The 5G Hype Machine
The network management team at a large financial services company was getting hammered regularly by their senior leadership about when they could start taking advantage of 5G in their branch offices, and we were tasked with helping them understand the current state of deployment by surveying their existing strategic suppliers. While I can't disclose some of the information as it was received under NDA, there is a massive amount of information out there for the finding, though untangling the detail can be a bit of a challenge. I hope to simplify this for you all a bit.
5G - What it is
5G, or Fifth Generation, is a new architecture for wireless radio networks that is designed to accomplish several goals over 4G:
1) increase speeds and lower latency
2) increase the number of devices that can be supported on a given tower, as well as support low-power devices (think sensors)
3) increase flexibility for the mobile carrier to deploy new mobile applications within the mobile network
Some of the key architecture and technology improvements that enable this area:
1) deployment of additional mobile radio towers to increase density
2) deployment of new radio's that can operate at frequencies higher than 6 GHz, which has substantially more capability for delivering higher bandwidth at lower latency.
3) Simultaneous use of multiple lower-band frequencies to enable parallel downloads
4) deployment of virtualization technology, such a the Verizon Multi-Access Edge Compute platform, to enable largely as a function of deploying radio's that operate in frequency ranges higher than 6 Ghz (which does both at the same time)
5) separated control plane from data plane
The key here is that 5G isn't necessarily a speed, it's an architecture designed to pave the way for more speed, and this is why you see several carriers marketing their 5G networks already despite them not having met all of the key objectives. Take this commercial from T-Mobile claiming they rolled out their nationwide 5G deployment, and the rebuttal commercial from Verizon calling them out for having slower speed than their own 4G network. Or how about Sprint (now a part of T-Mobile) suing AT&T over false advertising of their 4G network as 5Ge. Confusing, isn't it?
How can these companies get away with this? Well, it comes back to the architecture. In this implementation guideline developed by GSM Association (an industry trade organization composed of mobile operators), you can achieve an initial deployment of 5G by deploying a New Radio (NR) for the data path, while using the "legacy" 4G network as a control path. The NR does not have to be deployed using the upper radio spectrum, so if you don't own any (like T-Mobile didn't until they purchased Sprint), you can advertise 5G availability without offering you, the end-user, much of anything in the way of a speed or latency improvement.
Why is this higher spectrum bandwidth such a big deal, anyway? High Spectrum radio waves (also commonly referred to as millimeter wave, or mmWave for short) will be able to support download speeds up to a massive 10 Gbps, or between 100 - 250x what you currently get on 4G, an lower latency by up to 90% (10ms for low-band down to 1ms in high-band).
Getting it to your mobile phone is another matter entirely. Radio waves in that spectrum have an extremely high attenuation, meaning they don't travel very far before the power degrades to unusable levels. They are also easily impeded by virtually anything, including building infrastructure, and even clothing! This is why the health claims over 5G are so ridiculous...the likelihood that these radio waves can hurt you at the kind of power levels we're talking about is pretty low.
The ramifications of this very simple fact create material challenges for deployment. First of all, just for a half-way reasonable level of outdoor coverage, new radio towers have to be deployed approximately every two blocks. In order to pass through the exterior barrier of your home, you will need an external antennae to carry the signal into your home (or directly to a WiFi router), and office building will require a Distributed Antennae System (DAS), which you would typically outsource to a firm like WWT. You could also deploy a Cradlepoint with an external antennae system up to the roof, though I couldn't find any of these for sale yet. No real reason for them to be, given the state of 5G deployment today...the market just isn't out there yet. Even with some 10 million active 5G devices out there (most of which are in Korea), the 5G deployment is expected to take place over the next 5 years.
There is also a limited selection of mobile devices that can support 5G. Even Apple won't have a 5G device available on the market until the anticipated release of the iPhone 12 in September.
Oh..the but the applications!
