How Edge Computing Will Deliver on the Promise of 5G
Previous years at the Mobile World Congress Barcelona — the premier world conference for mobile consumer electronics — focused on topics like bringing the next billion online. This year at the event, you couldn’t walk a minute without someone touting the promise of 5G, the next generation of cellular mobile communications that promises to bring dramatically higher data rates, reduced latency, and massive device connectivity.
The only problem is, for the average consumer, 5G isn’t that thrilling. We can accomplish almost everything we can imagine with 4G or even 3G connectivity. Add to that, most providers, phones and local infrastructure still can’t support it. As you walked past all the flashy stands and suits paired with sneakers, 5G started to feel like a lost investment.
That’s because it’s not meant for us. Like most technological advancements, industry will come first. The promise of 5G is in swarms of drones and fleets of autonomous vehicles. It’s in lightweight industrial wearables that process crucial information rapidly. It’s in robots. And, yes, it’s in the augmented reality game we’ve all been waiting for, “Harry Potter: Wizards Unite.”
This promise of 5G will be most likely delivered on the edge, taking it from conference marketing to real-world feasibility. In fact, 4G can already deliver a 5G-like experience with edge computing.
A Primer on Edge Computing
So what exactly is edge computing? “The edge is anything that is close to the users,” said Sunay Tripathi, chief technology officer of MobiledgeX, which recently touted the world’s first public mobile edge network deployment with Deutsche Telekom.
This could be your devices at home like Apple TV or Alexa or the servers right across the street servicing your neighborhood. For him, it’s how your home Internet connects to the world.
Tripathi says, “Keep it local, as close to the users as possible, relieve the strain on the public cloud.” MobiledgeX focuses on device mobility and leveraging nearby edge servers.
Tripathi explained that “It doesn’t matter what kind of device you have. The nearest commute to that.”
He says that MobiledgeX leverages the trillions in investments operators have already made in their networks to support sub-30-millisecond communication between users on existing networks — anything above 30-millisecond and there will be a lag that will negatively impact performance.
When 5G becomes available, speeds could drop to as low as ten to 20 milliseconds. MobiledgeX automatically searches for where the network isn’t already oversubscribed.
“If you’re a mobile user — be it a group of cars or trucks or drones — you connect to the nearest tower and we connect the infrastructure close to you. We don’t change the developer workflow,” Tripathi said.
MobiledgeX still deploys in the cloud, using lambda-based containers in the backend to stream it.
With MobiledgeX, devs are only paying for what they’re using, while it automatically finds the nearest edge computer location to serve the user for that short time.
Tripathi offered the example of a 15-minute immersive video game. Either the devices and/or the game developers are connecting via the edge, but instead of buying a CPU for two to three months, the developer only needs to one CPU in the public cloud for that brief time.
In the traditional path to connectivity, if you have 10,000 users in 10,000 locations, you’d have to pay 10,000 times. “Instead of more static deployment of public cloud, you be more dynamic on the edge [with] a lot more sharing,” Tripathi said.
Where 5G Will Really Make a Difference
Tripathi said that this first year of MobiledgeX has been about narrowing down the use cases, like robots 3D mapping a space to help workers be more productive or rich video streams that require a lot of data on multiple devices to be processed in real time.
He continued that 5G processing on the edge is needed to pursue the trend toward more natural interfaces, which combined video and audio — “things we can see and things we can feel” — with lighter devices.
“The digital world is coming to the place where if you look on the client side, we are increasingly not using keyboard, but, if you look on the backend, it’s still designed for more keyboard and static applications,” Tripathi said
If you are dealing with a lot of people with audio and visual in a way that demands more processing needs, then you need edge computing. However, it also is necessary when smaller devices are needed.
Echoing this, Justyna Janicka, co-founder at 1000 Realities, told The New Stack how HoloLens, Microsoft’s mixed-reality headset, is far too bulky for practical Industrial Internet of Things (IIoT) applications. In the image adjacent, she is showing how the combination of 1000 Realities’ app running on MobiledgeX with 5G allows for more lightweight smart glasses to use augmented reality to quickly scan the temperature of servers in a data center. Without 5G on edge computing, such a lightweight device couldn’t handle this processing speed.
