5G abstract

T-Mobile Expands Mid-Band 5G

T-Mobile just launched mid-band 5G in a bunch of cities. Here’s the key bit from the press release:

T-Mobile lit up a supercharged 5G experience in another 121 cities and towns with mid-band 5G, delivering up to gigabit-per-second peak speeds and average download speeds around 300 megabits-per-second for capable 5G devices.

The expansion into mid-band 5G is exciting. While T-Mobile has been destroying its competitors in terms of 5G coverage, T-Mobile has relied mostly on slow, low-band 5G.

Twitter sweepstakes

In the press release, T-Mobile mentions a sweepstakes to encourage Twitter users to discuss T-Mobile’s 5G:

To celebrate, T-Mobile is giving away $100,000 on Twitter over the next several months, with #5Gsfor5G.

Interestingly, T-Mobile’s President of Technology, Neville Ray, criticized this sort of thing in a Tweet he shared the same day the sweepstakes launched:

DISH, MATRIXX, and Dynamic Pricing

DISH and MATRIXX Software just came out with a press release titled: “DISH selects MATRIXX Software for dynamic pricing and monetization of its 5G network”.

I’m a huge advocate of variable-rate pricing. Varying data charges based on how congested cellular networks are would bring huge efficiency gains.

While the press release seems promising, it’s full of corporate jargon and light on substance. Here’s the key bit from the press release:

MATRIXX’s API-first architecture is proven to deploy quickly and cost-effectively. Combined with DISH’s AI strategy, it will determine network availability and utilization, dynamically changing prices throughout the day. MATRIXX’s cloud native, continuous integration/continuous deployment (CI/CD) pipeline then automates pricing updates.

Cutting Through Bullshit Around 5G Latency

There’s a lot of unrealistic hype going around about 5G. Most of the hype focuses on the blazing-fast speeds 5G can offer.

Today, T-Mobile’s CEO, Mike Sievert, made a big deal about 5G latency rather than 5G speed:1

Latency measures the time delay involved in data transfer. It’s common for 4G connections to have a latency of about one-twentieth of a second (50 milliseconds). Some 5G technologies may be able to push latency far lower.

The video Sievert shared is allegedly a demonstration of latency under different technologies. It’s not remotely fair. The phone supposedly demonstrating a 4G LTE connection is about five seconds behind the real world. Latency isn’t anywhere near that bad with 4G. If it was that bad, normal voice conversations and video chats using 4G wouldn’t be possible.

Rethinking “Nationwide”

T-Mobile and AT&T started describing their 5G networks as nationwide once the networks covered over 200 million people. I’ve seen multiple people suggest that this is related to FCC rules. Allegedly, the FCC only allows networks to be described as nationwide when they cover over 200 million people. I’ve searched around, and I can’t find any FCC documents mentioning such a guideline.

As far as I can tell, the 200 million number comes from the National Advertising Division (NAD), a self-regulatory body for the advertising industry.1 Here’s an excerpt from a 2014 NAD publication:

NAD noted in its decision that it has applied a consistent standard for ‘coast to coast’ service for the past 10 years. In general, a wireless network can claim to be nationwide or coast to coast if the provider offers service in diverse regions of the country and the network covers at least 200 million people.

200 million people would make up about 60% of the U.S. population.2 I don’t think a network covering 60% of the U.S. population is nationwide in the common-sense meaning of the word. If networks with such lackluster coverage are advertised as nationwide, consumers will be misled.

The NAD should update its approach. The exact meaning of nationwide isn’t clear cut, but I think even a loose standard should be something like this:

Nationwide network: A network that covers at least 85% of the U.S. population and offers service in some parts of every state.

The NAD should probably frame its standard in terms of a percentage of the U.S. population covered (rather than a raw number of people covered). In 2004, 200 million people would have been almost 70% of the U.S. population.3 The NAD’s standard made more sense then. As the country’s population has grown, the NAD’s standard has become weaker.

AT&T’s 5G Coverage Now “Nationwide”

This morning, AT&T announced that its 5G network is now nationwide. By “nationwide,” I believe AT&T means that the service is estimated to cover over 200 million Americans. There are still plenty of parts of the country that are not covered by AT&T’s 5G service.

AT&T shared a few other noteworthy items in its announcement:

  • AT&T will add 5G support to the Unlimited Starter plan (the carriers’ most basic postpaid unlimited plan) at no extra charge on August 7.
  • Business customers on the Unlimited Web-Only plan will also get 5G support starting on August 7.
  • Cricket Wireless will begin offering 5G service on August 21.

On all of these plans, 5G service will only be available for subscribers with 5G-compatible devices.

5G Coming To Cricket Wireless

Today, AT&T came out with a press release stating that 5G service will be coming to Cricket Wireless. Here’s the relevant bit:

On Aug. 21, Cricket Wireless customers will be able to activate 5G service on the Samsung Galaxy S20+ 5G. More details soon!

As far as I can tell, Cricket has not yet updated its own web pages about 5G.

I expect Cricket subscribers will be able to access both AT&T’s sub-6 5G and AT&T’s ultra-fast millimeter wave 5G. While the S20+ will initially be the only 5G phone offered by Cricket, we should soon see the carrier offering more 5G-compatible devices. It may also be possible for Cricket customers to bring their own unlocked 5G phones to the service.

Tracking 5G Strategies and Deployments

I’ve created web pages to track the 5G strategies used by each major network in the United States. The pages go into detail about networks’ 5G strategies, compatible devices, and coverage profiles. I plan to update these pages regularly as 5G deployments move forward.

