IEEE Vet: Carriers Capping LTE Services To Avoid Fixed-line Cannibalization 118
alphadogg writes "Roberto Saracco isn't buying carriers' claims that they need to put data caps on their LTE services due to excessive traffic causing massive engineering challenges. Saracco, a senior member of the IEEE and the director of the Telecom Italia Future Centre, said during an interview Tuesday that the major reason carriers are placing data caps on their LTE services is to prevent users from going exclusively with wireless data services and ditching their landline connections. 'You're always going to want to make the maximum amount of value,' he said. 'And you don't want to have your fixed-line network being cannibalized by mobile.'"
Idiot (Score:4, Informative)
With any wireless service, you have a limited amount of bandwidth. That bandwidth is shared by everybody connecting to a tower.
If you have more than 1 person trying to use as much bandwidth as they can, it will just degrade the service for everyone.
You could get another frequency to operate on, or use more directional antennas so that less people connect to each transmitter, but that will only delay the inevitable.
In these days of ever increasing bandwidth demands, there is no way that wireless can supply that demand.
With a wired connection, you can add more cables.
With a wireless connection, that is not an option.
That's what people don't seem to understand (Score:5, Informative)
There is just a limited amount of bandwidth you can have. Look up Shannon's Law and then realize that is what you are up against. To increase the amount of bits per second you get you can either increase frequency bandwidth, which is infeasible with wireless since you have to share with others, or increase SNR, which is infeasible with wireless since there is natural noise all around. So basically there is only so good it gets. Now technologies like MIMO (having multiple antennas on transmitters and receivers to get array gain) can help increase what you get, at the cost of more complex transceivers, but still, there's only so much. It isn't as though you can just say "Let's do 100 spatial streams, no problem!" and it'll work.
Also that bandwidth? Shared with EVERYONE on the segment. So if you have 100mbps of effective WiFi throughput and you have 10 users on the AP, you all share it. It isn't like a wired connection where you each have that amount of bandwidth back to the switch, you all share the same 100mbps. So if four people are going full bore, you get only 25mbps each max (in reality, the more that share the less you get because of collisions).
Only thing to be done is to build out the network, make segments smaller. That is feasible and a good idea to an extent, but can only be done so much. Remember that you have to have antennas and hardware for each segment, there are a limited number of places you can stick all that. Also you have to have a cable going to it. The ultimate in segmentation might be Picocells, little devices you can hook to your net connection to provide local coverage for your house. Of course you need a wired network connection so...
All that and the best wireless can't compare to wired at all. Take the consumer space: The best you can get right now is 802.11n. If you have a 5GHz network (which reduces range) 40MHz channel you will get at best 150mbps of raw signaling per stream, with 4 streams max, though in actual practice I've never seen equipment with more than 3. So 450mbps raw signaling best case. Effective throughput you see probably 150-200mbps tops, wireless has a lot of overhead between link rate and effective rate. Oh and all that applies in a homogenous n network only, no older clients on it. Also not really anything faster for general use, this is as good as it gets right now.
Wired? 1gbps, full duplex no problem. Getting hard to buy equipment that doesn't support that speed anymore. So long as you cables aren't too long (100 meters or less) or kinked/broken, you'll really get it too. You can come very close to the theoretical speed without much difficulty. Works fine in a heterogeneous network, older clients don't mess with newer ones.There's faster out there too, 10gbps is real and working well. At this point it is still too expensive for consumer use, but price is the only barrier, the tech is finalized and released. Price has been dropping rapidly too.
What it comes down to is wired is a good idea, if we want to be able to have lots of bandwidth. On wireless connections, we need to be nice and share more which means less heavy bandwidth stuff. The wider area the wireless connection, the more true that is. WiFi isn't bad, its range is pretty short, you don't tend to have a ton of people on one AP (though at an office it can get heavy at times). Still though if you regularly do large file transfers with servers, you'll want to go wired as WiFi will start dragging, particularly with multiple users doing it.
LTE, you need to be share even more. On a 20MHz channel, if you are lucky enough to have that big of one for your service in your area, you can get 802.11n like data rates (in ideal conditions) but you share with many more people. You could easily have 100+ people on the same segment, meaning that you have to share that much more. If all those people try to go full blast, speed will quickly plummet.
If you had everyone using LTE in their homes and trying to do something like stream Netflix HD, it just wouldn't work. The cells would
Re:Not mobile (Score:4, Informative)
Free landline internet to those that make their internet connection available to mobile users of the same carrier.
At least one UK broadband provider provide WiFi routers that present two networks: a private one, and a less-private one. The less-private one is available for use by anyone with that provider (so in return for potentially sharing your bandwidth, you can potentially get free WiFi. But probably only in residential areas.)
http://www.bt.com/btfon [bt.com] is one, but I think there's another.
Re:That's what people don't seem to understand (Score:4, Informative)
SSH isn't bad anymore if you have a new Intel chip, and software smart enough to make use of it. AES-NI is no joke, if you've a CPU with it you get amazing amounts of throughput with little usage. If you have an AES-NI capable processor (any 32nm Core i series except i3, and the new 22nm Ivy Bridge chips basically) you can test it real quick with Truecrypt. Have it run a benchmark and be amazed at the AES speeds.
Also with regards to SNR you can't increase S, at least not in the US. Power limits are set by the FCC. Never mind any technical problems (and there are many with trying to use powerful transmissions) you just aren't allowed to do so for cell networks. They have low power caps.