In a Nutshell

Decentralized data processing at the edge of the network
Communication with the cloud can involve individual or multiple levels
Enables local utilization of valuable (production) data
Broad use of technology from IoT sensors to wearables
Ideal for low latency, economical consumption & good service

What is Edge Computing?

With edge computing, data is processed where it is generated. Instead of centrally in a cloud, the data is evaluated and used either in the immediate vicinity or even locally in the device itself. This forms the “edge” of the network, or simply put, the ‘corner’ or “edge”.

In contrast to centralized cloud computing, this form of decentralized data processing offers the advantage that data does not have to be transported from the device to the cloud, evaluated and then sent back again. Especially in particularly time-critical contexts (e.g. autonomous driving), the possibility of fast and decentralized information will be absolutely essential in the future.

The aim of edge computing is therefore to significantly reduce the latency between the collection of data and the data-based response. Real-time data should therefore also enable real-time decisions. At the same time, local pre-processing of the data helps to save bandwidth in the network - communication between IoT devices/sensors and the network is limited to the most important data packets.

From Cloud and Fog to Edge - Three Computing Levels

In cloud computing, all data flows together in the cloud to be processed there. Edge computing shifts this task further to the edge of the network and only makes pre-processed information available to the cloud. However, companies can also take advantage of a third “layer” - fog computing.

This concept basically represents a “middle ground” between edge and cloud computing approaches. For example, a company could rely on site-specific servers that communicate with the IoT modules and sensors on site. These servers in turn send pre-processed data to the higher-level cloud infrastructure.

It should be emphasized that cloud, fog and edge computing are not exclusive concepts. For example, a company can rely on both fog and edge computing to optimize performance within the cloud and save resources within the network.

How Edge Computing Works

Edge computing processes can be presented in a simplified form as follows:

Collecting Data
Large amounts of data are generated as part of a (production) process involving (IoT) sensors and related devices. These processes potentially include everything from simple car journeys to the operation of large-scale production lines. This data is then collected in real time.
Local Data Processing
This data is analyzed on (or near) the device. The relevant process data is collected, evaluated and further processed. Information that is irrelevant to the process is discarded. This requires the necessary computing power of the edge devices (IoT sensors and the like) or an edge gateway.
Data-based Descision-Making
Time-critical decisions in particular can be made without significant delays based on the continuously processed information. For example, data analysis in / on the device can detect the signs of a heart attack in a patient and trigger an alarm. Or stop production processes if sudden deviations in the process are registered.
Data Exchange with the Cloud
While the edge computing approach “delegates” the most important decisions to the device level, the exchange with the higher-level cloud also takes place at the same time. Thanks to the pre-processing of data at device level, only the objectively important data/information is forwarded as an aggregate to central data processing.

What are Edge Devices?

Edge devices are a key factor in this process of decentralized data processing - they generate all the data. Depending on the environment, the data is then either processed at device level or - as is often the case in IoT environments - forwarded to an edge gateway.

These then take over the bundled processing of all incoming data. Typical examples include the following devices:

IoT Sensors (manufacturing / logistics / etc.)
Camera Systems / Image Capture Systems
Smart Thermostats / Smart Home Devices
Intelligent Traffic Lights / Traffic Guidance Systems
Wearables / Smartwatches

These and many other devices are what make edge computing in its current form possible in the first place. The immense amount of data generated every day - both in private contexts and in industry - could hardly be processed, let alone meaningfully evaluated in real time, without this additional computing capacity.

Graphical representation of the different levels of edge computing

The Advantages of Edge Computing

The possibilities offered by the edge computing approach are of interest to a large number of companies. Accordingly, numerous devices designed for this form of cloud infrastructure are already in use across all industries. It is precisely these advantages of edge computing that are of practical relevance:

Low Latency
Latency times can be significantly reduced with the help of edge computing. As the data is processed locally and does not have to be sent to the cloud first, there is no loss of time at this point.
Gentle Use of Bandwidth
As the edge devices involved only forward aggregated and pre-processed data to the central, higher-level cloud, these devices require considerably less bandwidth than those that are in constant communication with each other. This is particularly advantageous in regions/areas with poor network connections.
Higher Level of Security
As the edge computing infrastructure is based on decentralized data processing, a (temporary) failure of the cloud has no immediate impact on the functionality of the devices. They continue to work autonomously, so there are no security gaps (e.g. in monitoring systems).
Stronger Data Protection
If devices and data centers are constantly exchanging (personal) data, this represents a potential point of attack for cyber criminals. As large volumes of data are processed entirely locally, the risk of data theft and data breaches is generally reduced considerably.
Better Service Quality
Due to the fact that the functionality of edge devices (such as wearables and the like) is not linked to a continuously stable internet connection, these devices also work much more reliably from the user/company perspective. This in turn has a positive effect on product satisfaction.

When considering the technology, it is important to note that it does not replace traditional cloud computing, but rather complements it technically. You can find an overview of the main advantages of cloud computing here.

Where Edge Computing is Used - Uuse Cases

The concept of edge computing has been successfully used in various sectors and industries for many years - often without this being immediately apparent to customers and users. Yet a whole range of processes would be practically inconceivable today without this technology or would be too vulnerable to be of any real use. Here are some examples of applications for edge computing technology.

Industry 4.0

In industrial production, IoT sensors provide a continuous flow of data that allows conclusions to be drawn about current processes. As part of this production data acquisition, process-relevant data is collected and often forwarded directly to an edge gateway, where it is then analyzed and processed.

This location-based and real-time data analysis using edge computing makes it possible, for example, to detect irregularities or predict maintenance requirements (predictive maintenance). Such technologies are of central importance for Industry 4.0 and the smart factory concept (LINK).

Autonomous Driving

Vehicles generate huge amounts of sensor data per second (e.g. via lidar / radar / cameras), which must be processed immediately. Edge devices in the vehicle analyze this data locally, make decisions (e.g. braking, swerving) and only communicate important events or statuses to central systems such as a cloud.

This process is used by driverless transport systems (DTS) in warehouse or production logistics, for example. Here, autonomous floor vehicles transport materials back and forth between machines without any manual intervention. The vehicles recognize obstacles themselves, can actively avoid them, steer independently to their destination and at the same time remain in communication with the local control and guidance system.

Retail

In the retail sector, edge computing enables stores to process large volumes of (customer-related) data directly on site. For example, cameras and sensors can analyze customer behaviour without personal data having to leave the store.

Here, advertising displays can be dynamically adapted to customer behavior or other influences (time of day, etc.), sensors in refrigerated shelves automatically monitor their temperature, the quantity of goods on the shelf is continuously compared with stock levels - from furniture stores and fashion stores to drugstores and grocery stores, the application possibilities are almost endless.