Edge computing: the nervous system for 5G networks

By Chris Pearson, President, 5G Americas

It never ceases to amaze me how much new technology gets packed into wireless networks as they are evolving. One of the most exciting things in the development of 5G is the establishment of computing power at the edge of a network, which is radically altering the way wireless is delivering data payloads from an end user’s mobile device to the cloud. Edge computing is one of these new developments and it is a technology enabler for 5G networks.

What does Edge computing do?

Edge computing enables new applications which require lower network latency for real-time operations, such as augmented and virtual reality for events, video and speech analytics, remote monitoring for video security, and others. Edge computing will be needed at scale for augmented reality applications, such as image recognition or at events, where computation on a mobile device would drain the battery, but processing in the cloud is too slow due to network latency for real-time uses.

Consider this: By 2022, 82 percent of all Internet traffic (both business and consumer) will be video, according to Cisco’s Global Mobile Data Traffic Forecast Update 2017-2022. Of that, 80 percent of video traffic consists of just ten percent of all the video titles (or files) available, according to caching software provider Qwilt. Because video content is so unevenly distributed around the Internet, it’s critical to put the right files at the Edge so users can quickly retrieve them on their devices – at the right time.    

For instance, smart cities alone will call upon numerous applications that are better enabled with Edge computing such as flame and smoke detection for first responders or improving communications between vehicles and people.  Other uses for edge computing include video and speech analytics, real-time object tracking, motion detection, event detection, audio translation and transcriptions. Edge computing enabled 5G networks can bring healthcare, security, and manufacturing to remote locations for operations that need real-time capabilities, such as Telemedicine, video surveillance – or remote monitoring of factory equipment.

What a smart city video analysis system might look like

But how does edge computing do all of that exactly?

Think of it this way: 5G networks take advantage of cloud-native concepts, such as containerization and micro-services through techniques like software-defined networking (SDN) and network function virtualization (NFV). As a network grows, it becomes more distributed over space, however if similar types of functions were grouped together, you could use the massive power of scale to do more powerful things. For instance, imagine you had 10,000 home computers each with its own CPU, graphics card, audio card, and RAM. If you grouped all the RAM together, you’d have a lot of RAM capable of doing many more things.

These cloud-native concepts of SDN and NFV allow different parts of a 5G network – even spectrum – to be shared or use open-source hardware or software, due to the separation of the control and user functions. This “disaggregation” of control and user functions even allows for the creation of new applications and services, for which open source initiatives will be critical to the development of standards in how these apps and services will be created.

As networks become more distributed, edge computing ensures data processing power moves near the periphery “or edge” of a network away from the core, allowing network traffic to be handled more precisely and efficiently – depending on the local needs of the user. One great achievement would be to allocate just the right amount of network and computation resources specific to a particular use – that’s what network slicing could do. And it’s achieved when you integrate edge computing power into the network.

How else will edge computing change the 5G wireless network landscape?

Edge computing can use innovative artificial intelligence and machine learning technologies to improve the management of data workloads across networks. In fact, edge networks themselves can be designed as autonomous and intelligent systems that sense the context around their environment and application, applying network resources in real-time.

Some major areas for artificial intelligence and machine learning include intelligent platforms, radio access network optimizations, end-to-end application delivery, network analytics and management, and subscriber and service management. Additionally, data from 5G wireless networks will be used for edge computing-based training of sophisticated artificial intelligence algorithms, enhancing new forms of emerging autonomous services and applications.

Machine learning is being applied across the entire wireless network

As wireless networks develop, imagine edge networks becoming “the nervous system” which connects the sensory cells at your fingertips with the core network, which is the brain. Edge computing will work as an interpreter for the sensory signals of the Internet of Things – billions of smart devices scattered in billions of different places around the world – with the data centers running algorithms of thought. Nerves need to parse and filter the noise from the signal, but they also serve as local decision-makers to help the mobile devices know what to do in real-time.

Just as every nerve is different depending which function it serves, from bicep curls, to digestion through the gut lining, to blood vessel constriction, so too is every Edge Computing architecture going to be specialized to the type of function or use-case it’s representing. In short:  Edge computing is complex and one size does not fit all.

Ultimately, a new reference architecture for edge computing-enabled 5G systems is being shaped that will have broad implications for how wireless networks operate in the future. This reference architecture will combine the elements of network function disaggregation with new models of programmability, new interconnections between radio access, transport and core networks, and implement artificial intelligence for self-organization. Work from dozens of standards bodies and open source edge architecture organizations will be required.

In the future, these changes may lead to entirely new Internet architectures that can include options for information-centric networks (or their hybrids). For instance, a network overlay may sit on top of, and allow for, the interoperability of existing models for Wi-Fi, Bluetooth, cellular, TCP/IP or Ethernet, or even more exotic options that stack different capabilities.

Imagine a world where the Internet of Things is communicated in real-time, sending many exabytes of data to the “nerves” of edge networks, transporting that data to core networks, all to be processed by sophisticated algorithms that turn the data into insightful, actionable knowledge.

That’s the world we’re heading into. It’s one of excitement and unimaginable wonder. I’m proud to be part of an industry that’s helping to bring forward a 5G world.

Chris Pearson
President, 5G Americas

Viewpoints are the expressed opinions of independent wireless industry analysts and stakeholders. They do not necessarily reflect the opinion of the 5G Americas association or its member companies. 

Find out more about edge computing and 5G networks

'5G at the Edge' explores radical edge computing architectures in wireless 5G networks.

Share this post

Share on facebook
Share on google
Share on twitter
Share on linkedin
Share on pinterest
Share on print
Share on email

Sign up to receive our announcements