maxsilver

Kevin Fitchard came up with that number and has been quotes by other websites giving credit to Gigaom. Everybody else quotes 37,000 which lines up with what I remember.

I'm not sure why you keep posting incorrect information.

T-Mobile itself (and more reputable wireless news sources) are accurately reporting 51,000 towers for their entire network, and 37,000 sites for their 3G/4G HSPA+ network alone.

Just because "everyone quotes 37,000", doesn't make it true. It just means places like "Huffington Post" can't be bothered to fact check their articles.

Here's Neville Ray (CTO of T-Mobile) explaining how, after the T-Mobile + MetroPCS merger, the combined entity has roughly 60,000 cell sites (roughly 10,000 of which are planned to be dismantled, bringing them down to about 52,000)

Source :http://www.fiercewireless.com/story/t-mobile-expand-metropcs-footprint-100m-pops/2013-05-15

If you don't believe T-Mobile itself, or GigaOM, or Fierce Wireless, perhaps you'd believe Robert : http://s4gru.com/index.php?/topic/3450-tmobile-4g-lte-vs-sprint-4g-lte/page-17&do=findComment&comment=160656

You're new here so you haven't gotten used to AJ's tone.

He's not accusing you of lying.

He's expressing his skepticism at these numbers given his experience with cell networks and comparing Tmobile vs Sprint.

I didn't think he meant it against me.

It sounded like he was accusing *T-Mobile* of lying.

[deleted, website accidentally double-posted same message.]

No, there is absolutely confusion, considering that the supposed site density does not line up with T-Mobile's in market coverage problems.

It seems to line up to me. T-Mobile purposly and measurably underpowers their AWS sites (which you can measure easily by going to a market where T-Mobile and MetroPCS both use AWS, often on the same co-located towers). Again, I don't know why T-Mobile does this (faster speeds? reduced interference? just speculating), but I have measured them doing this.

Additionally, these numbers seem large. But AT&T and Verizon both have even more cell sites than T-Mobile does. (I don't have a source right offhand, but I believe AT&T has something like 65,000 or more sites in the US). Both of these providers also have low-band spectrum, that T-Mobile doesn't.

T-Mobile's towers are often cheaper than AT&T's, so that impacts it some. (In my market, AT&T rents the absolute best tower location, on expensive colocation sites. T-Mobile often has the same number of sites, but in less ideal locations on lower utility poles or building rooftops)

These things, in summary, are :

1) Reduced power output on 3G (AWS)

2) No low-band spectrum

3) Cheaper, sometimes less ideal locations

4) Still less total cell sites than some of their competitors.

Wouldn't all four of those seem to account for the discrepancy?

Math could still be fuzzy.  Otherwise, T-Mobile should be a world beater in urban areas.  But that is not the case.

I don't understand -- are you accusing them of lying about their site numbers?

The confusion comes from the fact that some articles seem to conflict.

It'd be nice to hear it straight from the horse's mouth i.e. an article quoting the CEO or CTO.

Your absolutely right. Here you go :

A T-Mobile press release, from T-Mobile corporate, still on the T-Mobile.com website, with the 51,000 figure included :

http://newsroom.t-mobile.com/phoenix.zhtml?c=251624&p=irol-newsArticle&ID=1804179&highlight=

The line is "[even though we're selling sites to Crown Castle] T-Mobile’s nationwide network remains unchanged today, consisting of approximately 51,000 cell sites, the vast majority of which are leased from third parties, as is common in the industry across the US."

That's from September 2012, a few months before the MetroPCS + T-Mobile announcement.

- - -

There is absolutely no confusion around this fact. T-Mobile realy does run 51,000 sites.

There's a lot of confusion, but pre-merger T-Mobile had approximately 37,000 sites. MetroPCS had 11,000 plus 6000 DAS. so together they have 49,000. Now from the articles it seems that they will deactivate 10,000 of the 11,000 macro sites Leaving them wih 38,000 macro sites plus 6,000 DAS.

