Feels weird defending a competitor, but Samsung has officially announced details of their new phone's US processor choice that conflict with this post. What this company witnessed was probably an engineering sample with different hardware than the eventual shipping version.
Makes me wonder about how solid the rest of their information is. Carriers have test labs that can deliver any sort of signal to a phone. So maybe some of the test results are just from an LTE node at a test lab and have nothing to do with that being a launch city. If something isn't announced it may never even see the light of day anyway.
Given that, I think it is very possible they misidentified the device. As many have noted, the designs of the S3 and S4 are quite similar. And unlike an S4, it would make sense for T-Mobile's LTE S3 to have a Snapdragon S4, like all of the other US S3s.
Op here, we're pretty sure its the Galaxy S IV. The model number, which is what we use to identify it, is the SGH-M919 and all other specs e.g. the screen resolution match up to the S IV, rather than the S III. The other comment on it being an engineering device and the chipset for commercial launch is a fair point though.
It's not super-useful to measure the speed of a network nobody is using yet. AT&T gives me 5-6 megabits in a skyscraper in midtown Manhattan, in the middle of the afternoon. I'd be very surprised to see T-Mobile, with its limited resources and limited spectrum be able to come close.
Which is ultimately why it was a total disaster that the DOJ nixed the ATT/T-Mobile merger. Now you've got two carriers with not enough spectrum versus Verizon that's sitting on a fat 20 MHz nationwide band.
Real world speeds will undoubtedly differ, as the article describes, but it's at least an improvement over no data or the 'peak' data speeds that operators like to quote that bear no relation to the actual speeds consumers experience.
The primary challenge to the mobile operators is backhaul from the base stations, not spectrum. At any given tower/site, you've only got a certain amount of fixed facilities from the local wireline carrier that are available. In other cases, you're using point-to-point microwave links to backhaul until you can offload that traffic to a fiber-connected location. In either situation, you need to think about the entire picture before you go ahead and focus on "spectrum" as the reason why your link speed differs from what you expect. There are many moving parts and they all can cause congestion.
And the AT&T/T-Mobile merger was nixed because they tried to pull a fast one on the FCC/DOJ by claiming that they needed the spectrum/infrastructure to upgrade their network. When a document surfaced that showed that their cost to upgrade their existing network was a fraction of what they were paying for the T-Mobile network, it became clear that they were buying them to eliminate price competition.
When a cell reaches capacity, you subdivide the cell. e.g. You turn a single 3-sector cell into three smaller cells. Each cell operates in either a different frequency range or in the case of OFDMA (LTE) via a remapping of the timeslots and a lower upper power range.
Exactly the same way that you can re-use the 2.4Ghz band for WiFi a bazillion times over.
So the challenge isn't the available spectrum; it's the available dollars to install base-stations and do backhaul. QED.
That doesn't mean that availability of spectrum isn't a challenge, it just means it's a challenge that can be overcome by throwing money at the problem. You're taking about a complex multi-factor optimization problem. The fact that you can address a challenge, vis-a-vis a competitor, along one axis of the design space by moving along a different axis of the design space does not mean that there is no challenge.
If one competitor has more spectrum than another, it can support similar numbers of users at similar densities at lower capital costs. Alternatively, it can offer users higher speeds and higher capacity for similar capital investment.
In other words, if AT&T and Verizon both have $1 billion to build cell sites/backhaul in Chicago, and Verizon has 20 MHz of spectrum and AT&T only has 10 MHz, then Verizon can support more users at higher speed and its service will be perceived as much better. Alternatively, it can spend less than $1 billion on capital costs and offer the same service level as AT&T at a lower price while maintaining similar profits.
Also, making cell sites smaller is not a magic bullet. It's just one axis along the design space. Inter-cell interference is not a trivial problem, and having more spectrum can allow you to go to a point in the network design space where you have to worry less about that problem.
Except that more spectrum isn't free and is a finite resource. It isn't fair to dismiss one carrier as not being viable (which is what I took from your original post) if they can compete along a different axis more efficiently when compared to another carrier.
Don't forget that over time, technology advancement tends to favor later entrants to the market versus early ones. It's entirely possible that T-mobile can operate an LTE network with triple the base-station density of AT&T because they waited until the equipment was at a cost that would make the deployment cost-efficient.
