Hacker News new | past | comments | ask | show | jobs | submit login
LTE Has Slowed by 50% in the US This Year (twinprime.com)
492 points by wkoszek on Oct 19, 2016 | hide | past | web | favorite | 159 comments

A quick reminder that 4G != LTE. The 4G specification requires a minimum speed [1] so LTE was launched to avoid exactly this minimum. It seems that the companies did it right by launching LTE instead of 4G as they could have lost their 4G status, while now they could drop as low as 3G speeds and still be called LTE (which is ironic on itself).

This wasn't commented at all in the article, using 4G and LTE interchangeably which I find troubling.

[1] 100Mbit/s for high-speed transit areas and 1GBit/s for low-speed transit areas, https://en.wikipedia.org/wiki/4G#Technical_understanding

My go-to reference on the situation is https://blogs.oracle.com/ksplice/entry/essay_3g_and_me.

Like Keith says, more important than whether you have 3G, 4G, or LTE is:

""How much spectrum has the carrier licensed in my city, and how much is allocated to this kind of modulation?

How many other people am I sharing the local tower with? In other words, how big is my cell, and how many towers has the carrier built or contracted with?

How much throughput are my cellmates trying to consume?

How much throughput has the carrier built in its back-end network connecting to the tower?

You might notice that all of these meat-and-potatoes factors involve the carrier spending money, and they all involve gradual improvement in behind-the-scenes infrastructure that's hard to get customers excited. Persuading you to buy a new cell phone with a sophisticated modem and sign up for a two-year contract is a different story. So they don't sell you something measurable where they could be held accountable; they sell how sweet it feels to be using a sophisticated radio modem protocol to talk to them.

If the carrier sold you "384 kbps Internet access anywhere in the coverage area, outdoors," that would be something you could hold them accountable for. The carrier might even have to put a brake on signing up new customers until it could build new towers or license more spectrum for everybody to share, if it made that guarantee.""

I don't buy it. My carrier (AT&T) is almost certainly rate-limiting bandwidth to phones in my area, doing deeper inspection on their transparent proxies, or both.

In 2014, as measured by the Ookla app on an iPhone 6, I was routinely 50/17 and 60/20 speedtest to various servers in the region during business hours at home and work. I know this because I took a screenshot in disbelief.

Today, service varies from 1.5/1 to 10/3 during the day, and 10-15/3 at 5AM. I don't buy that usage has increased that much.

Which market?

In the case of the NYC market, AT&T had the worst spectrum planning of any major carrier. They are suffering because they chose to rely on low-band spectrum rather than densifying their network, and are now trying to band-aid the solution by using carrier aggregation of oddball downlink bands. It doesn't change the core problem of having too many people on each sector.

The reason they have to stake 2G dead this early is because they are spectrum starved and already only running it in HSPA guard bands. They need every resource they can get since they haven't been spending money at throwing towers in every possible place in the city for years. They've been using it to buy a satellite company, a Mexican carrier, and WCS spectrum that's worthless without densification. Priorities.

>> In the case of the NYC market, AT&T had the worst spectrum planning of any major carrier.

They've been having a hard time since the late 1990's when they introduced the "one rate" plans and basically oversold their coverage area in NYC.

I distinctly remembering long articles talking about how the networks were so saturated, dealers were still activating phones on their network knowing the customers wouldn't get a signal and would have zero reception in NYC, even after being told by several courts to stop signing up customers.

Here is the text of the class action lawsuit that made it all the way the NYC Supreme Court: http://www.whafh.com/modules/case/docs/2556_cid_3_AT&T%20Cel...

EDIT: some of the articles I alluded to are listed in the class action starting on page 9.

"all the way the NYC Supreme Court"

In New York, the "Supreme Court" is the trial-level court. Above that is the "Supreme Court, Appellate Division" and above that is the "Court of Appeals".

It's been a long time since buying and erecting towers was part of the carrier zeitgeist. I appreciate your frustration- but I don't think your understanding of the fundamentals of how carrier networks operate reflects reality.

You might find this article helpful: http://www.cell-tower-leases.com/Cell-Tower-Lease-Rates.html

For rural markets your logic applies, they simply don't need many new macro sites. The biggest issue in these markets is getting decent backhaul to the towers.

But that logic does not work in an urban market. Which is where the speed complaints are coming from. The other carriers are all beating the drums on densification and focusing on building out DAS systems, what's AT&T doing?

AT&T backed off their densification commitment after their Leap purchase by arguing having more macro cells and spectrum works. The complaints and speedtests speak for themselves.

See this article: http://www.fiercewireless.com/wireless/at-t-drops-goal-deplo...

Good counterpoint.

Upstate NY, mostly Albany area.

i've had phone (3g unlimited[0]); business DSL (connected 0.4miles from the trunk!); and Cable service from AT&T.

I can say with 100% of confidence, they oversell like crazy and are limited on both wireless spectrum as their pipes capabilities.

Even if they have much better coverage than Tmobile where i live and work, i will still endure it and not give them my service anymore.