Nope, there is absolutely no confusion. Pre-merger T-Mobile has 37,000 HSPA+ sites, but another additional 14,000 EDGE / GRPS sites, for a total of 51,000 sites, all of which existed pre-merger.

Source : GigaOM (that article was written one month before MetroPCS and T-Mobile even announced their merger)

http://gigaom.com/2012/09/28/t-mobile-sheds-its-towers-in-exchange-for-a-2-4b-infusion/

Everyone's always saying they have a smaller footprint than Sprint, but I'm not so sure about that.

I think Sprint has a larger "footprint" (in that, they cover more square miles). But I believe T-Mobile actually has a larger "network" (in that, T-Mobile owns and runs more cell sites than Sprint does).

I know in my market, this is definitely true -- Sprint has more land covered than T-Mobile, by a very large margin. But Sprint sites are spread out very far apart, often to the very edge or further than PCS can be run.

T-Mobile's covers significantly less ground than Sprint, but the coverage areas they do have are easily twice as dense, in terms of towers per square mile.

- - -

Sprint used to manage something like 70,000k sites, but I think as part of Network Vision, they are turning that down to just 38,000k sites. (I know the T-mobile numbers for sure, but I'm less confident about these Sprint numbers). If this is true, then Sprint (after network vision completes) will have a smaller network (in terms of number of active sites) than T-Mobile does.

Source #1 : http://www.wirelessweek.com/news/2012/03/sprint-plans-44-drop-cell-sites

Source #2 : http://gigaom.com/2012/02/08/sprint-can-barely-wait-to-rid-itself-of-nextel-network/

Those numbers do *not* take Clearwire into account, so when Sprint buys Clearwire, if they don't throw those Clear sites away, they could easily match T-Mobile in density, and surpass T-Mobile in number of sites, in those markets.

1) can you provide a source/link for the 50k towers? Over in the "tmo LTE vs Sprint LTE" thread this has split even the mods.

It's from an interview Fierce Wireless did with Neville Ray. I don't have the link on me at the moment, but I can dig it out and post it here in a few hours.

2) what do you mean "for a fixed price"? Tmobile has domestic roaming limits which would make no sense if its users could have unlimited roaming for one fixed price to Tmobile.

Oh, no, I must not have been clear. Your absolutely right, T-Mobile has strict limits on a subscribers usage of roaming.

I meant T-Mobile pays a fixed price for usage to AT&T, for the right to roam. As in, AT&T can't randomly jack up the rates it charges T-Mobile for each minute/sms/kilobyte, as they've agreed to a specific payment rate for the next few (7?) years.

The total cost isn't fixed to my knowledge (I assume T-Mobile pays for every single minute/sms/kilobyte), but the rate is fixed. This lets T-Mobile sign customers up for contracts on service (back when they still did 2 contracts per customer) without fear that AT&T will jack the prices up on them.

They had 35-36,000 of their own sites the rest of the 51,000 are MetroPCS sites. They don't need to upgrade the MetroPCS sites since those will be shutdown.

This isn't true. T-Mobile has 50k+ of sites total in the US. (the sum of both GRPS/EDGE sites and HSPA+/LTE sites). T-Mobile owned all of these sites before MetroPCS ever merged.

37,000 of them are already semi-modernized already (in that, they run either HSPA+ 21 or HSPA+42, today). All of these sites will get LTE (eventually) if they haven't already gotten LTE.

The other 25% of those sites are GRPS / EDGE, and will remain EDGE-like speeds.

Sometimes, when the equipment fails, TMO puts HSPA+ on old EDGE sites. But it's usually just an equipment replacement, not a backhaul upgrade, so they get "2G" only speeds. (Similar to what Sprint did pre-network-vision where they slapped EVDO radios on a single T1 line, and called it "3G", T-Mobile is replacing failed GRPS / EDGE sites with HSPA+ 21 radios, but leaving the backhaul at a single T1 and calling it "3G")

MetroPCS's sites don't figure into either of the two above numbers in any way (those numbers were announced before the TMO / Metro deal was officially announced).

Its could also be AWS. I hear that due to lower allowed transmission power it has a reduced coverage radius compared to PCS.

