33 Comments

Summary:

Who has built the fastest network? The most resilient network? The network with the most room to grow? Gigaom takes a look at how the carriers stack up on LTE given their recent flurry of activity.

cell phone tower / cellphone tower / antenna
photo: Shutterstock / Pavel Ignatov

Given that President Obama delivered his State of the Union address on Tuesday, I figured this was as good a time as any for an update on the state of LTE in the U.S.

Since 4G technologies arrived early here, LTE probably seems like old news. New 4G market launches are hardly cause for headlines and LTE is pretty much standard in any new smartphone or connected tablet. But the carriers have been busy in other ways over the last year.

They’ve expanded their networks into new spectrum and implemented new technologies that squeeze more capacity, speed and range out of their existing towers. While it might be easy to view LTE as simply a status marker — either you’ve got it or you don’t — the reality is that no two LTE networks are the same.

This is my attempt to explain what the four nationwide carriers have done to differentiate their networks in the last year. I’m not trying to make a judgment about who has the “best” network per se, and I’m not delving too much into speed tests or other benchmarks. But I’ll describe what exactly each of these carriers has built and what those technologies will mean for the consumer mobile subscriber.

Verizon’s new network beast

Verizon Wireless was the first major operator to launch LTE, and last year it started showing its age. Its speeds started slowing down as more than 40 million LTE devices began connecting to its networks. In some major markets, the LTE network got so overloaded that it began kicking people down to 3G.

But right after Verizon installed the last base station of 4G network No. 1 in June, it started building a second, much more impressive one. While the first network was focused on coverage, No. 2 is focused on creating capacity in major cities where mobile data demand is highest.

Verizon's LTE coverage in dark red

Verizon’s LTE coverage in dark red

In many major cities like New York, Chicago, Boston and Seattle, Verizon has built new 40MHz systems, tripling its capacity. In some cities in the western U.S. — notably San Francisco and Los Angeles — Verizon was only able to field 20MHz or 30MHz systems, but even those deployments equate to at least a doubling of capacity.

What this means is Verizon is now able to stack tens of millions more LTE devices onto its networks without overloading them. And for the speed-obsessed out there, Verizon’s network has become the one to latch onto. Early tests clocked downlink connections of 80 Mbps, though the network is sure to slow down as it fills up with subscribers.

Verizon is uncharacteristically quiet about the rollout, and that’s likely because it will be a while before all of its customers can access it. To connect to the new network, subscribers need a Verizon device that support the 1700 MHz/2100 MHz Advanced Wireless Service (AWS) band. About 20 percent of Verizon’s smartphone base — including the latest iPhones and iPads, Motorola’s new Droids and the latest Samsung smartphones — can today, but nearly all new devices going forward will have the necessary radios.

T-Mobile’s LTE miracle

T-Mobile has been boasting a lot about its 4G networks of late, and in truth it has a lot to brag about. The country’s smallest national carrier went from zero to nationwide LTE in just eight months, and in November it doubled the capacity of its network in at least 40 major cities.

T-Mobile’s 20MHz network is pretty much on par with AT&T’s network as well as Verizon’s first generation network. But because of T-Mobile’s small size, its fewer 4G customers have a lot more bandwidth to play with. The result is T-Mobile’s current customers are starting to see consistently higher speeds than the other carriers.

T-Mobile US CEO John Legere at CES

T-Mobile US CEO John Legere at CES

That led T-Mobile CEO John Legere to claim that T-Mobile has the fastest network in the country “bar none,” which is a bit misleading. T-Mobile’s numbers are based on Ookla speed tests across the current installed base of smartphones, not the newest capabilities of today’s devices and networks. Here’s how that translates: if you compare two older iPhones on both networks today, T-Mobile’s will most likely be the fastest. If you’re buying a new iPhone 5s, Verizon probably will give you far faster speeds.

Legere’s bravado aside, T-Mobile has an intense network, and it’s only going to get more powerful in the next year. The “Un-carrier” plans to double its network capacity once again in several dozen cities, which will give it an LTE monster to match Verizon’s.

