Jump to content

T-Mobile LTE & Network Discussion


CriticalityEvent

Recommended Posts

Dish trying to get a deal cooing with Deutsche Telekom. My guess is they're angry at Sprint for not letting them have Clear.

 

Sprint will let them have Clearwire if they come to an agreement with T-Mobile.

Edited by bigsnake49
Link to comment
Share on other sites

I love how quick people are to judge when it comes to the high data use with the "Get a fiber line" comment. I live in an area that can't even truly be considered rural, as it's only about 20-30 minutes outside of Cleveland Ohio. Out here there is NO fiber, and nothing offering speeds even close to it, best and most reliable here is a 6mBps DSL line, at times when traffic is high or when there are network problems it gets slightly better speeds than dial up. VZW and AT&T both have LTE here, it's faster and more reliable than my home Internet is for me, hell many times even Sprint's overloaded 3G network is too. Point is when Sprint LTE gets here, I will use what I pay for. A reliable, fast, better than my DSL network, until a service like FIOS is out here (could be many years, if ever) I will be highly dependent on my wireless connection, as are many others around our country. Not everything can have a blanket statement applied to it, this is one of those situations.

Link to comment
Share on other sites

I love how quick people are to judge when it comes to the high data use with the "Get a fiber line" comment. I live in an area that can't even truly be considered rural, as it's only about 20-30 minutes outside of Cleveland Ohio. Out here there is NO fiber, and nothing offering speeds even close to it, best and most reliable here is a 6mBps DSL line, at times when traffic is high or when there are network problems it gets slightly better speeds than dial up. VZW and AT&T both have LTE here, it's faster and more reliable than my home Internet is for me, hell many times even Sprint's overloaded 3G network is too. Point is when Sprint LTE gets here, I will use what I pay for. A reliable, fast, better than my DSL network, until a service like FIOS is out here (could be many years, if ever) I will be highly dependent on my wireless connection, as are many others around our country. Not everything can have a blanket statement applied to it, this is one of those situations.

 

Are you in an awful, horrible situation? Absolutely.

 

You should still get the 6Mbps line though.

 

I certainly hope you aren't eluding to any sort of violating the T's and C's, BTW. I'll give you the benefit of the doubt.

Link to comment
Share on other sites

Well I don't tether and do all necessary network usage on the device for one thing, plus considering I'm on a BB that greatly compresses my data through BIS, I use around 1GB a month. When NV and LTE comes, and I move to a Q10 without compression my usage will increase, where it will end up is yet to be seen. However I'm hoping it won't be more than 6-8 GB. I'm simply pointing out what is fact for many across the US, when a carrier advertises "Unlimited" then that service should be true to form. Even more so with LTE making data transmission cheaper for the carrier and more efficient. And yes, I do have the 6 mBps line, it's provided by a smaller alternative ISP/Telco called WindStream, which for hardwire services is similar to Sprint.

Edited by Epic4G25
Link to comment
Share on other sites

Agreed, If that is real, Sprint is woefully mismanaged for keeping such a user on the network.

 

we still dont know how many lines this Is for so I am reserving judgement here.

Link to comment
Share on other sites

Anyway you look at it, it will still come down to resource/bandwidth management. I will not take sides in this ongoing debate or lean towards any of the extreme usage scenarios and limitations we are discussing. It is possible that someone can use much more resources than another if they are in a less congested area of the network since the overall bandwidth is shared by less individuals.

 

There first needs to be a consensus on what is typical usage scenario but that still needs to be weighed against the % of network utilization as well. Another similar shared resource would be cable internet. A typical $100/mo cable internet plan can provide speeds of up to 50mbps/6.25MBps but is shared amongst whatever dedicated bandwidth path the provider has allocated to a particular sector because of finite resources and infrastructure shared between given user of that sector/segment/branch/node/trunk/whatever the hell you want to call it.

 

Now let's say this $100/mo plan comes with a soft cap of 400GB/mo (3200gbits or 3200000mbits) (because it does).