This is a common myth. AWS is really close / essentially identical in terms of transmission power and coverage radius. Many people believe AWS is lower, because T-Mobile chooses to run their sites at reduced power. But that's not a symptom of the AWS spectrum, that's a choice T-Mobile makes. (And I'm not a T-Mobile engineer, so I don't know why they make that choice.)

You can see this for yourself by doing a site survey. Find a tower where T-Mobile and MetroPCS are colocated, you can measure the output on AWS. MetroPCS (usually) runs their AWS at full power, and you can measure this at a noticable increase over T-Mobile, who (usually) runs their AWS slightly underpowered.

I don't know why they do this, but I have measured it, and can confirm it does happen.

There is only one national roaming partner for T-Mobile and thats AT&T and they are not playing ball.

I don't believe this is true. My understanding is that, as part of the T-Mobile + AT&T aquisition failing, T-Mobile got a complete, nationwide, 3G roaming agreement at a fixed price for a long length of time (something like 7 years).

However, T-Mobile has a bad habit of purposely blocking roaming in many areas. AT&T is available, it works, runs fine, but T-Mobile specifically disallows all of their subscribers from connecting to AT&T in many areas.

This leads to a big split in usable coverage. Sprint's roaming with Verizon works almost everywhere, so a user gets a lot more effective coverage. T-Mobile's roaming with AT&T is blocked (on T-Mobile's side) in a good chunk of areas, so users get less usable roaming coverage.

WiWavelength, on 20 Jun 2013 - 09:40 AM, said:

No, that is questionable. For the amount of revenue that wireless operators rake in, $1000 per month for a site lease is definitely on the low end.

I'm not saying there aren't more expensive sites (I'm sure there are ones that cost more somewhere) But that's what I was quoted for various sites in urban West Michigan. (I've called everyone here at least once, except Crown Castle, who I presume has the most expensive sites)

WiWavelength, on 20 Jun 2013 - 09:40 AM, said:

And your backhaul estimate is completely inadequate. A symmetrical 50 Mbps fiber link would be barely enough to service one 5 MHz FDD LTE carrier on one sector of a site. Multiply that by two or three for additional LTE carriers, as well as the CDMA1X and EV-DO carriers, then multiply again by three sectors, and now you have your actual backhaul requirements.

A 50x50 line can handle a single 5x5 LTE carrier, since 5x5 LTE maxes out at 37.5mb/s.

But, in theory, you would be correct. If Sprint wanted to run enough backhaul to fully handle all their air interfaces, at maximum load, all the time, they'd need roughly 160x160 to 200x200. (That's assuming a typical tower, with four LTE sectors, and EVDO + 1X service)

However, the vast majority of Sprint sites have never had enough backhaul to max out the air interface, and this isn't changing with Network Vision. Their old "3G" sites had just 1.5 to 4.5mbps of backhaul, even though they could push much more through EVDO rev A if they had more backhaul on the sites.

Most new Network Vision sites typically have just 100x100. This isn't coming from me, this is coming from Marty Nevshemal, Sprint's Vice President of Strategic Programs.

According to Sprint, they run 100x100 to most sites, and pay roughly $1,500 to $3,000 for backhaul at their sites. (Marty says "20 times more than the 4.5 used previously" - 20 * 4.5 = 90)

Source : http://www.fiercebroadbandwireless.com/story/sprint-ethernet-backhaul-gives-us-20-times-more-bandwidth/2012-08-15

Now, there's nothing stopping them from buying more / using more backhaul, and it wouldnt suprise me if they had a few sites in NYC or somewhere that were set up for "high capacity". (Fiber can scale up pretty easily.) But I don't think it's a stretch to assume that the sites here don't have anything more than 100x100 behind them right now.

... but the launched market coverage maps are not accurate initially...

I don't know about the other maps, but the Lansing one isn't just generous on LTE. It's claiming LTE coverage in places where there is no PCS EVDO service at all (even according to Sprint's own maps).

If you flip between the two, the boundary lines for LTE claim to go out farther than Sprint's own map for voice service. It's showing LTE data coverage in places where they claim to only have voice roaming.