[0] remember they charged extra if you had an iphone? ha! always used android and nokia. Also of note, they managed to overcharge my all-unlimited account for $800 above the contract for two consecutive months.

Netflix, HBO go, those sports streaming things, increased proliferation of smartphones, Pokemon go, web pages are continually getting heavier.

Seems reasonable to me. 2 years is a long time to be comparing against. Your carrier should be upgrading their infrastructure, but if they are not these numbers seem ok.

Our local time warner affiliate has been awesome in contrast, and their network management is awful.

Seems weird.

Mobile data usage has gone up a LOT in two years.

I've never known a carrier (or isp) to offer any speed without a "*experience may vary" type disclaimer. Is this different in the US?

As a rule all consumer services are on a "best effort" basis. Service with guaranteed performance exists, but is marketed exclusively to businesses and is much more expensive. It turns out that providing guaranteed service costs a lot more even if the end result is the same 90% of the time, that last 10% is a real margin killer. Consumers would generally rather accept the occasional less-than-stellar service than pay nearly twice as much for it.

Yup, you'll see this if you've worked with datacenters too, you can get a 100mbit connection on a gigabit link on which you mostly get gigabit connectivity, but only 100mbit of it is guaranteed for much, much cheaper than you can get a guaranteed gigabit link. And in my experience (for personal and small business purposes at least) that guarantee is rarely worth its name.

The end result, more often than not, is 10%. At this point, it has nothing to do with industrial guaranteed service levels, it's marketing X and delivering Y<<X.

Who cares about the price, as a customer I want to at least be able to know what I'm buying before committing 6, 12 months. There can be no competition or progress when we continue allowing ISPs to promise theoretical bandwidths that they don't come close to, ever.

> Who cares about the price?

I do.

Nope. Physics is the same everywhere.

And conveniently, from a user's perspective, damn near impossible to discern from poor network management.

You either meant inconveniently or from the ISPs perspective.

"And conveniently [for the ISPs], from a user's perspective, damn near impossible to discern from poor network management"

pdpi for the win. (And ambiguous phrasing for the loss)

The speed of light in vacuum is a constant, but depending on the medium (ie., frequent scattering), the time it will take for a signal to reach from point A to point B will be much more than the time it takes for a light wave to cover the same distance in vacuum.

God, no. I think this is one of those "would be nice if" hypotheticals.

Carrier frequency as well. Depending on carrier, rollout status, and device, you can have a vastly different experience in a given area indoors vs outdoors.

Seems like it would be much more expensive than the elastic demand model used now unless you used the guarantee.

The problem with the "elastic demand model" (or the "sell, don't deliver" model) is that it's not working out in the marketplace.

ISPs, instead of expanding network capacity when they lose their side of the "elastic demand" bet, throttle companies like Netflix or extort them into financing their lacking capacity.

That's not really true. Wireless carriers are expanding as quickly as they can. The problem is they're out of bandwidth, so the only way to expand capacity is to shrink the cell size.

And you can't just throw up hardware wherever you want. The bureaucracy involved (from the FAA to the FCC to local planning boards to code inspections to historical societies to nimby obstruction) in getting a single site up is staggering. Those sites are going up, but it takes time.

I wouldn't really consider selling a service to be extortion.

> using 4G and LTE interchangeably which I find troubling

I think in Europe HSDPA, HSUPA and HSPA+ were more consistently called 3.5G, 3.75G and perhaps 3.9G, so 4G and LTE can be synonyms.

In America, operators are marketing HSPA+ as 4G, so then there is a need to specify when 4G is LTE and when not.

Exactly. I can confirm this is the case in Singapore too. An HSPA+ network is labeled as 3G on the iPhone. In the US, when I see 4G on my phone I know it's going to be slow.

I call that FauxG! ;)

At least in the nordics the system seems to be something like this:

1G = NMT

2G = GSM


4G = LTE

Which is also what the standard bodies, mainly 3GPP, that produces all the technical documents for these systems (architecture, protocol specs., functional behavior, frequencies, coding schemes, security algorithms and so on) calls their technology.

> using 4G and LTE interchangeably which I find troubling

It follows the colloquial usage though.

> It follows the colloquial usage though.

....set by Marketers working for Cell phone companies who are very happy to trick you into thinking you have a faster service than you do.

Source: Worked for a cell phone company that was ecstatic to roll out new towers and tout "4G for all". They were LTE towers, with a back haul incapable of 3G speeds.

> The 4G specification requires a minimum speed [1] so LTE was launched to avoid exactly this minimum.

LTE is a name for a specific connectivity technology, it's a parallel to UMTS in the previous generation, not to 4G or 3G. It was going to be launched independent of whether it met the 4G minimum criteria. Standards-wise, LTE predates 4G. It initially looked like WiMAX might reach the 4G criteria first, but it tanked and then we were left with waiting for LTE-Advanced.

Traditionally the way that carriers deal with bandwidth congestion is to wait until people start screaming (and networks start breaking) before they invest in innovation. There are a bunch of technologies that could ease congestion and deliver significantly better wireless performance but they would require an investment that doesn't make sense for carriers (it's not like carriers can extract more money from you if the network is better...).