The big knock on T-Mobile has always been its lack of mobile broadband coverage outside of the big cities. But if you happen to live in one of those cities, you’re set when it comes to 4G. For the foreseeable future T-Mobile will have capacity out the wazoo, allowing it to pile millions more subscribers onto its networks while maintaining impressive speeds. And by the end of the year, it truly will have the fastest network in the country.

AT&T plays Dr. Frankenstein

To be honest, AT&T’s days of making big 4G headlines are probably over. Ma Bell had a big run over the summer after several independent studies all crowned AT&T the 4G speed king, but since then it’s been forced to back down from those claims.

AT&T simply lacks the big hunks of contiguous airwaves necessary to deploy the 40MHz behemoths that Verizon and T-Mobile are building. But that doesn’t mean AT&T lacks resources. It has spectrum in bands all over the map, and it’s using every megahertz it can. It’s cannibalizing old 2G and 3G capacity for LTE in the 1900MHz PCS band and it’s tapping into the virgin bandwidth locked up in its AWS licenses. It has boutique airwaves at 2.3GHz and it’s eyeballing the cellular band as well.

frankenstein

By stitching all of those frequencies into the same system, AT&T won’t get the speediest of networks, but it will get a lot more bandwidth to divide among its customers. Basically AT&T is adding many more lanes to its mobile broadband highway even though the speed limit will remain the same. What’s more, AT&T is delving into new technologies like small cells, which will surgically insert bandwidth into the highest-demand niches of its footprint, and self-optimizing networks (SON), which will allow cells to literally grow and shrink, following subscribers as they move through the network.

In its marketing, AT&T has stopped focusing on speed and is now hyping its network reliability, which makes sense given its situation. Sure, the speed demons among us should look to T-Mobile and Verizon, but for many subscribers the more important thing is maintaining a consistent, resilient connection at a respectable speed. And that’s where AT&T’s network is heading.

Sprint fans its Spark

Spark is an apt name for Sprint’s new 4G network. Spark could just as easily explode into a roaring blaze as it might fizzle into nothingness. Sprint has been making a big deal about the new tri-band system it launched last fall as an overlay to its existing LTE network, but at least for now, Spark is nothing to brag about.

sparkSprint is advertising Spark as a revolution in mobile broadband, combining three disparate bands in a kind of super-LTE system. First off, Spark isn’t really accessing three LTE bands, rather just two: its main LTE network in the PCS band and the 2.5 GHz airwaves it got when it took over Clearwire. The third band — the 800 MHz band it cannibalized from its old Nextel networks — won’t come online until later this year.

Second, Spark isn’t splicing together those bands (a technique known as carrier aggregation) it’s just running them as separate networks similar to the frequency patchwork AT&T is stitching together. Devices either access one network or the other, not both simultaneously.

Finally, Sprint is advertising 50 Mbps speeds on Spark, but those are really peak speeds. When pressed, Sprint said it expects average speeds of 5-12 Mbps. Spark is basically an average network by U.S. standards, and that average network is available only in 12 cities. The primary LTE network Sprint is building uses only 10MHz of spectrum, supporting half the capacity of any of its competitors, and it’s not even nationwide (Sprint has yet to launch LTE in San Francisco).

The sad thing is Sprint has the spectrum resources to build the biggest, baddest LTE network in the country. It just refuses to do so. As of now, Sprint’s mobile broadband service is the slowest in the country, it has the least amount of capacity and it has the smallest coverage footprint. Maybe one day Sprint will truly live up to it’s 4G potential, but I doubt its customers will wait that long.

LTE image courtesy of Shutterstock user Inq; Frankenstein photo courtesy of Flickr user annabellaphoto

  1. “The sad thing is Sprint has the spectrum resources to build the biggest, baddest LTE network in the country. It just refuses to do so.”

    Are they refusing, or do they simply lack the money to do it? The level of investment required is certainly not cheap.

    Reply Share
    1. They’ve got the money now that they were bought but SoftBank so they are refusing to do so.