What existing 1X 800? You mean Nextel service?

Oops, sorry. Yes, Nextel is what I meant. Fixed.

 

I think you are underestimating the effect of SMR

 

That's a completely fair argument.

I'm using the existing Nextel service as my metric for SMR.

But I've heard that the newer Network Vision equipment should add approx 6db to that, and I've never seen any Network Vision refarmed 800 first hand, so it could work a lot better than I'm expecting.

You mentioned "6 miles". Yes, they *could* turn up SMR to go crazy distances. But I would expect that shouldn't ever happen inside dense urban areas, if only because too many people would connect to a single tower.

 

Co-locating with other carriers cost less but is still significant as you have to get your own backhaul and lease. Still at least $100,000 after everything.

This is way, *way*  too high a guess. I have some actual numbers for West Michigan I can throw in here.

I called on an American Tower within the city limits, right on the freeway (AT&T and MetroPCS are on it). It's about $5,000 in fees to get started ("site surveys", "environmental impact", other fake fees they throw on). It's $1,000/month to be on the tower itself, and backhaul is about $2,000 - $3,000 a month (depending on how much you want, the $3,000 figure gets you 50x50 Fiber line from AT&T). 

This doesn't include electricity (which is small, but meaningful cost) and the cost for equipment + climbers to install it, but assuming its up and running, to maintain a 4G LTE tower is only about $4,000 - $5,000 per month.

If you assume an average revenue per subscriber of $40, you only need about 250 of those subscribers on a tower to give you enough revenue to maintain the tower with 50% margins.

Towers are expensive, no doubt. But they aren't *that* expensive.

Yeah, the 800mhz will help, but I don't think it will fix the problem. It's more like a band-aid.

Really, they just need to fill in the holes, and get on the Towers + DAS already up and running here.

It would only take about 5 towers + 4 DAS leases for Sprint to fix Grand Rapids, and another 10 towers or so to do most of the suburbs + Holland + Grand Haven.

I see.  I don't understand why Sprint didn't add a tower to the business district of St Joseph long ago though after they took control.  I didn't work in the area until recently, but I do remember pretty much every time I'd stop at the bank in that area I would have to try several times just to send a text outside.  When people can't use their phones when they're on their lunch break or out shopping, they're unlikely to stay with that carrier. 

 

Also, I've mapped Cleveland road multiple times on Sensorly, so it apparently doesn't even get 3G to this day. 

The iPCS work here has always been bad, and has never been fixed. They spaced their sites to the maximum limit available for 800mhz, so there are gaping coverage gaps in 1900mhz coverage, by design.

There are entire neighborhoods of Grand Rapids (close to 25% the total area of the city) that have never had working 1900 voice/data/sms from Sprint, ever, due to iPCS bad site spacing. I happen to live in one. 

There's plenty of towers here to rent space on, and they are cheap (MetroPCS and Clearwire can afford to be on them). But Sprint's not on them. Sprint even had one Nextel site (rooftop) in the middle of one of these PCS deadzones. They decomissioned the whole thing. They are literally paying a lease on a rooftop in a deadzone, but wouldn't put PCS up on the site (and that building already had 60x60 Metro Ethernet backhaul available, so that's not the issue) 

It's even sillier out on the suburbs. Some areas have miles of weak / poor coverage. (Jenison, Hudsonville). Other areas have perfect spacing with towers less than 2km apart of each other (Alpine, 28th st in Cascade, 28th street in Grandville).

In areas where Sprint's sites are less than 2km apart, their service is good. More or less competitive with everyone else.

In areas where Sprint's sites are more than 2km apart, service is almost useless. (In Grand Rapids, they have sites at 4km apart! In neighboorhoods with 3 times the density as the suburbs.). It's no suprise that data there is useless, phones stay on weak -105dbm of 1X all day long. (Not enough signal to be usable, but just enough to prevent you from getting roaming)

West Michigan is very hit - or - miss for Sprint.