That is to say, subscriber ARPU does not increase with network investment, so why invest in the network until it becomes a drag on subscriber growth?

Source: I was a manager at ATT when the network in San Francisco basically died with the introduction of the iPhone 3G. It stayed that way until ATT added new towers and upgraded the software on the towers for better spectrum utilization.

Living in SF, I haven't really noticed this, which the article bears out. I was actually just commenting in one of our Slack channels at $work that it's still kinda weird to me that the internets are, on average, at least 4x faster (throughput, not latency) on my phone vs my home internet service (bonded DSL).

EDIT: Out of curiosity, I just checked again, first on LTE and then on WiFi:

  LTE:   30ms ping  64mbit/s down  23mbit/s up
  WiFi:  24ms ping  6mbit/s down   2mbit/s up

There are a few high-rises in SF with gigabit fiber optic to the premises. It's nice, but it makes me sad that "anti-techie" sentiment is so strong that most of the city won't get it.

Srs. My ISP is Sonic, and I'm happy to give them money, because they give me every reason to believe they share values that are important to me, but god, I wish the NIMBY would get out of the way of them rolling out FTTP city-wide.

I suspect it's the complete lack of actual cable vaults beneath the streets in suburbia that is the issue here. Those things should be /required/ by regulation.

Actually there should be three isolated types:


Clean flows (water in, gas, telecom/data)

Outflows . . .

Why the wired Internet is so bad in SF is another question entirely (one I'd like to have the answer to, because I'm so sick of my AT&T copper based DSL, but my hatred for Comcast is very strong due to their billing practices ).

DSL sucks. I'm sure our local ATT has lost the vast majority of it's DSL users in the past few years because of rollout problems. Their fast connection is around 20mbit/s download and around 2mbit/s up. Our cable company is offering 50mbit/s down 5mbit/s up for the same price. The cable company is also offering 100mbit, 200mbit, 500mbit, and 1gbit connections too. On top of that they offer phone service at half the price.

I think ATT is trying to go out of business here.

ATT also has data caps (now 1TB and $10 per 150GB), compared with the cable companies (in my area) and google which don't. That doesn't seem to stop about 1/2 of my neighbors from using them though (you can usually tell those people because they leave the default ATT SSIDs on their wifi APs).

Once a year or so, it seems they hire some kids to come through the neighborhood selling service door to door. Every time I ask them, do you still have caps? Usually they stare at me blankly, but I then politely tell them I'm not even going to consider their service as long as I'm going to have to worry about a internet bill that could be larger in a single month than what I pay all year for my cable subscription (which literally got 10x better within a couple months of google announcing they were going to target my city).

Comcast just implemented 1TB bandwidth caps in most of the U.S., including where I am, starting on Nov 1.

DSL sucks, but it's not Comcast, which is my only other alternative. I'm just out of LoS for Monkeybrains, and my building is both too small and too old for any of the fancy options.

Complete speculation, but I was told that ATT is getting out of DSL and is working to lay fiber everywhere as a replacement (they are putting it in my neighborhood now). I currently have Comcast 125mbit/s down 25mbit/s up, and I have no complaints. BTW, the best way to deal with Comcast support if you ever need it is through twitter.

Does not surprise me. I switched my internet contract to unlimited LTE here in Austria and it has been great. Would do again.

(Building here has fiber though so I might switch if it turns out to be an issue)

Most tech journalists (the Verge, Gizmodo, that sort of thing) live in either San Francisco or New York, as do the kinds of techies who would raise a big stink over exactly this issue, so carriers have a strong incentive to keep performance there, and only there, high.

Mobile networks in a way are seemingly destined to be victims of their own success. I find that no matter what mobile bandwidth I'm getting, I can always use more. For example, considering adding a dedicated hotspot to my existing plan, just for my car. The better it works, the more I want to use it. And by "it", we are talking about a fixed physical infrastructure, otherwise known as a capital investment.

If the US made any sense, this would be a public works infrastructure project rather than private enterprise. The tower density and spectrum that will be needed for effective substitution of traditional wired internet services with wireless towers will be prohibitive for any reasonable company to simply invest their way into, and big teleco has proven historically to love charging more for the same now than competing to make the service better once it starts looking expensive to keep up.

> If the US made any sense, this would be a public works infrastructure project rather than private enterprise.

No thank you. My Verizon plan is expensive, but it works pretty damn well, unlike most of the public services here. My wife and I took Amtrak every day between Wilmington-Baltimore and Baltimore-DC for about two years. It was regularly late (often very late). Trains broke down in-between Wilmington and Baltimore regularly, and my wife would be stuck in the train for hours waiting for a replacement engine to arrive. Many of the tunnels along the route are decades past their design life.

Before that, I lived in Chicago, where ancient lead water pipes poison kids. Before that I lived in Atlanta, where ancient sewer systems dump untreated sewage into the river whenever it rains. I also lived in Wilmington, DE, where bus drivers would just randomly decide to end their route 15-30 minutes early. Now I live in D.C., and even though I'm within 0.3 miles of a metro station both at work and at home, I take Uber because Metro trains keep breaking down and/or catching on fire.