I intentionally omitted Sprint from the list. T-Mobile has been pulling the plug on Sprint’s 5G service. The spectrum Sprint used for 5G will be redeployed in the New T-Mobile’s network.

Verizon’s 5G Mobile Hotspot & Tethering Policies

Verizon’s higher-end unlimited plans come with monthly allotments of full-speed 4G hotspot data.

  • Play More Unlimited – 15GB
  • Do More Unlimited – 15GB
  • Get More Unlimited – 30GB

Subscribers that burn through all of their full-speed 4G data can continue to use more hotspot data with a 4G connection throttled to a sluggish 600Kbps speed.

Verizon doesn’t prominently advertise it, but the company has separate policies for 5G hotspot data. Here’s a bit from a section of Verizon’s website that gives additional details about one of the carrier’s unlimited plans:1

Unlimited 5G Ultra Wideband Mobile Hotspot

An unlimited 5G Ultra Wideband mobile hotspot means your smartphone becomes an ultra-fast and reliable Wi-Fi connection for your other devices. Service available in select locations and requires a 5G capable device.

Yesterday, I was exploring my Verizon online account for the Play More Unlimited plan and found this:

Screenshot of Verizon account interface showing a 5G hotspot allotment

Verizon is currently allowing 50GB of full-speed 5G hotspot use each month on the Play More plan. After 50GB of use, speeds are throttled to 3Mbps. While 3Mbps isn’t a great speed for hotspot use, it’s still fast enough that most people could browse the internet and do work on a laptop without overwhelming frustration.

I’m guessing the Do More and Get More plans have the same 50GB allotment, but I’m not sure. It’s odd that Verizon’s 5G hotspot allotment on the Play More plan is so much larger than the regular (4G) allotment.

I wonder if the large allotment is a promotional thing. Will we see the allotment shrink as more subscribers get 5G-compatible phones?

Post publication update: As of January 1, 2021, Verizon appears to have dropped the data cap and throttle for Ultra Wideband hotspots.

Opensignal’s Report on U.S. 5G Performance – No Big Surprises

Earlier this week, Opensignal released a report on the performance of 5G networks in the United States. Opensignal’s report puts some numbers and data behind two things that were already clear:

  • T-Mobile is destroying the competition in terms of 5G coverage, but T-Mobile’s 5G isn’t very fast
  • Verizon’s 5G is outrageously fast, but the coverage profile is terrible.

Opensignal primarily collects its performance data by crowdsourcing data from tests that run in the background on regular people’s phones. It looks like the company restricted the data underlying this report to include tests run from 5G-compatible phones.

Speeds

Verizon destroyed the competition with an average 5G download speed of 495Mbps. The other major networks in the U.S. had 5G download speeds averaging around 50Mbps. Verizon’s dominance in download speeds is due to the company’s focus on rolling out millimeter wave 5G.

Coverage

Unlike Verizon, T-Mobile has focused on deploying sub-6 5G. This type of 5G is great for covering large areas, but less impressive for delivering high speeds. Unsurprisingly, T-Mobile dominated in terms of 5G availability. According to Opensignal’s data, T-Mobile subscribers were able to access 5G 22.5% of the time. Verizon did about fifty times worse with an availability score of 0.4%.

While 0.4% is low, it’s still a better availability score than I would have predicted for Verizon. I wonder if Opensignal’s crowdsourcing approach might lend Verizon’s availability scores a leg up. If living near a Verizon 5G deployment makes a Verizon customer more likely to purchase a 5G phone, selection bias can creep in and cause Opensignal to overestimate Verizon’s actual 5G availability.

Silly press releases

Following Opensignal’s release of its report, T-Mobile published a press release. The company bragged about the network’s excellent 5G coverage without mentioning that the network got demolished in the download speed results.

Verizon published its own press release bragging about ludicrous download speeds. Verizon’s awful 5G availability score was not mentioned.

Download Speed Experience – 5G Users

Opensignal’s report included a metric called Download Speed Experience – 5G Users. The results for this metric were calculated by looking at users with 5G-compatible phones and tracking their average download speed even at times where they did not have 5G connections. In some sense, this single metric does some accounting for both 5G speeds and 5G availability.

Verizon and AT&T tied for the top spot:
Results graph showing AT&T and Verizon tied for the top spot

The metric is interesting, but I don’t think it quite captures how users will feel about the quality of their download speed experiences. The marginal value of a 10Mbps boost in download speeds that moves a subscriber from 5Mbps to 15Mbps is much greater than the marginal value of a 10Mbps boost that moves a subscriber from 500Mbps to 510Mbps. Collapsing a distribution of download speeds into a single, average download speed masks this reality.

5G abstract

5G+ Brilliant Marketing From AT&T?

A while back, AT&T began calling some its 4G service 5GE. It tricked consumers into thinking they were getting 5G when they weren’t. It was bullshit.

Now, AT&T is calling its millimeter wave (mmWave) service 5G+. In some ways, the 5G+ label could confuse consumers. Verizon’s 5G service uses pretty much exclusively mmWave right now. Verizon doesn’t call its mmWave service “5G+.” To Verizon, mmWave is just “5G.”

Still, I’m not annoyed by AT&T’s 5G+ branding. Not all 5G is the same. Lower frequency, sub-6 5G, is typically way slower than mmWave 5G. Laypeople don’t realize how different sub-6 5G is from mmWave 5G. If we want consumers to understand the difference, carriers need to help make that happen. Calling sub-6 service “5G” and mmWave service “5G+” makes a clear distinction for consumers.