It will be nice, that as part of Network Vision, all the towers get backhaul + data service. But we really need someone to come up here and finish deploying towers in West Michigan.

No device is going be the perfect full-strength signal all the time no matter the distance from the tower. What he is trying to get is something we all wish we had, perfect signal all the time no matter where we are in reference to tower location.

Perhaps I've mis-spoke above.

To be absolutely clear : he's **not** asking for "perfect signal" or "perfect coverage". He's not complaining about Sprint service.

However, knowing that signal fluctuates, and knowing he happens to be in areas of poor reception most of the time, he's just asking for a device that handles this situation the best.

It's obviously not ideal, but we've clearly seen wide variations in RF performance between devices, and we can't buy every single phone to test them ourselves.

Not asking for Magic. Just looking for recommendations on which device is ideal if you know your signal situation will be poor.

Get an Airave. They cost nothing and improve coverage.

That's great advice, and that would help him at home.

But not at work, or his wife's work, or when shopping, ect.

He'd like to improve the situation everywhere, rather than band-aid one particular spot, even if that improvement is relatively minor.

Backstory :

I've been asked to make a recommendation on a Sprint device for a friend, who lives in a medium sized city, in a very dense urban neighborhood. Nearest Sprint tower is very far away from his house, he gets about -99dbm outside, and -108dbm inside, and the current (not yet NV upgraded) tower pulls about 10-30k down on 3G data speeds.

I told him he might get slightly better reception after NV (by about 6dbm) but hes looking for a phone that can help him through that wait. He wants a phone that has LTE and has the best RF performance or best data reception / quality, above all else. Preferably one with SVDO.

I knew enough to tell him to stay away from the Galaxy Nexus, for the obvious reasons, but other than that, I don't have enough experience with these to make a good recommendation. From some limited testing in store, it looked like the Photon LTE was a good pick (and it ran the fastest 3G speed tests in store), but I'd prefer some insight from folks who have used these.

Anyone have some experience here? Recommendations for a heavy data user who knows they'll be spending a lot of time in a very weak coverage area?

Western Michigan is a former iPCS affiliate market. iPCS was one of Sprint's worst affiliates -- and that is saying something. So, Western Michigan markets are not representative of Sprint markets as a whole.

Wow. I've been digging through various cell information for a long time, but I never knew this before.

That explains a lot. Especially why people kept talking about Sprint having a "dense network." -- I was pretty confused, as their cell sites here are spread out super thin, with less of them (per square mile) than other carriers (even less than either T-Mobile or MetroPCS.)

Thanks!

Sprint is doing every tower well Verizon is doing every other or every third tower.not the same as Sprint

This is technically true, but not a fully accurate representation of the situation.

Verizon's LTE rollout, and Sprint's LTE rollout, seem to be about the same amount of work.

I'm not sure how it is in other markets, but in Michigan, Verizon is only upgrading about half of their towers to LTE, and Sprint is doing most-to-all of them.

However, Sprint here has very little overlap between tower coverage -- their cell sites coverage barely overlap -- only just the minimal amount for a call handoff. Verizon on the other hand, has a lot of redundant tower coverage, and lots of overlap between sites.

So, it's technically true that Verizon isn't upgrading "all" of their sites, and that Sprint is upgrading all of theirs. But Verizon has so many extra, overlapping sites that even if they only do half of their sites, the total number of towers upgraded will end up pretty similar to Sprint's.

And while they haven't started this yet, Verizon will be using AWS for LTE on all of their tower sites in a year or two, so eventually, all of their towers (even the redundant ones) will be running LTE.

(And, by the time Verizon is actively deploying AWS LTE, Sprint will likely be re-purposing Clearwire spectrum to do something similar with their sites)

This seems very unlikely, considering Sprint's current plan to purchase Clearwire will give them more than double the spectrum T-Mobile / MetroPCS will have.

If anything, it might be a requirement on Sprint to divest some of their Clearwire EBS spectrum to other players. (This seems very unlikely to me too, but if any divestitures were to occur, I'd expect them in the Sprint/Clear deal before the T-Mobile/MetroPCS deal)