I'm not a libertarian nutjob, but I don't want the government running my cell service. Our roads are awful (I just came back from Munich), our transit is awful, etc. There are a tiny handful of competent public infrastructure organizations in the country (e.g. New York's water system, Metro North), but most are a disaster. Verizon may be evil, but unlike say WMATA it doesn't have to regularly shut down major sections of its network because it spent decades neglecting its infrastructure.

Amtrak does not work not because it's a public service. Many railways in Europe are public and the work great.

Many public services in Europe work great. But I'm talking about the US.

Amtrak being late wasn't entirely their fault, The railroads that owned the tracks often delayed the passenger service giving preference to their freight service. They got dinged for that not too long ago.

Wilmington to DC is northeast corridor, which is all owned by Amtrak. That's why delays are usually only tens of minutes rather than hours (except when equipment breaks down and the delay is hours).

> The tower density and spectrum that will be needed for effective substitution of traditional wired internet services with wireless towers will be prohibitive for any reasonable company to simply invest their way into

It will be prohibitive for the government too, so lets just stick to wires.

Let's unpack this. If this is done at the Federal level, ala the Interstate Highway system, it would never work. Wireless infrastructure is still incredibly immature compared to road technology. And the idea that the government would be able to spend the amount of money required to keep up to provide the same or better service than the private sector has never been proven, and IMHO unlikely. Look at how NASA has been run the last 30 years, and how it's been spending its money on their current launch system.

The problem is that the capital expenses are high in this industry, so the players naturally want to recoup their costs and focus on profits. Eventually technology will shift, demand will rise, and one of the telcos will take the plunge in an attempt to gain marketshare. Their competitors will react, and for a while it'll be golden for the users. Then the cycle will repeat.

let's unpack the unpack.

> Wireless infrastructure is immature and [the government can't spend].

The exact same thing could be said for the computer industry in the 40s and 50s, which was almost entirely funded by the government, precisely because only it could spend enough money for big risk long term stuff. Ditto the internet itself. Ditto the airplane. Therefore there is ample evidence that the government can spend, and spend big, on strategic stuff, and there is ample evidence that, over decades, it pays off bigtime.

> capital expenses are high, recoup etc

That's why you want the government in, because it has a much longer investment recoup horizon, and the government, unlike private networks, benefits from the general indirect economic growth advantages of good mobile infrastructure. That's precisely why it is the government that builds roads, often at a project-specific NPV loss, but at a huge NPV positive for the economy, which it taxes. In other words, only the government is able to capture the collateral positives of good infrastructure, giving it a higher incentive to spend more than private companies.

> competitive cycle rinse repeat etc

Quarterly targets are a strong disincentive to long term capital intensive projects. Most such projects undertaken by the private sector have, at a minimum, government concessions to allow restriction of competition, pricing guarantees, or outright government financial support. Notice the whole thing where municipalities cannot expand their internet service? There you go.

In general the whole dogma that the government is always useless at everything, is very pre-2008, ie. dangerously incorrect.

Not sure about the downvotes, unless it's because of the new HN motto where everyone downvotes things they disagree with.

<In general the whole dogma that the government is always useless at everything, is very pre-2008, ie. dangerously incorrect.>

You set up a good strawman. And the implication that since 2008 (what exactly happened there?) we're somehow more capable of having well run, long term, government projects in the US is not supported by much evidence.

Yes, government can have a longer recoup horizon; but it also has an extremely short political horizon. If something doesn't succeed on the first attempt, it becomes fodder for whichever party is in power and supported it.

And while government has had successes in the past, as the investment folks remind us, is no guarantee of future successes, especially in something as complicated as creating a continent-wide wireless infrastructure. Imagine the cost, the rent-seeking, the cronyism, the NIMBYism that will infect such a project.

Government is not an agency like DARPA that is free to investigate and develop technologies without concern for budgets and satisfying constituents. Look at how politicized the FCC is. Imagine handing it perhaps the largest communication system in the world, and expecting it to manage it. Keep it upgraded. Understand which technologies are worth something, and ones which are dead ends. Oh, and make sure Grandma in Broken Bow, Nebraska has as good a connection as someone in Berkeley or San Diego.

And back to your strawman; show me large Federally run government programs implemented since 2008 that are on budget, meeting original goals, and not having huge unintended side effects.

You don't even need a wholly public-only solution. Licensing out state-owned towers to third parties, in addition to providing baseline public service, would be adequate to allow third party companies to compete too.

You don't need to try to artificially restrict spectrum use, just make intentional obstruction illegal just like blocking a road is. Then if state technology calls too far behind due to budget cuts people move back to private solutions until the state can get its act together again.

Of course, the problems with US infrastructure run a lot deeper than that. I imagine in a more democratic country you would see the maintenance and upkeep of such infrastructure much easier to maintain.

If you want to throw government agencies around: The NSA seems to be doing a fine job keeping up with current technologies.

Possibly. Other than Snowden releases, I don't know much about the inner workings or capabilities of the NSA to judge them. But consider the budget that the NSA has to work with.