      It really is impressive that T-Mobile has absolutely leapfrogged Sprint in LTE coverage and speeds. Two years ago they looked like a limp, dead fish and now they are stirring up the industry. No way the FCC or DoJ allows a merger.

      Reply Share
      1. They just got bought, I would imagine you can’t upgrade your network in a few months.

        Reply Share
    2. Sprint’s just now completing NV 1.0, which around ~33K sites finished by now, this is LTE 1900 rollout. Sprint will be starting to roll out LTE on bands 800/2600 later this year, and they’re not doing it in a slow manner – it takes time to completely REBUILD a network. T-Mobile/Verizon are simply adding overlays and that is why their rollouts are going faster. Sprint is completely removing all equipment at the sites and adding new equipment. It also takes time to refarm the Nextel Spectrum.

      Reply Share
    3. It’s already in the works. Once they roll out the 20 MHz TDD in the first phase of Spark, all they have to do is add an additional carrier card to each station. Further, they will be aggregating the TDD chunks of the 2500/2600 MHz so instead of 3 separate 20 MHz channels there will be a 60 MHz channel available in a 3:2 download:upload ratio. There will be some overhead due to the aggregation, so it won’t be a literal tripling of speeds, but it will be impressive.

      Reply Share
      1. I believes these are all still big “if”s. Sprint has been talking about doing the same thing all the way back to 2007 since the early days of WiMAX. Sprint has always hedged its bets, but at certain point you’re no longer hedging, you’re just static.

        Do you guys really believe, getting 100 markets covered with TD-LTE by the end of 2015 is really aggressive? I’m sorry but that is a laughably slow pace that to me says Sprint is giving itself plenty of wiggle room in case some other option comes up.

        Reply Share
        1. Hi Kevin,

          I thought I’d point you to this post over at S4GRU. It’s very insightful as to how coverage actually stacks up in the U.S. It came from Sensorly, which is a website/app that generates coverage maps using crowd-sourced data. If you don’t have time to check out the link, in short it basically shows that announced coverage and actual coverage aren’t equal. Verizon over exaggerates their maps a lot. Users on Verizon as well as, Sensorly prove this. AT&T and T-Mobile are playing a numbers game. They boost numbers in order to make their networks seem larger. And Sprint’s network is just as large as AT&T’s in real world use.

          Here is the link: http://s4gru.com/index.php?/topic/5456-factual-coverage-comparison-verizons-lte-comparison-but-using-sensorly-data/

          Reply Share
        2. lickemlowbr78 Sunday, March 9, 2014

          This information was very very helpful I’m glad i have version and sprints sucks i had ta learn that first hand

          Reply Share
  2. While I have recently switched to Sprint and hate it. I can attest that its 4g speeds are faster where I live than Verizon. But that is when the 4g tower isn’t down. The only thing keeping me happy with sprint is it is much much cheaper than verizon or att and I get unlimited everything.

    Reply Share
    1. When the 4g tower isn’t down? Once they are launched, they don’t just take it off for fun. If you have a down tower, report it to customer service and techs can look at it for you.

      Reply Share
  3. Verizon LTE where I live (Columbia, SC) is terrible, its terribly overcongested. When I first got LTE i was getting 28M down/14M up. Now, I cant even get 2M/1M. It’s slower than ATT 3G network that I left 2 years ago. Pathetic.

    Reply Share
    1. Feeling your pain, Derek. Same in San Francisco: VRZ was super fast when I switched from Sprint 18 months ago, now it’s frustratingly slow. It’s borderline false advertising…

      And then one day I was skiing at Squaw Valley and tested VRZ: 45 MBps. Wow! If only we could get that in the city…

      Reply Share
  4. Trust me I know what I’m talking about. Sprint has no chance of becoming anything larger than it has already in regards to the acquisition of T-Mobile and AWS LTE spectrum. I work as an contractor for erricson and we deploy cell sites. Recently T-Mobile has been requesting we install spark contingency relays , which points to combining the two. Leak alert: that’s the definition of Spark. Think.

