the problem here is that it IS illegal in Canada to retain that data, even if it was handed to you (in this case via unencrypted broadcasts!). to protect people from themselves, Canada maintains laws that a company (or person/group) can only KEEP data that's not personally identifiable, unless the person they gathered information about waves that right.
it astounds me how often people just can't wrap their mind around this!
I honestly know people that refuse to backup files that they NEED because they think of them like paper documents. "if it's still there today, it'll still be there tomorrow. it's in a safe place!"
I even have a friend who teaches that thinks that he can't backup student's submissions because, get this: he think's it's plagiarism. he feels that making copies of documents is like taking it for your own.
ever time I yell at him, telling him what a horrible idea it is to keep everything on a single external hard drive. does he listen? not yet!
quite randomly: a large customer of mine was on it's knees last night for much this same reason. our major carrier (who does NOTHING to provide uptime guarantees, but the owner decided on them based on price) suffered an unexpected outage last night at 8:00PM due to an electric pole being hit by a car. the 2 hour UPS in the NID died at 10:00PM, and their phones were out until 3:00AM, preventing them from responding to alarms for that period of time.
suffice it to say: if they won't budget for guaranteed up time after that event: I'll walk away from them. no interest in working for a company who leaves peoples lives at risk because they can't be bothered to put their money where their mouth is.
If those 300,000 users switched right now to using 3G (read "4G speeds") regular T-Mobile phones, wouldn't that cause more strain on the network and not the other way around?
No, not at all. it would be exactly the same issue.
regardless of what speed the phone/tower can communicate at: they still can only talk about what frequency to use, when to send data, etc at a set rate. these handsets are trying to use that control frequency more than others, meaning they're slowing other people down.
for example, in cases like twitter: as much as we as users want to be able to send 60 "are there any new messages?" a minute, that would be like plugging twenty computers at home into a hub and running once/second twitter clients: and being surprised when they slow each other down. in this case, the air around us is the hub. until we find a way to better segment the air: there's not much of it to go around to all the people that want to use it.
I'm glad to read a comment from somebody that understands this topic. thank you for contributing!
it's the equivalent issue to storming a million dollar CRS-1 with 24 byte packets. it's wonderful that the beast can handle 2 Million 1500byte packets per second: but it's not going to do 125 million 24 byte packets: it's still only going to do 2 million 24 byte packets. (IF that!)
Airtime is the limiting factor here: not bandwidth.
as mentioned a pile of times in the comments here: it's not a bandwidth issue.
if you work in DSL, then you've rarely had spectrum to ever worry about. DSL uses what, less than a single megahertz in bandwidth? the entire voice spectrum is 20KHz, (before compression) I think (I'm not in DSL, I could be wrong) VDSL2 has a whopping 30MHz per loop. take a look at cablelabs: they manage to squeeze 500 customers onto a single RG11 trailer oscillating in the neighborhood of 4GHz. they initially suffered from similar issues to this: if somebody had an open cable device, they could send out broadcasts on every RF band, and consume all the channels of bandwidth to prevent other cable customers from getting anything! they solved this, by porting each customer into the cable equivalent of a switch, through the use of frequency filters. (now each customer can only broadcast on their set frequency)
unfortunately: you can't segment the air. Air works like a large hub, with only 32 total ports (under the current GSM standard) if you think of it like that, each cell phone virtually reaches out to the tower and plugs it's wire into a port that's available. if the tower doesn't respond: then all the ports were full. if it does, it assigns you a port number, for a set length of time. you do what you need to, and then unplug and carry on.
the problem here is that open handsets can plug and unplug more often than other handsets, meaning that they consume more access than they "should" degrading the hub for other customers. maybe surprisingly: when all handsets are strictly regulated by the carrier, that 32 port hub can be shared by as many as twenty thousand people, with little to no noticeable problem. (ie: they know how to stand in line, they each know they get the same treatment, etc)
in this EXACT CASE: the issue was with the one control port. the air hubs have 32 ports, but only one is a control port. everybody issues commands to the one port constantly: stuff like setup and tear down (which port to use and for how long you have it), SMS, and network status all flow over this port 24/7. open handsets are able to issue commands to it more than the protocol expects them to, so it chokes.
with your analogy it would be like saying: "I'm sorry, you can't connect your PS3 with it's ten Ethernet ports to my 32 port hub. other customers need those other 9 ports."
Finland is a whopping 338,000KM^2. that's what, 250KM by 1350KM?
North america is 73 times bigger than Finland. (not that the shape is ANYTHING similar,) that's about 18,250KM X 98,550KM
we have issues. to run fiber ALONE to a tower on the other coast would require ~18,000 KM of Fiber, with 180 powered repeaters. at $20/meter installed (regular here for 32-48 strand, powered) that would run you a bill of $360 Million, for a line just from the Pacific to the Atlantic. that band, (if well planned) will reach a potential of 4% of the land mass (4% will be within 10KM of the nearest access to that fiber)
for the same money, you could put fiber within 10KM of EVERY KM^2 in Finland.
sure. and if you know basic electronics, some simple physics and a little RF theory: you can build a transmitter that can shut down communication in most RF bands by elevating the noise floor so high equipment has no ability to filter it out.
a rooted phone that you buy at the store requires an 8 year old to figure out and even accidentally cause confusion on a network. (though I'm sure you can find an 8 year old who is skilled enough to do it from scratch, he'll be likely one in a billion)
face it, people are educated to be worker drones these days. but even a drone will have an hour or two to play with a rooted phone at a command line, the average person just wants the thing to work though.
Yeah, they do: but so does everyone else. they all just buy the industry standard equipment within their budget: and sell access to it. if there was better equipment for decent prices, they'd consider buying it: keeping customers happy DOES mean they make more money. (though in many cases like this, it's cheaper/easier to just get regulations in place, it's easy to tell a customer: "sorry, you can't do that with your phone on our network: but you can't do it on any network! it's regulated!"
the problem with walled gardens is that they all share the same air. (and the soil to some extent.)
like phone providers: there's only so much of it to go around. when density increases, you have to tighten and restrict the walls further and further, else more and more things get in and out.
that's a pretty awesome comparison. did you ever see the coverage of the Niue installation of the OpenBTS project? they required a single antenna at 125W to cover the entire 260KM^2 island.
there was NO way to get a carrier to spend the FORTUNE to install a cell tower for ~1700 people. these days, they'd hardly even know how to do it for such a small area. even if you had 100% of the islanders, on a $50/month plan, it'll JUST cover your maintenance +bandwidth costs. you'll be out the install costs for years to come before it ever turns a profit.
see that band covering all of 9% of canada? that's it. we have a few other providers that provide CDMA coverage across ~50% of SK/MB/ON but none of them will handshake off GSM handsets. at any given time, (Even in some major cities!) you seriously may be on one single tower, with the rest of the entire town.
I did the math somewhere in this comment thread:
to cover 100% of the area in germany, would require ~58 towers. (~357KM^2 total area / 6100KM^2 single cell size@750W/sector = ~58) assuming population maintains the current density of 248 people/KM^2
20% of north america, (which ABOUT covers the MAJOR population centers), would require 809 towers.
I'm not saying that putting up ONE tower SHOULD cover such a huge area, but if you push a BTS to 1.8KW/sector, you get a cell 90KM in diameter. (and SHITTY battery life as each handset needs to use the 2W 850MHz band to ever reach the tower) yielding 20,000KM^2 bubbles. for ~$450K installed and less than $50K a year in maintenance, a "startup" in Europe can place the tower within range of a major city (like greater london) and blanket the surrounding area: yielding as many as 16 million potential customers. if you get 5%, that's 800K people: at a $10/month plan with shitty limits, that's $8 Mil a month in revenue. (yes, 800,000 people sharing one cell that's only build to handle ~250/min will suck. but for $10 a month, who cares!:P)
here in north america, people travel a much greater distance just to go home. if you don't cover their work AND home, you have no chance to sell that consumer. meaning that carriers constantly push old technology to it's limits to cover their customers geographical restrictions. it's not unusual here to find that 5-10% of the employees in a larger origination will travel much more than 75KM just to get TO work. you know how large a cell that would take to cover?
The issue with looking at it like a signaling system has plagued us for years, unlike a network that you can add capacity in any form you like: (even in SS7, you can always add an additional signal channel) in radio communications: you can't. it's like dealing with Multicast:
if everybody joins the group: everybody has to listen to every conversation and "hopefully" will ignore each other. this includes towers!
I'm TOTALLY with you on not advertising it if you can't support it: but when your competition advertises it, (even through they're lying through their teeth) it's kinda hard not to. it's hard for a marketing dude person to watch hundreds of thousands of subscribers leave, (through a company memo weekly saying "why is this happening? is it your fault?") because he feels "it'll hurt the company more to lie about it". totally a sad though. when faced with "how can we stay in business, how can I help you keep your job" what would anybody else in marketing do?
the tracking is provided by getting an echo from the handsets. if you've rooted the handset: you just don't reply to the echos. (or better yet, fake IMEI/SIM combos, and the real phones WILL respond at incorrect locations.)
it's all about protocol manipulation. if you know how to talk the talk, people (devices) will listen, and do as instructed.
Germany is a whopping 357,021KM^2. hosting a population density of ~248 people per square kilometer.
Manitoba, (ONE of the provinces in Canada) is 649,950KM^2. one province alone. we boast one of the LOWEST population densities in the world: at 1.9 people per square KM.
North america is a staggering 24.7 Million square kilometers. with a total population density of 22.9 people per square kilometer.
to even cover 20% of North america with cell phone service, would require 800 towers. at an average cost to deploy of $300K for the shitty old 2G equipment, and an installation cost running likely near $500K to get your fiber trenched from your NOC to the remote tower: that would require $640.7 Million to deploy.
the cost to cover 100% of germany with $500K devices, and $300K installs (shorter fiber runs) is half. (about $360 Million)
the excuse is simple. do you know anything about how GSM works?
you have a whopping four approved frequency ranges. in each of those ranges, you have eight full 33.75Kb/s channels per band, with frames every 4.6ms.
from that, you have to timeslot a HUGE number of users. it's one thing to say: "just put up more towers!" but every one of those towers share the same 32 channels. as much as you want to pack population density: there's no way to go much beyond 400 devices/band/geographical area. and that requires a LOT of multiplexing. allowing for a whopping 46 "medium quality" voice calls.
until we get a HUGE chunk of spectrum, or manage to overcome much of known physics: it's not going to get much better. if everybody used audio frequency to communicate at once at a density like that, you'd never be able to understand anyone. it's just be a constant drone.
when you do it that way, you establish a DS0 (or greater) signaling channel between you and an IP gateway via a control message.
when you disconnect, it sends the control message to close the channel.
a cell phone, to save battery life, does this and disconnects every time you want something. every request, every "sync inbox" every "refresh tweets". everything involves control messages due to how the Signaling System works.
The towers were not built to deal with this. they were build with what is basically a 28K modem for control, and some number of DS0 channels for data.
they have no issue with you using your 5GB of transfer a month. what they have issue with is that they can't include the extra ~0.3GB in control protocol overhead in your bill.
also: when you try to issue 1GB in control messages a month, and only use 1GB in bandwidth with that 1GB in control. you can't charge for bits that needed to get transfered due to their network errors, or for the ICMP keep-alives the towers send to the phones. they're complaining that they can't charge for the control: because newer open devices are capable of more then they had planned for in overhead.
it's illegal to prevent a handset from sending a 911 call. flat out: illegal.
so when you root 10 handsets and have them try to sms/establish 21 - 911 calls a minute, what do you do? you legally can't stop them, but they've now consumed your entire T1 backhaul? what's the call?
So the carriers in the US must have total control or their network is going to explode, eh? How is it that you can buy whatever device you want and connect it to whatever network you want here in Europe, eh? Why haven't the mobile networks in the EU exploded yet, then, eh?
Because all the devices you have follow the protocol that all the carriers agree on.
if you had a rack of ~100 cell phones, that were all able to broadcast rotating IMEI's and SIM ID's of known users, you could single handedly collapse the entire infrastructure for minutes to hours at a time. and based on how the protocol works, it would wreck havoc on the providers. but being a good bunch of people, you don't! so it all just works.
here in north america: you may only have one tower in reach of you at any time (outside of a major city) meaning that you can do exactly the above with a single rooted handset.
for the record: I'm an android developer, and I fully, FULLY support open handsets globally. however: people need to know how brittle the system is: and without spending hundreds of millions of dollars in patch jobs: it will only get thinner and worse over time.
remember, the issue they were complaining about was protocol, not bandwidth.
the problem they're having would be the same as allowing somebody to attach an FPGA with an ethernet jack to your CAT-6500. as much as you want to limit the port to only allow certain communication: there's nothing stopping the node from abusing the protocol.
the major difference here is that: the GSM stack assumes you trust your Mobile Stations (MS) because when it was written: telco's had full control over them. TCP/IP has no such restrictions (though I'm sure if you look at the standard long enough you'll find that you can send a physical signal to them in a particular way they will respond in a fashion you don't expect.) the analogy here would be that a GSM BTS acts much like a hub (almost exactly) and when one user keeps storming it with packets, nobody else get's a chance to talk.
personally: I don't see this as either a telco OR a user issue: it's a protocol issue. we need to take another look at the wireless communication protocol, and find another way of allowing untrusted users their fair access to the medium. (in this case, a pretty narrow band of wireless spectrum). telco's need to push standard-makers/suppliers for something they can sell these days: not some old outdated protocol that allows things like this to happen.
unfortunately: Air IS like a hub: only one person can talk on it at a time. so it'll be a hard fix.
Slashdot Mirror
the problem here is that it IS illegal in Canada to retain that data, even if it was handed to you (in this case via unencrypted broadcasts!). to protect people from themselves, Canada maintains laws that a company (or person/group) can only KEEP data that's not personally identifiable, unless the person they gathered information about waves that right.
and might survive the next hundred or so years without our hearts exploding. :P
it astounds me how often people just can't wrap their mind around this!
I honestly know people that refuse to backup files that they NEED because they think of them like paper documents. "if it's still there today, it'll still be there tomorrow. it's in a safe place!"
I even have a friend who teaches that thinks that he can't backup student's submissions because, get this: he think's it's plagiarism. he feels that making copies of documents is like taking it for your own.
ever time I yell at him, telling him what a horrible idea it is to keep everything on a single external hard drive. does he listen? not yet!
Somehow I don't think an officer would understand: "all I did was DD the contents to an ISO!"
which scares me. basic IT != rocket science.
ever try to use skype from your 3G phone to make a 911 call while being chased for your life?
funny: it's sure easy to do with a $50 unlocked Nokia POS phone.
sometimes there's more to a service than one single feature.
quite randomly: a large customer of mine was on it's knees last night for much this same reason. our major carrier (who does NOTHING to provide uptime guarantees, but the owner decided on them based on price) suffered an unexpected outage last night at 8:00PM due to an electric pole being hit by a car. the 2 hour UPS in the NID died at 10:00PM, and their phones were out until 3:00AM, preventing them from responding to alarms for that period of time.
suffice it to say: if they won't budget for guaranteed up time after that event: I'll walk away from them. no interest in working for a company who leaves peoples lives at risk because they can't be bothered to put their money where their mouth is.
If those 300,000 users switched right now to using 3G (read "4G speeds") regular T-Mobile phones, wouldn't that cause more strain on the network and not the other way around?
No, not at all. it would be exactly the same issue.
regardless of what speed the phone/tower can communicate at: they still can only talk about what frequency to use, when to send data, etc at a set rate. these handsets are trying to use that control frequency more than others, meaning they're slowing other people down.
for example, in cases like twitter: as much as we as users want to be able to send 60 "are there any new messages?" a minute, that would be like plugging twenty computers at home into a hub and running once/second twitter clients: and being surprised when they slow each other down. in this case, the air around us is the hub. until we find a way to better segment the air: there's not much of it to go around to all the people that want to use it.
I'm glad to read a comment from somebody that understands this topic. thank you for contributing!
it's the equivalent issue to storming a million dollar CRS-1 with 24 byte packets. it's wonderful that the beast can handle 2 Million 1500byte packets per second: but it's not going to do 125 million 24 byte packets: it's still only going to do 2 million 24 byte packets. (IF that!)
Airtime is the limiting factor here: not bandwidth.
as mentioned a pile of times in the comments here: it's not a bandwidth issue.
if you work in DSL, then you've rarely had spectrum to ever worry about. DSL uses what, less than a single megahertz in bandwidth? the entire voice spectrum is 20KHz, (before compression) I think (I'm not in DSL, I could be wrong) VDSL2 has a whopping 30MHz per loop. take a look at cablelabs: they manage to squeeze 500 customers onto a single RG11 trailer oscillating in the neighborhood of 4GHz. they initially suffered from similar issues to this: if somebody had an open cable device, they could send out broadcasts on every RF band, and consume all the channels of bandwidth to prevent other cable customers from getting anything! they solved this, by porting each customer into the cable equivalent of a switch, through the use of frequency filters. (now each customer can only broadcast on their set frequency)
unfortunately: you can't segment the air. Air works like a large hub, with only 32 total ports (under the current GSM standard) if you think of it like that, each cell phone virtually reaches out to the tower and plugs it's wire into a port that's available. if the tower doesn't respond: then all the ports were full. if it does, it assigns you a port number, for a set length of time. you do what you need to, and then unplug and carry on.
the problem here is that open handsets can plug and unplug more often than other handsets, meaning that they consume more access than they "should" degrading the hub for other customers. maybe surprisingly: when all handsets are strictly regulated by the carrier, that 32 port hub can be shared by as many as twenty thousand people, with little to no noticeable problem. (ie: they know how to stand in line, they each know they get the same treatment, etc)
in this EXACT CASE: the issue was with the one control port. the air hubs have 32 ports, but only one is a control port. everybody issues commands to the one port constantly: stuff like setup and tear down (which port to use and for how long you have it), SMS, and network status all flow over this port 24/7. open handsets are able to issue commands to it more than the protocol expects them to, so it chokes.
with your analogy it would be like saying: "I'm sorry, you can't connect your PS3 with it's ten Ethernet ports to my 32 port hub. other customers need those other 9 ports."
Finland is a whopping 338,000KM^2. that's what, 250KM by 1350KM?
:P
North america is 73 times bigger than Finland. (not that the shape is ANYTHING similar,) that's about 18,250KM X 98,550KM
we have issues. to run fiber ALONE to a tower on the other coast would require ~18,000 KM of Fiber, with 180 powered repeaters. at $20/meter installed (regular here for 32-48 strand, powered) that would run you a bill of $360 Million, for a line just from the Pacific to the Atlantic. that band, (if well planned) will reach a potential of 4% of the land mass (4% will be within 10KM of the nearest access to that fiber)
for the same money, you could put fiber within 10KM of EVERY KM^2 in Finland.
you know, I kind of want to move to Finland.
sure. and if you know basic electronics, some simple physics and a little RF theory: you can build a transmitter that can shut down communication in most RF bands by elevating the noise floor so high equipment has no ability to filter it out.
a rooted phone that you buy at the store requires an 8 year old to figure out and even accidentally cause confusion on a network. (though I'm sure you can find an 8 year old who is skilled enough to do it from scratch, he'll be likely one in a billion)
face it, people are educated to be worker drones these days. but even a drone will have an hour or two to play with a rooted phone at a command line, the average person just wants the thing to work though.
Yeah, they do: but so does everyone else. they all just buy the industry standard equipment within their budget: and sell access to it. if there was better equipment for decent prices, they'd consider buying it: keeping customers happy DOES mean they make more money. (though in many cases like this, it's cheaper/easier to just get regulations in place, it's easy to tell a customer: "sorry, you can't do that with your phone on our network: but you can't do it on any network! it's regulated!"
the problem with walled gardens is that they all share the same air. (and the soil to some extent.)
like phone providers: there's only so much of it to go around. when density increases, you have to tighten and restrict the walls further and further, else more and more things get in and out.
that's a pretty awesome comparison. did you ever see the coverage of the Niue installation of the OpenBTS project? they required a single antenna at 125W to cover the entire 260KM^2 island.
there was NO way to get a carrier to spend the FORTUNE to install a cell tower for ~1700 people. these days, they'd hardly even know how to do it for such a small area. even if you had 100% of the islanders, on a $50/month plan, it'll JUST cover your maintenance +bandwidth costs. you'll be out the install costs for years to come before it ever turns a profit.
No, as I stated in many location in europe, you can cover many countries with a single tower, where as in North america you can cover a single city.
:P)
:P)
(I did sound a little hazy there, it was late! I'm sorry!
and FYI: the dominant GSM carrier in Canada (Rogers) maintains very few towers. you ever see a GSM coverage map of north america? http://www.rogers.com/web/content/wireless_network
see that band covering all of 9% of canada? that's it. we have a few other providers that provide CDMA coverage across ~50% of SK/MB/ON but none of them will handshake off GSM handsets. at any given time, (Even in some major cities!) you seriously may be on one single tower, with the rest of the entire town.
I did the math somewhere in this comment thread:
to cover 100% of the area in germany, would require ~58 towers. (~357KM^2 total area / 6100KM^2 single cell size@750W/sector = ~58) assuming population maintains the current density of 248 people/KM^2 20% of north america, (which ABOUT covers the MAJOR population centers), would require 809 towers.
I'm not saying that putting up ONE tower SHOULD cover such a huge area, but if you push a BTS to 1.8KW/sector, you get a cell 90KM in diameter. (and SHITTY battery life as each handset needs to use the 2W 850MHz band to ever reach the tower) yielding 20,000KM^2 bubbles. for ~$450K installed and less than $50K a year in maintenance, a "startup" in Europe can place the tower within range of a major city (like greater london) and blanket the surrounding area: yielding as many as 16 million potential customers. if you get 5%, that's 800K people: at a $10/month plan with shitty limits, that's $8 Mil a month in revenue. (yes, 800,000 people sharing one cell that's only build to handle ~250/min will suck. but for $10 a month, who cares!
here in north america, people travel a much greater distance just to go home. if you don't cover their work AND home, you have no chance to sell that consumer. meaning that carriers constantly push old technology to it's limits to cover their customers geographical restrictions. it's not unusual here to find that 5-10% of the employees in a larger origination will travel much more than 75KM just to get TO work. you know how large a cell that would take to cover?
Well, I for sure haven't. out of interest... have you really? cause that sounds like the weirdest thing to "accidentally do". :P
The issue with looking at it like a signaling system has plagued us for years, unlike a network that you can add capacity in any form you like: (even in SS7, you can always add an additional signal channel) in radio communications: you can't. it's like dealing with Multicast:
if everybody joins the group: everybody has to listen to every conversation and "hopefully" will ignore each other. this includes towers!
I'm TOTALLY with you on not advertising it if you can't support it: but when your competition advertises it, (even through they're lying through their teeth) it's kinda hard not to. it's hard for a marketing dude person to watch hundreds of thousands of subscribers leave, (through a company memo weekly saying "why is this happening? is it your fault?") because he feels "it'll hurt the company more to lie about it". totally a sad though. when faced with "how can we stay in business, how can I help you keep your job" what would anybody else in marketing do?
and how do you propose you track them down?
the tracking is provided by getting an echo from the handsets. if you've rooted the handset: you just don't reply to the echos. (or better yet, fake IMEI/SIM combos, and the real phones WILL respond at incorrect locations.)
it's all about protocol manipulation. if you know how to talk the talk, people (devices) will listen, and do as instructed.
Germany is a whopping 357,021KM^2. hosting a population density of ~248 people per square kilometer.
Manitoba, (ONE of the provinces in Canada) is 649,950KM^2. one province alone. we boast one of the LOWEST population densities in the world: at 1.9 people per square KM.
North america is a staggering 24.7 Million square kilometers. with a total population density of 22.9 people per square kilometer.
to even cover 20% of North america with cell phone service, would require 800 towers. at an average cost to deploy of $300K for the shitty old 2G equipment, and an installation cost running likely near $500K to get your fiber trenched from your NOC to the remote tower: that would require $640.7 Million to deploy.
the cost to cover 100% of germany with $500K devices, and $300K installs (shorter fiber runs) is half. (about $360 Million)
THAT'S why we have shit.
the excuse is simple. do you know anything about how GSM works?
you have a whopping four approved frequency ranges. in each of those ranges, you have eight full 33.75Kb/s channels per band, with frames every 4.6ms.
from that, you have to timeslot a HUGE number of users. it's one thing to say: "just put up more towers!" but every one of those towers share the same 32 channels. as much as you want to pack population density: there's no way to go much beyond 400 devices/band/geographical area. and that requires a LOT of multiplexing. allowing for a whopping 46 "medium quality" voice calls.
until we get a HUGE chunk of spectrum, or manage to overcome much of known physics: it's not going to get much better. if everybody used audio frequency to communicate at once at a density like that, you'd never be able to understand anyone. it's just be a constant drone.
when you do it that way, you establish a DS0 (or greater) signaling channel between you and an IP gateway via a control message.
when you disconnect, it sends the control message to close the channel.
a cell phone, to save battery life, does this and disconnects every time you want something. every request, every "sync inbox" every "refresh tweets". everything involves control messages due to how the Signaling System works.
The towers were not built to deal with this. they were build with what is basically a 28K modem for control, and some number of DS0 channels for data.
they have no issue with you using your 5GB of transfer a month. what they have issue with is that they can't include the extra ~0.3GB in control protocol overhead in your bill.
also: when you try to issue 1GB in control messages a month, and only use 1GB in bandwidth with that 1GB in control. you can't charge for bits that needed to get transfered due to their network errors, or for the ICMP keep-alives the towers send to the phones. they're complaining that they can't charge for the control: because newer open devices are capable of more then they had planned for in overhead.
how do you propose they lock out those devices?
it's illegal to prevent a handset from sending a 911 call. flat out: illegal.
so when you root 10 handsets and have them try to sms/establish 21 - 911 calls a minute, what do you do? you legally can't stop them, but they've now consumed your entire T1 backhaul? what's the call?
So the carriers in the US must have total control or their network is going to explode, eh? How is it that you can buy whatever device you want and connect it to whatever network you want here in Europe, eh? Why haven't the mobile networks in the EU exploded yet, then, eh?
Because all the devices you have follow the protocol that all the carriers agree on.
if you had a rack of ~100 cell phones, that were all able to broadcast rotating IMEI's and SIM ID's of known users, you could single handedly collapse the entire infrastructure for minutes to hours at a time. and based on how the protocol works, it would wreck havoc on the providers. but being a good bunch of people, you don't! so it all just works.
here in north america: you may only have one tower in reach of you at any time (outside of a major city) meaning that you can do exactly the above with a single rooted handset.
for the record: I'm an android developer, and I fully, FULLY support open handsets globally. however: people need to know how brittle the system is: and without spending hundreds of millions of dollars in patch jobs: it will only get thinner and worse over time.
This has nothing to do with bandwidth
remember, the issue they were complaining about was protocol, not bandwidth.
the problem they're having would be the same as allowing somebody to attach an FPGA with an ethernet jack to your CAT-6500. as much as you want to limit the port to only allow certain communication: there's nothing stopping the node from abusing the protocol.
the major difference here is that: the GSM stack assumes you trust your Mobile Stations (MS) because when it was written: telco's had full control over them. TCP/IP has no such restrictions (though I'm sure if you look at the standard long enough you'll find that you can send a physical signal to them in a particular way they will respond in a fashion you don't expect.) the analogy here would be that a GSM BTS acts much like a hub (almost exactly) and when one user keeps storming it with packets, nobody else get's a chance to talk.
personally: I don't see this as either a telco OR a user issue: it's a protocol issue. we need to take another look at the wireless communication protocol, and find another way of allowing untrusted users their fair access to the medium. (in this case, a pretty narrow band of wireless spectrum). telco's need to push standard-makers/suppliers for something they can sell these days: not some old outdated protocol that allows things like this to happen.
unfortunately: Air IS like a hub: only one person can talk on it at a time. so it'll be a hard fix.