Excellent timing considering the HN thread [1] about T-Mobile being fined for network management.

"There is no doubt that the US will need to set up the infrastructure to keep pace with the rapid changes in usage and content expected in the future. Like any instance of supply and demand, we will continue to see a give and take in this market. As operators catch up to the current demand and LTE becomes faster, users will opt to use it over others – thus creating greater demand, supply scarcity, and decreased performance. At which point the cycle will begin again."

TL;DR Expect more network management in the future due to heavy demand of a constrained resource.

[1] https://news.ycombinator.com/item?id=12745255

>There is no doubt that the US will need to set up the infrastructure to keep pace with the rapid changes in usage and content expected in the future.

I feel like we've been here before. It's almost like it's 1996 again with the Telecommunications Act. Our infrastructure is bullshit and it's making the US economy fall behind its peers. No worries! Instead of actually directly fixing the problem, let's give away all the power to corporations to fix as long as they super-pinky-swear to meet infrastructure goals.

Of course we all know how this ends - they didn't meet those goals, service stagnated even more, there was never any recourse, and we've axed the option to directly fix the problem. How convenient for ISP stakeholders in 1996.

*fined for misleading customers about network management

Your average user doesn't care, as long as they're not charged extra for exceeding their quota.

The quotas on wireless ISPs are already very low and people exceed them very regularly. What can be done to actually improve the situation? What choice do consumers have?

But consumers do have a choice, at least in the US. T-Mobile offers unlimited data plans, and although they were just fined by the FCC for deprioritizing heavy data users in cases of congestion, that's still a significant improvement over standard data caps. (At least T-Mobile never did what AT&T used to do before they were sued, throttling users who weren't connected to congested towers.) Some web searching shows that Sprint just recently started up a similar offering. Sure, the more favorable terms come at the expense of network quality compared to Verizon or AT&T, but they're not that bad and that's the nature of competition...

They'll take my grand-fathered-in unlimited plan from my cold, dead hands...

Actually one more contract-violating rate increase and I'll probably be forced to give it up. The only way I still justify it is everyone is paying an arm and a leg to Verizonopoly in my area.

Voting for representatives that actually represent people and not corporations.

Which means voting for people supporting election financing reform.

I had this thought earlier: how about just electing people who don't give in to corporate lobbying. Keep corporate donations legal, I don't care, just actually look at the disclosure forms of where their money is coming from and don't vote for them if you disagree with conflicts of interest. Or if they have PACs that aren't transparent, don't vote for them.

People are unable to make their voting decisions outside of the TV commercials and news articles they read, so I don't have any real hope for this tactic.

...due to heavy demand of an artificially constrained resource.

If this worked the way free-market idolizers like to pretend there'd be competitors driving prices down and features/capabilities up. But none of that is happening.

Yeah. Saw this other article in the front page of HN and it inspired me to post this one too.

I think that control of content is one of the major reasons for this. If users were more able to readily (and for zero cost to them) cache content when connected to local networks then we would see less content transferred over 'higher cost' networks.

Of course streaming services (I'm thinking more of Twitch than Netflix) for live content production are 'rather difficult to cache' in their prime viewing time.

The thing most difficult to cache is the thing that all the providers most want: advertising.

If you removed the advertising from the web, caching gets really easy except for something like Twitch.

Anecdotally, I've been incredibly disappointed in LTE speeds for NYC lately. It's almost a joke how slow LTE is. Browsing the web feels like using ancient DSL.

There's definitely a material difference between providers. I'm on Verizon now, but T-Mobile and AT&T were both much better when I had them (and I'll be switching back as soon as I can).

Last Friday I couldn't even get a page to load on my iPhone with "full bars" on Verizon. Where does all the money go towards? Not fios, that's for sure (Verizon defaulted on a deal with the city to roll out fiber to all households[0]).


Since joining Verizon I've realized that bars are more or less meaningless. There are many times when I can't even make a phone call with "full bars."

Sadly, I've also been very disappointed with FiOS speeds lately.

Verizon's claims are pure BS. When I did some work for a company called CrunchButton, I figured out (using my fiance's phone) that Verizon had towers that weren't connected to anything. They were just there to provide you with 'full signal' (your phone never connected to these towers) so they could claim coverage in that area. You couldn't make a phone call from that area on the Verizon network. Meanwhile, my T-Mobile phone showed three bars and worked flawlessly.

Don't trust Verizon. They're a sham company.

Would you elaborate?

Verizon seems to think it's a better investment to spend money on ads saying their network is good, than actually spending money on making the network good.

I've been on Verizon 4 years currently with iPhone7+ using iPhone the entire 4 years living and working in Manhattan.

With the carrier aggregation now turned on I can get as much as 100 DL / 5 UL. They own the most LTE spectrum.

Verizon does have the best network NYC.

I completely disagree. I'm a Verizon user, my working partner is on T-mobile, and I regularly benchmark out of curiosity. It's not even close, in any neighborhood.

Since I demonstrated 100 Bit/sec DL, I assume you friend exceeds that on T-mobile?