There is much speculation about the types of new applications this technology will enable. Take this commercial from Verizon about the use of augmented reality for firefighters, or this article about farm automation for the robotic milking of cow, or this piece from techradar on the applicability to VR. Some of these ideas are pretty cool, but do we really need edge compute embedded in our wireless providers network to enable them? Take the cow milking example. The article suggests that 5G is a solution because the farm operator has a shitty PC and broadband connection that can't handle all of the sensor data efficiently enough for the robots to react in real-time to the images they have to have processed somewhere. Well...why does the farmer have a shitty PC and broadband? Because there's no fiber to his farm, and that's because there's no business case for it. If there's no business case to get high-speed connectivity out to a farm, what makes us think that there will be a business case to deploy massive quantities of 5G towers to rural areas with that low of a population density? Is the farmer going to be able to justify paying the rental fees on local compute vs. just upgrading their computer? I honestly don't know the answer to these questions, but it it feels a little optimistic to me.
And what about VR? Assuming there was an application other than gaming, do we really need to get our video feed from a nearby remote system rather than a local system? Maybe with untethered systems, like those outlined here, but if we're in a multi-user VR environment with people from across the world, we still have latency due to the speed of light in fiber that can't be overcome without some non-relativistic-based science (and I predict that communication via quantum entanglement is not likely to be in 6G or 7G).
And then there is the autonomous car application...how our cars will communicate with other cars in near-realtime to avoid collisions. Do we really want our cars to require network connectivity to be autonomous? I don't know about you, but I'd rather have all of the required technology on-board the car to react to road situations. Any other useful information, like selecting routes and what not, can be done using a slower-speed connection.
And what about the use cases for our financial services clients? There are a ton of examples out there of potential applications, but they all read like "more bandwidth to more places is good." Video Booths, Augmented Reality trading desks, automated financial assistants, new wearable devices and more advanced bio-metric scanning... c'mon people, this is like a brainstorming session gone bad that never reached it's logical conclusion. You could deliver all of these applications today without waiting for 5G. To find the real use cases for this technology is going to require not just some brainstorming, but some prototyping, market testing and ingenuity.
When you're a hammer, everything looks like a nail
As skeptical as I am about some of the proposed use cases for the increased bandwidth and distributed compute, there is no doubt that this new infrastructure will have its uses. There are applications out there today that demand localized compute available at low latency, like warehouse and factory automation, and they are absolutely candidates for this architecture. Let's remember, though, that there are many technical solutions available to solve problems, and not every nail is going to need the hammer of wireless paired localized cloud compute offered by your wireless carrier. Fiber deployments continue on their march to serving more locations, DOCSIS 3.1 is already pretty awesome (I've got 600 Mbps at home), and DOCSIS 4.0 promises speeds up to 10 Gig. The big cloud providers are already increasing the density of their cloud footprint to serve more areas in a local way. As Public Enemy said, don't believe the hype, it's a sequel.
The ulterior motive
Carriers have been trying to find a way to stop being viewed as a "dumb pipe" for decades. They miss the days of the circuit-switched telephony network with massive DMS-500 and AT&T 5ESS switches that had all of the "smarts" of the network. Thanks to the Internet an high-powered computers, they lost all that "cloud-based" intelligence in the system and got severely commoditized as a result, and they desperately want to get it back. They envy the success of the big cloud providers and want to cash in on the cloud trend. But can they become a successful alternative to an AWS or Azure? While AT&T apparently marches towards deployment of their own Multi-Access Edge Compute (MEC) solution, Verizon is (smartly, in my opinion), partnering with AWS for it's MEC solution. Wouldn't businesses actually prefer a 3rd party cloud provider that is well connected to multiple networks, rather than one hosted inside a single carrier so they can leverage any and all carrier technology to serve their locations?
It's not an either/or
Don't get me wrong, 5G is going to be a great network upgrade, and it's an essential technology for wireless carriers to compete with terrestrial solutions, but I don't think it's going to be an either/or situation, but more of an 'and.'
So, what do you think? Do you get the feeling that some aspect of the 5G package is a solution looking for a problem? Leave a comment.
Siegel Group is a technology consulting firm specializing in enterprise network and organization transformation, working exclusively through Yates Ltd. If you would like to speak with me about a project we might be able to help you with, please reach out any time.