A dream combination: IoT and edge computing

The proliferation of Internet of Things (IoT) devices is the biggest driver of edge computing; in turn, edge technology is creating new applications within the IoT framework.

For the Internet of Things to work, many different sensors and microprocessors are needed, which are integrated into IoT devices, generating a large amount of information at any time and anywhere, which must be processed as real time as possible.

However, with traditional network architectures, this becomes increasingly difficult because data must be transmitted remotely to a central data center. This alone creates a certain delay, and in addition, data centers are faced with an increasing burden of processing data because the growth rate of information exceeds their capacity. And after processing, the processing results must be transmitted to IoT devices, which will take even more time.

Furthermore, IoT mobile devices are often unable to perform high-performance wireless communications everywhere (e.g. drones or unmanned vehicles), which would cause further delays or even downtime, making it impossible to react quickly to new requirements.

[[250375]]

Edge computing application cases

A good example is a self-driving car: driving on the road, an inattentive pedestrian walks in front of the car. Although the self-driving car's built-in camera can recognize this person, the image must be sent to a central data center for processing before the brake control is transmitted to the car. In this case, the reaction is a little too late.

As a result, the automotive industry is installing processors in vehicles that can instantly process camera images and respond to unexpected hazards in real time. This is edge computing, and it involves processing data at or near where it occurs—connected cars, each of which can now generate and consume terabytes of information every day.

Of course, the advantages of this decentralized data processing have also been recognized in various other industries. In manufacturing, for example, the data generated by networked devices must also be processed as quickly as possible so that machines can immediately adapt to new production requirements or faulty components can be detected and repaired proactively.

In addition, transmitting the data generated by this IoT to a central data center or cloud through the network can be time-consuming, resulting in delays and even data loss. In contrast, edge computing allows all data to be processed and analyzed in real time, thereby improving consistency and response time. Due to these advantages, the proportion of data created and processed outside of centralized data centers or cloud systems is expected to grow from the current 10% to 50% by 2022, according to a Gartner report.

Data Center Expansion

However, edge computing is rarely a standalone solution and is often used as an extension of a data center. While the technology is great for fast data processing, it doesn’t store much data, so it can’t identify long-term trends or perform comprehensive analysis.

The data is therefore processed, aggregated and compressed at the edge and then collected and transmitted to a central data center at regular intervals. This records and stores the information, which is then evaluated as part of big data analytics, which can be used to optimize processes or develop new solutions.

A specific application example is a police body camera, where a portable microcomputer on the camera or the camera itself can compress and encode the captured videos and then send them to a local edge center to speed up the upload process and reduce the load on the central network.

Retail point-of-service (POS) machines can also benefit from this process, for example, sending a customer’s shopping data to an edge computer that performs the necessary checks and transactions. Not only does this speed up the process, it also eliminates the possibility of sending sensitive information over the network and reduces the possibility of exposing it to attacks.

Safety

Edge computing is not more secure than traditional architectures, so enterprises also need to perform risk analysis and design an overall security architecture for this approach.

On the one hand, edge computing can improve transparency into where data comes from and where it goes, thus simplifying security management. In the case of a central data center or cloud system, high traffic volumes can be difficult to monitor for under-resourced businesses, and cybercriminals can take advantage of this to secretly intercept data. Therefore, edge computing often provides better control over these connections and their security.

On the other hand, a greater number of sensors also increases the attack surface, so more connection points need to be protected. Therefore, enterprises need stricter patch management, which can be quickly replicated and transmitted to various sensors that collect and send data. If sensors are not properly protected, hackers may attack them and seriously damage IoT systems, such as remotely controlling the brakes on a connected car. At the same time, hackers can break into corporate networks through unpatched vulnerabilities or merge IoT devices into botnets to perform DDoS attacks.

This means that only by taking comprehensive security measures can enterprises benefit from the Internet of Things and edge computing. In addition, secure edge solutions also allow new IoT applications to process more data on the spot and respond faster to new and complex requirements. In the near future, there will be truly intelligent robots, drones, machines and cars; and perhaps one day, there will be fully autonomous learning IoT edge systems.