Have you tested signal quality in large buildings in Manhattan?

I can say I get signals in these buildings when iPhone owners using say, AT&T get no signal.

I suspect that much of the difference you see might have something to do with your handset.

This report demonstrates that Verizon has the best network. http://cellularinsights.com/state-of-lte-advanced/

Lol, that report was written by a T-Mobile employee: https://www.reddit.com/user/milan03 . All it stated was Verizon has been active on densifying their network and is purchasing 4x4 equipment. Verizon does own the most spectrum in NYC, but they have not been leading on LTE-Advanced feature rollout.

T-Mobile already had that equipment deployed for over a year in some markets. They also are the only ones running 256QAM DL 64QAM across their entire LTE footprint (though you'd need a phone supporting it) increasing the spectral efficiency 15-20%.

If you can tolerate the boasting typical of a press release this is the CTO of T-Mobile discussing what they have already rolled out: https://newsroom.t-mobile.com/news-and-blogs/lte-advanced.ht...

I work at a quite large telecom company and one of the most repeated topics is that mobile traffic has been doubling every year for the last years. The trend is going to keep growing even more.

The nature of mobile networks is being a shared resource, as opposed to traditional DSL or Fiber which have a generally more dedicated bandwith.

This obviously implies quite a challenge for telcos, as expanding the network comes at a massive cost.

> With the onset of functionality like 4K video streaming, this number is set to increase to as much as 22GB/month.

Oh god why would anyone want to watch 4K on a cellphone. Go for 60fps instead, you'll get some value out of that on your 5" screen.

That's what I thought too. 4K doesn't make any sense on that small a screen.

I think streaming VR content is what is going to contribute to a much larger data usage.

I've noticed it big time in the northeast. I live on the road and work via a Verizon connection. Over the last 6 months over 7 or 8 locations I get a full Verizon signal (with a booster) and very low speeds compared to a year ago. And speeds increasing at off peak times (it's fast in the middle of the night) point to overloaded towers.

I know people love to hate on cell companies but it must be hell to try and keep up with demand that changes so rapidly.

I suspect it's securing the rights to build towers, and the rights to install fiber to those towers. Actually building the thing shouldn't be much of a big deal. getting a permit to dig up a mile of street, that sounds like a pain.

Hey look, another example of why trying to sell the rights to light sucks.

We are going to see AT&T / Verizon / etc go the way of Comcast soon. The cost to improve service will be high enough and the overhead of trying to get more spectrum when they hit physical limits annoying enough and their revenues large enough and the demand insane enough they are going to constantly try to buy each other out than actually invest anything until we have one big corrupt mess like Comcast is for physical wire service.

It seems like the inevitable outcome of having infrastructure services that should be public utilities instead be provided by private companies competing over who can exploit the state to get more unfair advantage, be it land access rights for wire carriers or FCC bribing for spectrum.

Sad, because they're the best hope we have against Comcast - at least wireless providers (I'm stuck with DSL over AT&T copper because I absolutely refuse to deal with Comcast any further, and my phone is faster than my home internet).

At least it's relatively easier to set up a cell tower as a new competitor than it is to wire the same number of homes; this is a big factor in why we've seen the explosion of wireless in the developing world (often in places where even basic electric service is unreliable).

And hey, the feds DID block T-Mobile and AT&T from merging, so there's hope yet that they won't just roll over to every request.

"From early 2015 to early 2016 there was a 56% increase in data usage according to Cisco."

And there you see the problem with data caps (common among mobile carriers but swiftly coming to cable). We have plenty of bandwidth today and are squeezed for more money in a few years.

We don't have plenty of mobile bandwidth though. 56% increase resulted in 50% degradation of speed. What happens next year to speeds with continued growth in usage?

Good point that we need to think about. I'm more worried about caps in landlines. A similar problem exists there, however. Here's a good article on it.


What I don't get is: why can't I as an app developer specify what kind of data transfer rate I need and have the phone choose which connection type it needs depending on the currently running software?

Like, if I'm doing push notifications or IRC, I'd tell the phone that I only need 2G speeds, and the phone only connects to something faster than 2G if I open the web browser.

Right now, my phone books into LTE as soon as it's in coverage mode - and it stays there, eating power like nothing else, instead of dropping into the relatively quiet and strong-signal 2G/3G/HSxPA cells and saving power.

Your phone is actually far more battery-efficient when it's able to only use LTE (i.e., if your carrier supports Voice over LTE). It's having to have both radios on (one for voice and one for data) that is hard on the battery.

Not exactly. When it's on LTE and doesn't support voLTE, the signaling is used here to search, connect, and handoff to HSPA or GSM (or CDMA). It increases the call setup time, but it isn't actually a major factor in battery life.

GSM will win most in battery efficiency due to it being a TDMA based standard that pulses the radio off and on. But LTE has improved a long way since the early days.

UMTS and HSPA were fine on battery life and they were CDMA technologies. The big difference is the extra work basebands have to do to handle new modulation technologies as well as the extra CPU time it takes to handle more data. More efficient CMOS processes, baseband designs as well as ARM core designs have been the main cause behind power efficiency in LTE-enabled devices.

