databases that centralise data from a wide range of connected devices, without any rigid structures that they could easily develop for any use of that data.
In line with this analogy, some analysts have transformed the term into “data tsunami”, referring to the inability of many companies to use the huge volumes of data. The main battle today is not how much data you can obtain, but how to acquire and process it optimally and efficiently for subsequent use.
To navigate this “tsunami” of data from thousands of IoT devices, Edge Computing technology is a solution under which data does not have to be entirely centralised; instead, part of it can be processed on distributed computers called Edge Nodes in the same place where the data is generated. Only the result or aggregate of this processing is then centralised, thereby avoiding infrastructure overload, removing unnecessary latencies, and mitigating security and data sovereignty risks, all of which are so important to businesses and citizens today.
Imagine, for example, an energy distribution company that wants to adjust its production, almost in real time, depending on production and consumption levels. The infrastructure to communicate, centralise and store all this data from thousands of sensors is so complicated that the return on investment may not be viable. However, through Edge Computing, each substation can analyse the information in real time and only communicate with the centralised infrastructure about key deviations that will have a significant impact on the network.
Edge computing is becoming of significant interest with new use cases, especially since the introduction of 5G. The Linux Foundation 2021 State of the Edge report predicts that the global market capitalization of edge computing infrastructure will be worth more than $800 billion by 2028.
A great deal of work has been done by large corporations in defining and explaining what Edge Computing is and its different case studies, resulting in a range of definitions and classifications. All of them group the different types of Edge depending on the location where the data processing takes place.
When data processing is conducted at the closest point to the network and furthest from the devices, we speak of «Fog-Computing» (a term coined by Cisco) or «Thick-Edge». This occurs at distances of 100m to 40km from the devices, and is carried out by very powerful edge nodes, or in some cases even embedded in the network core equipment itself. This is the case, for example, of some 5G communications towers, which can perform data storage and processing avoiding unnecessary latency when the communicating devices are on the same network.
If data processing is performed on network equipment or data aggregators located in the local network itself, it is named by the «Far-Edge» or «Thin-Edge». The physical distances in these cases can range from 1m to 100m, and is carried out by lower-powered Edge Nodes, at 1GHz and no more than 8GB of RAM, which in many cases act as data concentrators, IoT gateways, or even intelligent industrial automation equipment.
Finally, when the processing is embedded in the IoT equipment itself, we talk about the “Micro-Edge”, which often has reduced functionality as the devices usually have a very limited computing capacity to avoid price increases or battery consumption.
There can be no doubt that IoT at the Edge is one of the new enablers that will accelerate the digital transformation in businesses. However, it is not without challenges that any organisation must consider in the design and implementation phase. The most important challenges we have identified are:
• Cybersecurity: as a network of distributed resources, in many cases unattended, and often connected to critical elements, the security design, protection and monitoring requires special attention. Even more so when Edge Nodes can operate the connected equipment.
• Scalability: especially in the “Far Edge” or “Micro-Edge” computing environments, the number of devices deployed can be very large (thousands to tens of thousands). This means that the supply, installation, and maintenance of Edge Nodes can add hidden rollout hidden costs to the point of being financially inviable. In the case of industrial installations, in particular, which have extremely long lifetimes, it is essential to have tools that facilitate this lifecycle management of Edge Nodes in a remote, centralised and scalable way.
• Integration: the typology of the connected devices, especially in industry, is highly fragmented. There are no fully pervasive communications protocols or common data structures. It is therefore important that an edge computing deployment is based on open technologies, ideally standard or widely used by industry, to allow effective integration and evolution of different parts with the existing infrastructure. Monolithic, closed solutions with high integration costs should be avoided.
Across all sectors, industrial companies are undergoing digital transformation processes. Being able to connect devices, as well as collect and exploit this data, is becoming crucial to competitiveness. The industries where IoT Edge can have the most impact are those that work with large numbers of connected devices. The impact is also exponentially greater where these devices are in distributed locations and generate data at high frequencies.
In this sense, they are clearly positioned:
• Utilities: Business continuity is key for the critical electricity, gas or water services sector. Monitoring their assets to detect – or even prevent - failures is a basic operational functionality. However, assets are often in remote locations. In this case, Edge Computing allows real-time analysis, with processing much closer to the asset, which means much less dependence on connectivity and better response times.
• Renewable energy: Edge computing can have a major impact on the sustainable management of limited renewable energy resources, such as solar and wind power. Again, in a remote and highly distributed environment, it can avoid high dependency on connectivity and provide and robustness and security for such a critical service. Edge Computing algorithms can assess in – and even predict – in real time, supply and demand of energy resources, with substantial improvements in the energy balance. Companies seeking to reduce carbon emissions are increasingly looking positively at the use of Edge Computing combined with the Cloud in this regard.
• Smart Grid: With the emergence of distributed energy resources, such as electric cars, chargers, batteries, self-consumption solar panels, and other elements, local decision making can lead to very high energy efficiency improvements for companies or large communities. Given the complexity and variables involved in this kind of management, it is not expected to be left to the users and therefore must have a high degree of automation. Given that data privacy may imply restrictions on use, the heterogeneity of devices may complicate its integration in cloud platforms, and latency or errors may have implications for the business case, Edge Computing is now positioned as a high-potential architectural solution.
• Logistics and Mobility: In this case, where assets are not only distributed and multiple, but mobile, there is a growing trend in the use of Far-Edge technology with in-vehicle computing nodes. From the most basic cases of sensorisation for securing loads or critical elements inventories to advanced route optimisation or even semi-automation of driving, low latency response and overall system reliability are fundamental aspects when placing the computing at the closest point to the assets.
• Distributed Manufacturing: Distributed, or decentralised, manufacturing is understood to be manufacturing where the product is made in a network of several smaller, geographically dispersed facilities coordinated through computer networks. These networks have traditionally had very high installation and maintenance costs due to the need to transfer large amounts of data frequently between multiple sites. While cloud computing has been a significant improvement in these environments, the combination with algorithms at the edge can optimise investment, improve data security, and facilitate compliance with industry regulations that do not fit so well in cloud environments.
If you were interested in this article and want to know more about the possibilities of IoT Edge Technology, please contact us!