    -Milan from howardforums

    Reply Share
    1. If you’re asking us to believe you learn how to spell Ericsson properly…

      Reply Share
  5. Frank:
    Several facts you have are wrong on Sprint:
    Carrier Aggregation is part of their solution with the 2.5 GHz frequency only
    The Medium speed according to Sprint is 12 -16 Mbps
    I know Korea Telecom already has “channel” bond multiple frequencies on the higher RF spectrum, do you not think that SoftBank, who gleams its technologies from Korea and Japan are already 100 steps ahead of the European and Americas’ market. When Mr. Son mentioned that the US’s network was slow, he knew things that we are just finding out recently but what his home and ancestral countries have been employing for a long, long time.
    IF Sprint employs the same technologies as the above with the spectrum they have, they will be hard to beat (NOT number one, but a distinct threat for At&T for number 2…Verizon right now is untouchable). IF Sprint buys out T-Mobile (which in my opinion has a 0.05% of approval), Sprint will have to divest its higher RF spectrum to Verizon and AT&T and will lose the advantage. Sprint needs to forget T-Mobile and concentrate on its problems right now which will take about 3 years.

    Reply Share
    1. I meant to address the above reply to Kevin Fitchard, not Frank.

      Reply Share
    2. Yeah, I feel for you Kareny. The coverage thing is what flabbergasts me the most. I mean things like speeds, frequencies and capacity don’t matter to a lot of people, but simply ignoring places, like SF, Seattle, Denver, Minneapolis and Washington — that makes no sense. This is age of mobile data and the smartphone right? Not having LTE in SF is like not selling gasoline in Detroit.

      Reply Share
  6. Daniel Quintero Friday, January 31, 2014

    How is sprint #3

    Reply Share
  7. Sprint will be doing carrier aggregation between all of their LTE bands as soon as LTE-Adwanced becomes a finished product. It needs to be finalized first before they can get the software to upgrade all of the towers from their equipment vendors/manufacturers.

    Reply Share
  8. Kevin Fitchard, you are correct about AT&T which is why I left AT&T years ago. You are also correct about Sprint. Sprint indicated to me when I obtained cellphone service in 2011 that 4g LTE would be launch in San Francisco in a year. At the end of 2012, Sprint indicated that 4g LTE would be launch in San Francisco by the summer of 2013. Then I was told by the end of 2013. Then I read an article in which Sprint’s CEO indicated that Sprint would launch 4g LTE in San Francisco by the end of June, 2014. I gave up waiting & left Sprint for Verizon. Sprint needs to get a clue. Why should its customers pay for a 4g LTE smartphone & 4g LTE that they are not receiving?

    Reply Share
    1. Yeah, I feel for you Kareny. The coverage thing is what flabbergasts me the most. I mean things like speeds, frequencies and capacity don’t matter to a lot of people, but simply ignoring places, like SF, Seattle, Denver, Minneapolis and Washington — that makes no sense. This is age of mobile data and the smartphone right? Not having LTE in SF is like not selling gasoline in Detroit.

      Reply Share
  9. When I was out in SF for Google i/o last year, I could swear there was LTE with Sprint. Maybe just near Moscone? Sprints 3g is horrendously slow and don’t remember that experience when I was out there.

    Reply Share
    1. Sprint had decent coverage at I/O. I was there as well. I’m sure you recall how most networks (including the wifi at Moscone) were horribly slow for the first day or two.

      Reply Share
  10. Nicholas White Tuesday, February 4, 2014

    I’m not sure what you all are talking about when you state that Sprint has no LTE in SF – maybe they just haven’t “officially” launched it yet? I moved to Massachusetts from San Francisco in mid-2013, and I had had LTE first on my HTC EVO, and then on my HTC One, for several months by the time I left. My partner’s iPhone 5 also received LTE coverage. Maybe Sprint hasn’t gotten to the point where they are satisfied enough with the network to officially issue a press release launching it, but believe me, LTE is up and functional in San Francisco.

    Is that absurd and shameful? Yes. But it is factually incorrect to state that Sprint does not have LTE service in SF.

    Reply Share
    1. htc evo never had lte radios.. only wimax fake 4g.

      Reply Share