Maybe you remember a different history than I do. WCDMA/HSPA was poorly designed early on to tackle smartphones and had terrible battery life and performance.

To summarize pre-2012 HSPA:

- Basically, HSPA phones until around 2011 or so stayed in a higher power state much longer than they should have.

- People got pissy about having crap battery life.

- Manufacturers responded by doing proprietary hacks to send modem into lower power state.

- Cell networks essentially got DDOS'd as phones sent nonstop signaling to negotiate different power states.

- Phones now not only got poor battery life, but also barely operated on the now congested network.. remember the at&t iphone monopoly days before 2012 in major cities, it was a bad time.

3GPP Rel8 came about to fix the issue for good by moving power state control to the towers. Here's an article explaining the technicals of it: http://blog.3g4g.co.uk/2010/10/fast-dormancy-in-release-8.ht... Easier read: http://www.3glteinfo.com/fast-dormancy-in-3gpp/

Would agree though that process size and baseband optimization is helping the LTE situation. Differences are very small and Apple actually quotes better LTE life on the SE than 3G. LTE really came into its own in just the past 2 years.

FYI, AT&T will shut down their 2G network by the end of the year: https://www.att.com/esupport/article.html#!/wireless/KM10848...

Thank God I'm not in the US. I wonder what will happen to all the embedded stuff using low-power 2G chipsets because they transfer data via SMS or CSD/GPRS?

I'm especially thinking about car/truck theft systems. That's going to be expensive to retrofit these...

I work in the automotive industry and the 2G shutdown announcement caused a bit of an uproar for car dealerships that install GPS/starter interrupt devices in the vehicles they sell (for purposes of repossession). Most dealerships buy these devices in bulk, and as I recall from a conference I went to a year or two ago only one of the major providers was offering a partial discount on the price of more modern 3G units for those with soon to be obsolete 2G stockpiles.

Australia is also shutting down 2G next year. They're trying to free up spectrum for more 4G and possibly 5G

T-Mobile has committed to keeping their 2G network operational until 2020 for these M2M devices - and they support the same GSM bands as AT&T does. Unless these device manufacturers want to rush to put LTE radios in everything they'd be wise to just replace everything with T-Mobile SIM's and work on a better upgrade plan over the next 4 years.

I don't know how comprehesive this is, but clearly AT&T is not the only one (or US the only market affected): http://www.zdnet.com/article/optus-announces-2g-network-shut...

Currently seeing this with the Nissan Leaf. It has a 2G modem built in for their Carwings service. Nissan has sent out a letter that the car will need an update for the data features to keep working, but no details on cost or availability. My local dealer has no idea either.

Are you in the UK? We're installing "smart" gas and electricity meters, which use 2G, but industry analysts are predicting the cessation of 2G here by 2020. Awesome!

Maybe it's time for swappable wireless transceivers? Something along the lines of SFP?

Won't help much, because the antenna and coax cable has to meet the correct specs for the frequencies used by the transmitter.

T-Mobile isn't shutting this down anytime soon. Those customers will have to move there (same frequencies anyway).

At least from the operator's perspective it's more efficient if you are in LTE, as you occupy fewer time slots in the spectrum. To the extent that I've seen some operators in Europe offer cheaper data on faster technologies.

I think you're better off bursting the data with the highest speed possible and getting off the spectrum ASAP. Doing IRC on 2G speeds, you'll be transmitting or receiving a decent chunk of the time. With LTE, you'll quickly send or receive your data, then go back to leaving the capacity available for others.

This is part of the big drive to deploy "5G" network technologies in 2020. Better spectrum efficiency means you can support more customers doing data-intensive tasks with fewer cells, that was a big drive behind LTE as well.

With the bandwidth available to LTE connections already I think for the time being we're pretty much topped out on the bandwidth an individual terminal needs, just need to service more of them with less hardware.

If there are no incentives to get new hardware that supports technologies that service more people with less hardware, then deploying them will do nothing.

Because that would be a bad idea.

Why? The phone is idle 90% of the day and thus doesn't need high-bandwidth data in my pocket. In foreground any app could request high-bandwidth data - or, why not, high-bandwidth data while the screen is on and low-bandwidth when the device is in sleep mode.

Because with higher speeds, it returns to idle faster.

2G is also far less power efficient per byte.

The most efficient phone using today's technology would be a category 10 or higher LTE modem /w no 2/3G modems at all. Radio is on for a very short time, then switches back off.

Things have increased in speed so much that if power usage scaled linearly with capacity, using LTE would drain your battery in less than a second. Since they did not scale linearly, you are better off going faster. See "race to idle".

I remember getting on VZW LTE reasonably early, with the HTC Thunderbolt (don't even get me started; that device was trash) and I consistently got 60-80Mb down. Now I'm the same location, same carrier, infinitely faster LTE modem, I get maybe 5-10Mb if I'm lucky. Such a shame, could've been transformative.

I just tested my T-Mobile LTE using Ookla. I'm in suburban Boston. The local time was about 00:30.