Edge computing is cost- and CPU-effective when you have many users and want to create a rich, multimedia experience, where you can’t have the load dropping down to 3G.
Where the Edge Will Drive 5G (and Profits for App Developers)
The first widespread use case for 5G was augmented reality powerhouse Pokemon Go. Parties of hundreds of monster catchers would gather in parks, crashing mobile coverage for everyone near. 4G networks pinging to towers simply don’t have the bandwidth of streaming videos and augmented reality on a multitude of devices.
Germany is at the forefront of 5G with Deutsche Telecom having made a huge investment already. By 2020, we will see quite a few major cities around the world with 5G connectivity, although it remains to be seen if the phones will have caught up by then.
Currently, in some of the most crowded spots in the Western world — Times Square, New York, or Canary Wharf financial district in London — there are so many users on the public cloud that everyone gets bumped down to 2G because the network can’t keep up.
This happens because everything is pinging up to the public cloud, and the typical service provider only has about three to four peering points — they allow for the exchange of information between users. Imagine if a hundred people were battling it out on an augmented reality version of Fortnight.
Tripathi argues that between the device and the edge, there is plenty of bandwidth to enable this multitude of multipoint device-to-device interactions over video content and immersive experiences.
He also argues we will soon see revenue models shifting from the biggest cloud providers toward the people who are providing the infrastructure and who are owning the applications
“Currently the cloud providers are making a disproportionate amount of money because everyone has to live in the public cloud,” Tripathi said.
He continued that the infrastructure spend is already done and the cloud operators are already using it. Edge computing will just allow everyone to use it more efficiently.
Tripathi says that companies who will most benefit from this push to the edge are lots of small app development companies with 40 to 100 developers.
“Smaller guys are not aware about that edge is starting to take shape and they can actually build their applications to be edge enabled and reach a whole other set of clients and experiences,” he said, hoping he can raise said awareness.
These are the devs who think they can’t do rich videos on 4G bandwidth. Edge is deployable on today’s 4G networks already, which actually enables devs to offer 5G-like capabilities including video-rich interactions now.
“The hardest part is to reach out to these small developers. They just need to know that edge exists and what does it take to build — we [already] have stable software and edge locations,” Tripathi said.
While much of this year’s Mobile World Congress’s ubiquitous promise of 5G felt a bit far off, certainly edge computing can help app developers deliver that promise today.
How Edge Computing Will Scale 5G
So, how will edge computing help scale 5G?
For Tripathi, he sees this as simple because, by the end of 2019, we will see more than 20 billion devices connected and massively distributed around the world. The first couple billion are our ubiquitous smartphones, but the rest will be dominated by connected wearables, vehicles and drones. Suddenly infrastructure will have to deal with a billion odd devices spread around 10,000 locations.
It’s not just about the 70 percent living in cities but edge computing can also be used to bring that next billion online with smaller, more efficient localized servers in rural communities.
Over the next ten years, Tripathi pegs edge computing as having ten times the impact of what the public cloud has had over the last decade. He says edge computing will be able to handle double the scale with many more clients per second over many more distributed locations and architecture.
At the time of the first interview, about five application developers were using it along with about 8,000 users in Germany. When we met face to face a week later at Mobile World Congress, they had millions of users, including three edge servers set up around the conference space and dozens in key cities.
The United States is a unique use case for this because most countries see cell phone towers at least partially owned by the public. In the U.S., it’s neither service providers nor the government, but mainly third-party private businesses Crown Castles and American Towers who run them.
MobiledgeX is working with the The New Stack sponsor Packet who are setting up the bare-metal infrastructure on these towers.
“By next Mobile World Congress, we’ll have a much bigger edge in Europe for sure, and the edge in the U.S. will start taking shape,” Tripathi said.
“For developer communities, the world is a global market. You can start benefiting from edge in Germany while the U.S. and the rest of the world launch it.”