My whole point is that the place in the design space a carrier with less spectrum is forced to compete from is less efficient than the place in the design space of a carrier with more spectrum. The carrier with less spectrum is forced to spend more capital on building cell sites, spend more money on operating costs for more numerous, smaller sites versus less numerous, larger sites, spend more money debugging inter-cell interference, etc, all just to match the other carrier in performance at a given capacity. The carrier with more spectrum has more capital tied up in spectrum holdings, but that's a totally non-depreciating capital asset that's by and large cheaper than network infrastructure to begin with.
And I'm not writing off T-Mobile based on their current spectrum position. I'm writing them off based on the trajectory. If T-Mobile has to burn more capital deploying an LTE network with much higher density to achieve similar performance, then they are going to have lower margins than VZN and AT&T. Lower margins means less capital for building out the N+1 generation system.
Of course this is all theoretically consistent with the status of wireless as a market that breeds natural monopolies. VZN has a spectrum edge right now that's moderate nationwide, but fairly substantial in many major cities. That edge in spectrum holdings means they can serve more customers at a given performance level for dollar of capital spent on infrastructure. That means when the next spectrum auction happens, they'll be able to acquire even more spectrum and enlarge their lead.
I thought T-Mob just got a major spectrum influx due to the failed merger, plus they bought some from Verizon last year. And they're doing that merger w/ MobilePCS.
The problem with the T-Mobile/ATT merger was that it would have concentrated a lot of customers and eliminated a national carrier. Frankly, allowing Verizon to buy the CableCo AWS spectrum was a mistake as well. Competition works well when carriers are closer in strength to each other.
But LTE seems to handle traffic well. Verizon has 50% of its data traffic flowing over a 10x10 LTE deployment.
In terms of T-Mobile, they will be launching 10x10 LTE just like Verizon and AT&T have and, via the MetroPCS merger, will even be able to launch 20x20 LTE in many top markets in the future. In some ways, T-Mobile will have the nicest spectrum position with 50MHz of AWS in NY, LA, Detroit, Boston, SanFran, etc. It's also important to look at spectrum per customer. AT&T averages 97MHz, but has more than double the customers of the combined T-Mo/Metro while that's only 28% more spectrum - ~150% more customers with ~30% more spectrum. A similar case can be made with Verizon. With 50MHz, it should be easy enough to get a 10x10 LTE deployment out to match Verizon and AT&T on speed and capacity there.
Now, Verizon will likely augment in the coming years with a second 10x10 LTE deployment in the AWS band with the CableCo spectrum. But T-Mobile will likely be able to get 20x20 LTE at AWS in many areas. A lot of MetroPCS' customers are already VoLTE equipped and T-Mobile notes that they will likely be able to shut down CDMA in 2 years time.
To be frank, LTE holds up well. A 10x10 LTE deployment offers good speeds even as it starts taking the majority of data traffic in 20MHz. As traffic migrates away from PCS and cellular, there will be more spectrum for LTE. While data traffic is likely to continue to rise, I don't think it's unreasonable to believe that 40MHz could handle near 100% of the current data traffic for a giant like AT&T or Verizon. VoLTE will decrease the spectrum that voice communications are using.
One thing to note about the Verizon situation: yes, they got a new 20MHz chunk of unused spectrum and that's awesome for Verizon, but they only hold 12% more spectrum than AT&T and they have more customers to serve with it. As traffic on legacy networks drops off, I think we're going to see a lot of spectrum become available. The amount of data transferred over networks in the future will increase, but LTE Advanced is on the way and there's a lot of spectrum to be repurposed from UMTS and CDMA/EV-DO. It wasn't great that Verizon was allowed to open a nice spectrum advantage through their CableCo deal, but it's not as big as one would think at first glance.
In fact, the company in the worst position is Sprint. With 54MHz, they're around half AT&T and Verizon and going to be behind T-Mobile. However, if they get approval to buy out Clearwire that will change (even if Clearwire's spectrum offers a lot less range).
I'd be cautiously optimistic about the competitive environment. The T-Mo/Metro combination will give them a lot of contiguous AWS spectrum and given their low customer numbers and ability to push 20x20 LTE in many areas, they could end up with a great network for those in urban areas (especially if they hit 200M people covered by the end of 2013, which is their plan). Sprint may get approval for their Clearwire purchase giving them a lot of spectrum and the ability to move beyond their 5x5 LTE deployment currently in progress. Sprint also has SMR spectrum that they'll hopefully use to their advantage in creating more robust coverage. I'm not saying that Verizon and AT&T's days are over, but it's possible that we'll have better options from more competitors in the coming years.
EDIT: Also of note, AT&T has un-paired spectrum (download-only) that it got from Qualcomm. It's 6MHz across the country, but 12MHz in the Northeast Corridor and California. That's another 5 or 10MHz down and upload capacity usually isn't needed as much as download. While this spectrum isn't usable pre-LTE Advanced and its channel bonding, it's unclear that they would be using ...
17 comments
[ 3.3 ms ] story [ 52.1 ms ] threadMakes me wonder about how solid the rest of their information is. Carriers have test labs that can deliver any sort of signal to a phone. So maybe some of the test results are just from an LTE node at a test lab and have nothing to do with that being a launch city. If something isn't announced it may never even see the light of day anyway.
Given that, I think it is very possible they misidentified the device. As many have noted, the designs of the S3 and S4 are quite similar. And unlike an S4, it would make sense for T-Mobile's LTE S3 to have a Snapdragon S4, like all of the other US S3s.
Which is ultimately why it was a total disaster that the DOJ nixed the ATT/T-Mobile merger. Now you've got two carriers with not enough spectrum versus Verizon that's sitting on a fat 20 MHz nationwide band.
The primary challenge to the mobile operators is backhaul from the base stations, not spectrum. At any given tower/site, you've only got a certain amount of fixed facilities from the local wireline carrier that are available. In other cases, you're using point-to-point microwave links to backhaul until you can offload that traffic to a fiber-connected location. In either situation, you need to think about the entire picture before you go ahead and focus on "spectrum" as the reason why your link speed differs from what you expect. There are many moving parts and they all can cause congestion.
And the AT&T/T-Mobile merger was nixed because they tried to pull a fast one on the FCC/DOJ by claiming that they needed the spectrum/infrastructure to upgrade their network. When a document surfaced that showed that their cost to upgrade their existing network was a fraction of what they were paying for the T-Mobile network, it became clear that they were buying them to eliminate price competition.
Where the bottleneck is depends on market being served. In dense urban areas (i.e. the ones that matter), the bottlenecks is often spectrum. See: http://gizmodo.com/5847811/giz-explains-whats-wireless-spect..., http://www.androidpolice.com/2013/01/25/verizon-sells-att-lo.... This is both because the density of users in urban areas, and the fact that it's easier to provision urban cell cites with fiber.
When a cell reaches capacity, you subdivide the cell. e.g. You turn a single 3-sector cell into three smaller cells. Each cell operates in either a different frequency range or in the case of OFDMA (LTE) via a remapping of the timeslots and a lower upper power range.
Exactly the same way that you can re-use the 2.4Ghz band for WiFi a bazillion times over.
So the challenge isn't the available spectrum; it's the available dollars to install base-stations and do backhaul. QED.
If one competitor has more spectrum than another, it can support similar numbers of users at similar densities at lower capital costs. Alternatively, it can offer users higher speeds and higher capacity for similar capital investment.
See: http://gigaom.com/2012/05/14/atts-chicago-problem-why-lte-sl...
In other words, if AT&T and Verizon both have $1 billion to build cell sites/backhaul in Chicago, and Verizon has 20 MHz of spectrum and AT&T only has 10 MHz, then Verizon can support more users at higher speed and its service will be perceived as much better. Alternatively, it can spend less than $1 billion on capital costs and offer the same service level as AT&T at a lower price while maintaining similar profits.
Also, making cell sites smaller is not a magic bullet. It's just one axis along the design space. Inter-cell interference is not a trivial problem, and having more spectrum can allow you to go to a point in the network design space where you have to worry less about that problem.
Don't forget that over time, technology advancement tends to favor later entrants to the market versus early ones. It's entirely possible that T-mobile can operate an LTE network with triple the base-station density of AT&T because they waited until the equipment was at a cost that would make the deployment cost-efficient.
And I'm not writing off T-Mobile based on their current spectrum position. I'm writing them off based on the trajectory. If T-Mobile has to burn more capital deploying an LTE network with much higher density to achieve similar performance, then they are going to have lower margins than VZN and AT&T. Lower margins means less capital for building out the N+1 generation system.
Of course this is all theoretically consistent with the status of wireless as a market that breeds natural monopolies. VZN has a spectrum edge right now that's moderate nationwide, but fairly substantial in many major cities. That edge in spectrum holdings means they can serve more customers at a given performance level for dollar of capital spent on infrastructure. That means when the next spectrum auction happens, they'll be able to acquire even more spectrum and enlarge their lead.
But LTE seems to handle traffic well. Verizon has 50% of its data traffic flowing over a 10x10 LTE deployment.
In terms of T-Mobile, they will be launching 10x10 LTE just like Verizon and AT&T have and, via the MetroPCS merger, will even be able to launch 20x20 LTE in many top markets in the future. In some ways, T-Mobile will have the nicest spectrum position with 50MHz of AWS in NY, LA, Detroit, Boston, SanFran, etc. It's also important to look at spectrum per customer. AT&T averages 97MHz, but has more than double the customers of the combined T-Mo/Metro while that's only 28% more spectrum - ~150% more customers with ~30% more spectrum. A similar case can be made with Verizon. With 50MHz, it should be easy enough to get a 10x10 LTE deployment out to match Verizon and AT&T on speed and capacity there.
Now, Verizon will likely augment in the coming years with a second 10x10 LTE deployment in the AWS band with the CableCo spectrum. But T-Mobile will likely be able to get 20x20 LTE at AWS in many areas. A lot of MetroPCS' customers are already VoLTE equipped and T-Mobile notes that they will likely be able to shut down CDMA in 2 years time.
To be frank, LTE holds up well. A 10x10 LTE deployment offers good speeds even as it starts taking the majority of data traffic in 20MHz. As traffic migrates away from PCS and cellular, there will be more spectrum for LTE. While data traffic is likely to continue to rise, I don't think it's unreasonable to believe that 40MHz could handle near 100% of the current data traffic for a giant like AT&T or Verizon. VoLTE will decrease the spectrum that voice communications are using.
One thing to note about the Verizon situation: yes, they got a new 20MHz chunk of unused spectrum and that's awesome for Verizon, but they only hold 12% more spectrum than AT&T and they have more customers to serve with it. As traffic on legacy networks drops off, I think we're going to see a lot of spectrum become available. The amount of data transferred over networks in the future will increase, but LTE Advanced is on the way and there's a lot of spectrum to be repurposed from UMTS and CDMA/EV-DO. It wasn't great that Verizon was allowed to open a nice spectrum advantage through their CableCo deal, but it's not as big as one would think at first glance.
In fact, the company in the worst position is Sprint. With 54MHz, they're around half AT&T and Verizon and going to be behind T-Mobile. However, if they get approval to buy out Clearwire that will change (even if Clearwire's spectrum offers a lot less range).
I'd be cautiously optimistic about the competitive environment. The T-Mo/Metro combination will give them a lot of contiguous AWS spectrum and given their low customer numbers and ability to push 20x20 LTE in many areas, they could end up with a great network for those in urban areas (especially if they hit 200M people covered by the end of 2013, which is their plan). Sprint may get approval for their Clearwire purchase giving them a lot of spectrum and the ability to move beyond their 5x5 LTE deployment currently in progress. Sprint also has SMR spectrum that they'll hopefully use to their advantage in creating more robust coverage. I'm not saying that Verizon and AT&T's days are over, but it's possible that we'll have better options from more competitors in the coming years.
EDIT: Also of note, AT&T has un-paired spectrum (download-only) that it got from Qualcomm. It's 6MHz across the country, but 12MHz in the Northeast Corridor and California. That's another 5 or 10MHz down and upload capacity usually isn't needed as much as download. While this spectrum isn't usable pre-LTE Advanced and its channel bonding, it's unclear that they would be using ...