Results: 69.39 down, 20.70 up, 36 ms ping (to Norwood). The test used about 170 megabytes of my 3-gigabyte monthly quota.

I went to a wedding in upstate New York last weekend, and I on the drive up, there were some areas where I had no 4G (or LTE, or whatever my phone gets), but anywhere that it was available, it was substantially faster than what I'm used to from the densely-populated areas where I spend most of my time. I assume this was because there were simply fewer people sharing approximately the same bandwidth.

Many people mistake the Gb/s numbers for these modem protocols as Gb/s per subscriber, when in reality it's more so the total throughput of a tower modem shared among subscribers.

Outrage of course, ensues.

This is an interesting read, but the comparisons to other countries/regions omits any mention of population density. It's much easier to roll out public utilities in dense areas than sparsely populated ones, and western Europe and Korea are more densely populated than the US.

Not that this excuses the big drop in speeds, but it makes the comparative piece a bit less relevant/accurate.

More population density means smaller cells, more towers, more infrastructure, higher level of wired network throughput surface density required, more money.

It might be economically unfeasible to service remote areas, but covering a given area is much harder as population density increases, not less hard.

Yet other countries can do while charging the consumer less.

I agree that there are different challenges to serving high-density areas, but I don't agree that these challenges are necessarily harder than creating a far-flung network. These articles [1] [2] indicate that for wired high-speed internet, population density is a good thing for cost savings.

To the extent that high-speed wired internet is part of the infrastructure needed for high-speed cellular networks, the cost benefit would similarly favor dense populations.

Tower density would have to be higher, of course, and interference could be an issue, but I've not seen anything indicating that these challenges outweigh the cost savings from having a dense population. I'd be interested in seeing an analysis!

1: http://www.fool.com/investing/general/2014/11/30/why-cant-th...

2: http://www.cnn.com/2010/TECH/03/31/broadband.south.korea/

Wireless technology has solved this problem since you no longer have to wire up each individual home. It's probably more difficult in dense populations because you need more towers and spectrum is limited.

Meanwhile here in Tokyo my mobile data connection is faster than the wifi running off of my 1 Gb/s internet connection...

I switched to tmobile $30 plan a year ago and regularly got 25mbps around Manhattan, now days I get around 10, often 5.

Yeah, their bandwidth decreased noticeably.

With their massive subscriber growth I'm not surprised, it seems a lot of the network investment they've been pushing has been extending coverage and in-building service with their Band 12 licensing. With them looking at deploying a "5G" network by 2020 I'm not sure if there are plans to add new cells right now, they have plenty of backhaul but the physics of shared RF spectrum is biting them now that they've had so many consecutive quarters of net subscriber adds.

Here's hoping they do invest in some more cells for my dense metro bretheren, because I enjoy 60Mbps at least day-to-day here in Boise :/

22gb per month expected with 4K? Lawl I was doing ~40gb per month when I first got an iPhone 5s on att. Good times.

Thanks for ruining unlimited data.

There is no purpose of marketing unlimited data if it can get "ruined" by people who use lots of data.

Anyway, why do you need unlimited data if you don't use a lot of data?

How can you ruin what never existed?

What is dead may never die.

40gb is still low. I've seen people that use terabytes on their phones.

Nah, i was on AnR from ATT, they paid for it themselves. nice try tho.

I want to downvote this comment.

Des the actual report cover whether there are any differences across carriers? I know in the 3G era there used to be fairly significant variations for the companies which installed newer base stations without upgrading their back-haul capacity to match.

Just did a test on my iPhone 7 showing 4G over EE in the UK... 18 Mbps down / 1 Mbps up

I've bet pretty big on this development (and its continuation) by building http://www.anyfinetworks.com. Gonna be very interesting how it plays out! :P

Yes the more people that use it slower it goes since people have to share the same bandwidth. However given all the datacaps the faster the speed the faster you hit the datacap so I guess you can look at the positive side.

Oh yeah! I was visiting NYC from Canada and I kept complaining to my wife that the LTE speeds were so slow in the city. I just couldn't understand it!

Downtown Montreal here, speed right now: 83 mbit/s down 13 mbit/s up

Quite good, almost as fast as my home connection.

"Network speeds are not what they advertise or what you see in the Bay Area"

So why are the companies in San Francisco not getting sued for false advertising?

This is a comment directly from the author to his technical audience. Worded awkwardly and combining two thoughts, he is saying, just because your app works well in SF, does not mean it works well everywhere.

It seems that the companies did it right by launching LTE instead of 4G as they could have lost their 4G status

I have Sprint LTE in Denver and the latency and bandwidth are horrible. I often get 1mbps with really high latency.

Have they deployed band 41 in the area yet? I see band 26 being saturated more then band 25, but they both are slow at times.

> Verizon has the broadest LTE coverage at 95.3%, followed by T-Mobile with 91.7%.

higher than in Germany, great!

"Kill the Snapchat"

So as more people start using LTE, 3G becomes faster?

gimme a good ol hard line

Applications are open for YC Winter 2020

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact