5G — on the Edge

By: Dave Eyre

19, December, 2019

Categories:

5G - 5G conference - IoT -

  •  
  •  
  •  
  •  
  •  
  •  

Alright, so 5G is all about low latency and fast speed that helps the advanced smart technologies that will be deployed in the future. Wait a minute. Just give it a thought. Is fast speed and low latency enough for Internet of Things, smart homes, smart stadiums and autonomous cars to work smoothly?

Let’s consider this scenario where there are thousands of sensors gathering different types of data. This data is transmitted over the 5G network to the core network where it is processed and sent back through the small cells to the device for action. In all probability this to and fro between the sensors and the small cell network to the core network and back the same channel is expected to cause a certain level of latency that is unrelated to the 5G network speed.

Then, how will we achieve low latency?

The solution that firms have devised is what is called ‘edge computing.’ In 4G, the processing takes place in the core network. The core is more of a centralised hub that sorts and processes data. However, in 5G, the processing will be done on the ‘edge’ — comprising towers and antennas. At least part of that processing will be done on the edge.

What is this edge?

The edge is the part of the network, which is closest to the user. Moving data to the edge helps reduce the distance between the user and the data. This, in turn, reduces the time needed for data flow. The shift to the edge also reduces the data that a server has to handle, reducing the load and speeding things up. This seems to be a more feasible architecture, considering IDC’s estimate of 175 zettabytes of data to be created per year by 20251. An average person will have over 5,000 digital interactions a day2.

Edge types

Based on the application, there are different types of edge computing. The European Telecommunications Standards Institute has developed Multi-access edge computing (MEC) which has the cloud-enabled service environment closer to the edge but not on the user device. For this type of edge computing, the developers have to define the application elements that will be processed on the MEC host and those that will be sent to the cloud. Cisco has defined another standard called ‘Fog computing.’ In this form of edge computing, the fog nodes have storage, network and computing capabilities. The collected data is stored temporarily on these fog nodes and later transmitted to the cloud for long-term storage. Carnegie Mellon University and Intel labs have talked of Cloudlets or a mini Cloud that has the capabilities of a Cloud but in small proportions. These Cloudlets store and process data. As a device moves, the Cloudlet detects the next closest Cloudlet and transfers the data there for processing. These Cloudlets will transmit the data to the Cloud only if needed2.

It is expected that these types of edge computing will work together to deliver the seamless, truly low-latency experience.

What challenges can we expect?

Just as 5G-compatible devices will be required for smooth 5G network, these devices will also have to support edge computing. Telecommunications equipment providers will have to determine the perfect balance of processing power and the device size. The software design will have to be capable of routing data to different processing nodes based on the requirement. As devices will not be static, the edge computing network will have to be able to seamlessly shift to the nearest edge node. Particularly for smart systems, the edge nodes will have to be capable of making intelligent decisions for quick responses. For example, the edge node of a smart home will have to use algorithms to detect dangerous situations like an oven left on or tap left running while the owners are out of home. The communication among the edge computing types and the core network will have to be flawless. The data will have to be routed to the right nodes within a short time. This will need a control plane to manage the routing. Finally, this entire network architecture will come with significant costs and security challenges. With such a deeply connected network, security breaches and malware may spread easily.

All in all, 5G will be an impeccable interaction between the core network, the different types of edge computing technologies and the user devices (sensors and all).

Get a complete understanding of the architectural web that 5G is at the upcoming 5G Expo London. Check out what experts shared at the 5G Expo North America.

References:

1https://www.orange-business.com/en/blogs/why-5g-needs-to-team-with-edge-computing

2https://www.infosys.com/about/knowledge-institute/insights/Documents/edge-computing.pdf

Other references:

https://uk.usembassy.gov/get-smart-core-vs-edge-in-5g-networks/