 

 

If you are in a relatively uncongested area of the network you may see higher speeds on average than those in area with more users and network utilization. If you want to look at what absolute limits on even the above cable ISP plan you would find that if you could maintain peak speeds collectively for any period within that month you would hit your bandwidth cap within

 

64,000 seconds

1,066 2/3 minutes

17 7/9 hours

 

 

 

This means that 400GB/MO works out to about 2.4% of the theoretical maximum network throughput if you could someone maintain constant maximum advertised speeds.

 

 

 

Perhaps if we looked at mobile bandwidth/caps the same way we do with other limited bandwidth resources could have a consensus on what constitutes a fair amount of monthly user data.

 

 

 

Can anyone give me an idea of the maximum throughput/capacity available per pop a month when one looks at Sprint's or any networks total available bandwidth/spectrum/throughput at any particular moment in time divided by the number of subscribers so that we can find out what 2.4% of that would be? There would be a lot of math required her given the vasts amount of spectrum and load that the entire network can handle at any moment in time along with bottlenecks introduced by inherent hardware limitations and technologies. I know we have a MENSA member somewhere on this site.

 

 

 

I only put this out there because I believe that defining at least a methodology for calculating fair usage we could end what people on either end would find unreasonable.

 

 

It appears that most High Speed ISP's range from 1.8% to 4.2% of total advertised maximum speeds when you do the math. I have looked at the soft/hard caps of most broadband service providers to come up with the range, but maybe I need to do a spreadsheet here.

 

If you want to go through and do the math yourself (don't forget the difference between GB and gb) you will find that the theoretical maximum throughput of the above ISP plan is 16,425 Gigabytes/month while they soft cap you at 400 Gigabytes (or 2.43% of theoretical max).

 

I cannot imagine anyone utilizing or remotely receiving the maximum throughput but at least we can get a feel for what is commercially acceptable at this point in time.

  • Like 1
Link to comment
Share on other sites

Can anyone give me an idea of the maximum throughput/capacity available per pop a month when one looks at Sprint's or any networks total available bandwidth/spectrum/throughput at any particular moment in time divided by the number of subscribers so that we can find out what 2.4% of that would be?

 

This is not exactly what you are looking for, but it will get you started. I posted this about a year and a half ago.

 

Many people seem to have an ill formed notion (or just no idea) of the actual data capacity of an EV-DO carrier channel. It is certainly not infinite.

 

The maximum amount of downlink (i.e. cell site to mobile) data that an EV-DO Rev A carrier channel can deliver during a 30 day billing period is ~949 GB. Now, that assumes that the carrier channel can deliver the ideal maximum DRC Index 14 (3072 kbps) continuously 24 hours per day, seven days per week for an entire month. Real world performance does not even approach that limit.

 

Instead, in real world usage, most data activity occurs during only 16 out of 24 hours per day. And the carrier channel can never deliver the ideal maximum 3072 kbps all the time, as many mobiles frequently experience radio fading conditions that degrade S/N ratio below the >10 dB required for maximum DRC. So, the actual mean data rate is no greater than DRC Index 8 (921.6 kbps). And the practical maximum amount of downlink data that an EV-DO Rev A carrier channel can deliver continuously during a 16 hour day, 30 day billing period is ~190 GB.

 

Just 10 users each consuming 10 GB could soak up greater than half of the capacity of that EV-DO Rev A carrier channel. The rest of the users of that carrier channel would be left to share the remaining capacity. Depending on the number of other users and the extent of their data usage, average data speeds could suffer.

 

For those reasons, tiered or capped data is not inherently "evil." Rather, it can be an effective network management technique. Certainly, tiered data should not replace deployment of additional carrier channels as demand grows and supply permits. Regardless of the number of carrier channels, though, wireless data capacity is a shared, finite resource. And users should face reasonable limits or pay at least quasi proportionally for how much they use. Otherwise, the wants/needs of a few can impose upon those of the whole. See the "tragedy of the commons."

 

http://www.phonescoop.com/articles/discuss.php?fm=m&ff=8856&fi=2949309

 

AJ

Link to comment
Share on other sites

This is not exactly what you are looking for, but it will get you started. I posted this about a year and a half ago.

 

 

 

http://www.phonescoo...8856&fi=2949309

 

AJ

 

I certainly do not disagree with any of these realities, my main and probably only point is to find a concrete datapoint to define acceptable usage and I figured a comparative percentage should add some definitive numbers to that line of thought.

 

Once that is known and addressed we can all look at data tiers much more objectively without the subjective concept and resultant arguments on what unlimited means or what one defines as average or acceptable usage. Obviously no network capacity is infinite or truly unlimited in that regard nor can we pretend it is.

 

 

I thank you for that link, as it is a step in the right direction.

Link to comment
Share on other sites

Given your statement of "And the practical maximum amount of downlink data that an EV-DO Rev A carrier channel can deliver continuously during a 16 hour day, 30 day billing period is ~190 GB."

 

If we were to apply my relatively conservative percentage of 2.34% (comparatively) in the percentage range I listed above then we would be left with 4.446GB/mo to be an acceptable cap of allotted (unlimited) data on an all EV-DO Rev. A carrier. Obviously many would use less and a few would use more, but this would be my baseline range if I were Sprint in determining where throttling would occur in the future if we were not to go with hard caps.

 

Obviously, these numbers are subject to change (higher) in the future, given developments in spectral efficiency as well as the amount of available spectrum theoretically increasing or finally being utilized.

 

 

Even If Sprint were to institute throttling today based upon my above considerations they would still offer 2x as much full speed data as the nearest competitor. If my logic is flawed in any respects, I am more than receptive to anyone pointing out where my thought process is diverging from reality.

 

:tu:

Link to comment
Share on other sites

Network bandwidth is infinite for as long as the network functions at a normal mode of operation.

 

What is not infinite is the bandwidth per timeslice, which is both shared and finite. But it is also variable, dependent on network medium (and network segment), and the medium typically improves with both technology and time.

 

Bandwidth that is not used, is gone and wasted. But that is ok because bandwidth over time is a constantly renewable resource.

 

Conserving bandwidth, in under-utilized states, is pointless, because it does not increase available bandwidth in over-utilized states.

 

Given no outside-induced network traffic control or throttling (and all other things being equal), network over-utilization degrades equally for all all affected users (of the particular network segment that is over-subscribed), including both extremes of the 'data hogs' and '100meg'ers'.

 

100 people trying to cross the sahara with 1 pint of water, is not going to work.

 

Of course, neither will 1 person trying to cross the sahara with 1 pint of water.

 

A shared 56k modem won't serve 10 people very well.

 

Heck, a 56k modem won't serve 1 person very well either. Of course that did not stop us from using 300 baud modems with acoustic couplers and thinking how amazing that took 5 minutes to download a picture that would take miliseconds now.

  • Like 1
Link to comment
Share on other sites

Network bandwidth is infinite for as long as the network functions at a normal mode of operation.

 

What is not infinite is the bandwidth per timeslice, which is both shared and finite. But it is also variable, dependent on network medium, and the medium typically improves with both technology and time.

 

Bandwidth that is not used, is gone and wasted. But that is ok because bandwidth over time is a constantly renewable resource.

 

Conserving bandwidth, in under-utilized states, is pointless, because it does not increase available bandwidth in over-utilized states.

 

 

 

This is something to take into consideration as well obviously, which is why finite resources like broadband ISP scale based upon overall traffic.

 

There is nothing wrong with providing as a fairly large chunk of unused capacity to any one user/users if it is currently available, plus the faster speeds will reduce the timeslices that are eaten since they will be pushed through faster.

 

 

 

We still need to find common ground. Clearly 500mb/mo is far too low to work for everyone and 400GB is far too high to be acceptable.

 

 

 

 

ON A SIDE NOTE / AND GOING BACK TO MY PREVIOUS PERCENTAGES (what can I say I'm a numbers guy)

 

if we went to the higher end Sprint could make the base limit 10GB/mo for standard usage to remain competitive if and when they must introduce data caps, however the kicker could be that they could charge $1/GB a month for those going over the limit and at the same time subtract $1 per GB for every user that goes under that baseline. Sure, it may not be ideal, but it certainly is better than what the competitors are presently offering when you look at the per GB increase in cost, and the fact that there is no reduction for those who use less. This would certainly go well within the confines of Sprint's "simply/simple" mantra.

Link to comment
Share on other sites

IMO, your ~4gb/month is not necessarily unreasonable.

 

For evdo.

 

But, now do your 2% calculations for LTE.

  • Like 1
Link to comment
Share on other sites

IMO, your ~4gb/month is not necessarily unreasonable.

 

For evdo.

 

But, now do your 2% calculations for LTE.

 

When NV is complete and we can get a solid number on average throughput I will surely do that.

Link to comment
Share on other sites

It appears that most High Speed ISP's range from 1.8% to 4.2% of total advertised maximum speeds when you do the math. I have looked at the soft/hard caps of most broadband service providers to come up with the range, but maybe I need to do a spreadsheet here.

 

I like the methodology, but as long as we are doing some back of the napkin math, I would adjust some of the figures. The 1.8-4.2 percent figure is derived from wired broadband data caps. But is wired broadband an appropriate parallel for wireless broadband? I say not. The number of subs that share a cable node, for example, is almost certainly much lower than the number of subs that share a wireless sector. I ran some informal calculations a month or two ago. See below:

 

If we take Sprint's 55 million aggregate subs, then divide by 95,000 sectors (i.e. 38,000 sites multiplied by an average of 2.5 sectors per site), we get a result of about 600 subs per sector. Now, that assumes an even dispersal of subs over all sites. But that is hardly the case. Sprint has thousands of rural highway sites that see mostly/only transient use. So, at least 600 subs per urban PCS sector would be a conservative estimate and 1000 subs per sector would not be out of the question.

 

Also, 25 Mbps seems a bit on the high side for average 5 MHz FDD LTE sector capacity. Remember, 37 Mbps is the maximum throughput, but that assumes 64-QAM and 2x2 MIMO for all connections. Subs within a sector are generally so spread out that path loss (thus dynamic rate control) probably follows some sort of normal distribution. For that reason, I would assume the average throughput to be no greater than 50 percent of the maximum, hence 18 Mbps.

 

http://s4gru.com/ind...post__p__111617

 

Now, if we use the low end estimate of 600 subs per sector, the fair and proportional cap per sub would be 0.17 percent of the total. Admittedly, that would be a bit extreme, since it would not take advantage of the synergies that come from some using less, some using more than their fair share. But the 1.8-4.2 percent figure may be a bit extreme in the other direction, since only a relatively small number of the 600 subs would have to reach the cap in order to soak the sector completely.

 

AJ

  • Like 1
Link to comment
Share on other sites

I like the methodology, but as long as we are doing some back of the napkin math, I would adjust some of the figures. The 1.8-4.2 percent figure is derived from wired broadband data caps. But is wired broadband an appropriate parallel for wireless broadband? I say not. The number of subs that share a cable node, for example, is almost certainly much lower than the number of subs that share a wireless sector. I ran some informal calculations a month or two ago. See below:

 

 

 

http://s4gru.com/ind...post__p__111617

 

Now, if we use the low end estimate of 600 subs per sector, the fair and proportional cap per sub would be 0.17 percent of the total. Admittedly, that would be a bit extreme, since it would not take advantage of the synergies that come from some using less, some using more than their fair share. But the 1.8-4.2 percent figure may be a bit extreme in the other direction, since only a relatively small number of the 600 subs would have to reach the cap in order to soak the sector completely.

 

AJ

 

Continuing my napkin math and factoring in your new numbers based on 18Mbps in a 5x5 configuration, this would give us a theoretical maximum of 5,912 GB/month. Placing your lower per sector adjustment of 0.17% against that would still keep it in the 10GB monthly allotment range that I indicated earlier.

 

Since your numbers are much closer to including all variables we were looking at earlier inasmuchas accounting for users per sector, theoretical maximum (adjusted for real world network conditions) and average saturation, I will still stand by the 10GB/MO being an acceptable pre-throttle soft cap when NV is completed with all things being equal. Of course subscriber numbers will go up in the future and available spectrum and network capacity couple with spectral efficiency will increase as well.

 

I am just glad that at least we have identified a solid basis for diminishing returns on Sprint's end and where it will ultimately need to look at caps for its users. The present average probably has the majority of users somewhere in the 1-2GB/mo usage range, which means Sprint-uh-bank has some comfortable wiggle room going forward on an averaged basis.

Link to comment
Share on other sites

Continuing my napkin math and factoring in your new numbers based on 18Mbps in a 5x5 configuration, this would give us a theoretical maximum of 5,912 GB/month.

 

When I made those informal calculations of EV-DO capacity, I did so for a 16 hour day. I think using a full 24 hour day model is a bit unrealistic -- though, I do wish that wireless carriers would consider relaxing traffic shaping and/or caps for off peak data usage. So, per my 16 hour day estimates, the real world max for a 5 MHz FDD LTE carrier is somewhere closer to 3900 GB per month.

 

Do you think my narrowing the scope to primarily the waking hours is reasonable?

 

AJ

Link to comment
Share on other sites

 

Do you think my narrowing the scope to primarily the waking hours is reasonable?

 

AJ

 

Very reasonable. But you don't factor in the effects of 800 MHz and 2500 MHz spectrum. Admittedly, both are in the future, and if good ole' Charley Dish gets his way, 2500 may never be fully implemented. But the Sprint plan does include both. As Sprint adds more subs, and also as more subs start using more bandwidth, it is likely that both bands will come into play -- possibly as soon as early 2014 in some areas. Sprint may also be able to reallocate some current EvDO channels and also start using the US Cellular spectrum where available, in a slightly later time frame. So the actual number is likely to be well above 3900 GB per month, maybe much above. (And, yes, I use my recently-speeded-up Comcast WiFi at home [i really hate to praise the buggers!], and have an Airave, so I am offloading from the Sprint network as much as possible. Of course, I don't have LTE at home yet unless I stand on the south-facing side of the roof, but I still expect to depend heavily on WiFi when I do have reliable LTE.)

  • Like 1
Link to comment
Share on other sites

When I made those informal calculations of EV-DO capacity, I did so for a 16 hour day. I think using a full 24 hour day model is a bit unrealistic -- though, I do wish that wireless carriers would consider relaxing traffic shaping and/or caps for off peak data usage. So, per my 16 hour day estimates, the real world max for a 5 MHz FDD LTE carrier is somewhere closer to 3900 GB per month.

 

Do you think my narrowing the scope to primarily the waking hours is reasonable?

 

AJ

 

I will concede that, but I was basing my information on your post stating that this was assumed on a 16 out of 24 operational basis which in itself is still subjective, but even if my number were to be reduced by 2/3 to re-math what should have been proper math we would still be looking at a 6 2/3 GB/mo allotment before a cap would make sense from a diminishing returns standpoint.

 

I am not trying to demonize the initial (500mb) amount you suggested which many believe to be quite low, I am only trying to provide a factual basis of comparison that is relative to other ISP's since we are moving into a more data-centric world.

 

Whether that equates to is 6 2/3GB or 10GB/mo that is the break even point for Sprint is immaterial. My number were also based on your 0.17% much lowered percentage as opposed to my initial (and documented) supposition of a 1.8-4.8% rate that all other terrestrial ISP's charge for their bandwidth. 0.17% vs 1.8-48% is a substantial difference in a magnitude of 10x-463x less.

 

My napkin math will remain for the moment, unless someone can give me hard numbers from all network traffic network-wide since this is a rough and fair estimation of the midpoint for reasonable expectations on both a consumer and network side of the equation. Any alteration due to increased capacity, sectors, network density, spectrum efficiency, subs per sector, et al would only bring the 6.66-10GB/mo number up. This certainly doesn't factor in the 800mhz or 2.5ghz or any other "potential" spectrum advantages and was based purely on what network vision in itself can provide with LTE 1900 on all 38,000 towers.

 

This is obviously subject to very wide interpretation and speculation, but I think that by erring on an ultra-conservative side with nowhere near max utilization of available resources this is a fair representation of where we can all draw the line on what is or is not excessive use/allotment.

  • Like 1
Link to comment
Share on other sites

I try my best to only use my unlimited data at night like 10pm-5am. Though I still use way more than I should. It's what happens when you only have one ISP able to get to your house and they don't give a crap. I can't use youtube much on it with out them throttling me. Also using more than 100-150GB will do it. It's a soft cap that they say they don't have, but it's there. But mostly it tends to fail alot even after replacing two modems and cables.

 

I have a unlimited At&t data plan and they have been very nice to me, even with 150GB of data use they didn't slow me down. I only used it at night and my unlimited data Verizon line in the day, but only use about 20-30gb a month.

Link to comment
Share on other sites

I try my best to only use my unlimited data at night like 10pm-5am. Though I still use way more than I should. It's what happens when you only have one ISP able to get to your house and they don't give a crap. I can't use youtube much on it with out them throttling me. Also using more than 100-150GB will do it. It's a soft cap that they say they don't have, but it's there. But mostly it tends to fail alot even after replacing two modems and cables.

 

I have a unlimited At&t data plan and they have been very nice to me, even with 150GB of data use they didn't slow me down. I only used it at night and my unlimited data Verizon line in the day, but only use about 20-30gb a month.

 

How are you using 100gb a month on a wireless connection? What ISP do you have that throttles you?

 

Makes me glad I live in an urban area where Verizon has deployed FiOS.

Link to comment
Share on other sites

 

 

How are you using 100gb a month on a wireless connection? What ISP do you have that throttles you?

 

Makes me glad I live in an urban area where Verizon has deployed FiOS.

 

At&t throttles DSL at 150gb and uverse at 250gb, though I'm not sure it's actually implemented, they haven't even figured out a way to meter it last I checked.

Link to comment
Share on other sites

At&t throttles DSL at 150gb and uverse at 250gb, though I'm not sure it's actually implemented, they haven't even figured out a way to meter it last I checked.

 

Ahh, good to know.

 

That's insane, using Netflix and Hulu would eat that up without batting an eye.

Link to comment
Share on other sites

Ahh, good to know.

 

That's insane, using Netflix and Hulu would eat that up without batting an eye.

 

I have had found my softcap (gasp 400gb/mo) reached as well. they do tend to throttle on a basis of overall network usage. if you are a tech friendly family that enjoys streaming whether, amazon, vudu, hulu or netflix and also have many devices such as tablets and smartphones that are regularly wifi offloaded you can hit the soft cap (which usually varies by the tier of internet you pay for) pretty damn quick

Link to comment
Share on other sites

How are you using 100gb a month on a wireless connection? What ISP do you have that throttles you?

 

Makes me glad I live in an urban area where Verizon has deployed FiOS.

 

I have a unlimited AT&T wireless data plan that does not throttle. Also youtube, netflix and hulu to a 46in HD TV will use the data very fast.

 

Time warner cable says they don't throttle, but there network is so poor that they may as well be throttling me. They block web sites like youtube/netfilx if you use it too much. many users have been getting nailed by time warner cable lately. If you youtube "throttling" time warner cable is at the top. Even higher than AT&T throttling it's so bad.

 

They limit it to the point even 140p video is unplayable on youtube.

 

But at lest I have 0 packet loss now. (Before it was 58-60%. You don't even want to know how bad that was. FIO's guys count your lucky stars, you have it made.

Link to comment
Share on other sites

Guest
This topic is now closed to further replies.

×
×
  • Create New...