Not logged in
Log in now
Create an account
Subscribe to LWN
An unexpected perf feature
LWN.net Weekly Edition for May 16, 2013
A look at the PyPy 2.0 release
PostgreSQL 9.3 beta: Federated databases and more
LWN.net Weekly Edition for May 9, 2013
I keep hoping that phone communication will move to VOIP over HSPA+ or whatever data connection you have. Rather than trying to fix GSM (what looks to be an IMMENSE project), just leave it behind.
Guess that's kind of a long shot.
GSM security testing: where the action is
Posted Sep 27, 2010 6:10 UTC (Mon) by dambacher (subscriber, #1710)
Posted Sep 27, 2010 6:13 UTC (Mon) by rahvin (subscriber, #16953)
Posted Sep 27, 2010 6:58 UTC (Mon) by Cyberax (✭ supporter ✭, #52523)
Cell phone towers are mass-produced and are pretty cheap. You probably can't make anything much cheaper as it is on large scale. On small scale it might be possible, but it'll still probably be cheaper to buy a nanocell.
And while GSM is nice, it's now clear that the 'path forward' is 4G which is going to use VoIP over an IP data network.
Posted Sep 27, 2010 8:21 UTC (Mon) by freddyh (subscriber, #21133)
Posted Sep 27, 2010 8:37 UTC (Mon) by Cyberax (✭ supporter ✭, #52523)
As for competition, there's enough of it in the hardware cell-tower market. They are produced in millions, and basic GSM cell phone infrastructure (i.e. no GPRS, EVDO, etc.) is now VERY cheap on per-subscriber basis. Use of non-carrier-grade hardware is quite likely to _increase_ costs because of increased maintenance overhead.
No, the next battlefield will be 4G technologies. They have the potential to uproot established telecoms. 4G phones will be simple IPv6 nodes, and calls between them are going to be simple VoIP streams over data network which can be (by design) transmitted over the WiFi and public Internet.
PS: I worked as a network architect in a small cell-phone company.
Viability of open GSM stacks and equipment
Posted Sep 27, 2010 12:39 UTC (Mon) by sladen (subscriber, #27402)
Yes, going forward, the number of modems/radio interfaces available for backhaul on a device is seemingly always increasing (GSM/GPRS/EDGE, Bluetooth, W-CDMA/HSDPA, 802.11b/g, LTE...), but on a phone the GSM Um is going to be present for a very long time virtually anywhere in the world (except in Japan).
Instead of thinking about the costs of initial hardware certification, consider instead every business case that you may have encountered (in designing networks) where a mast had been planned but wasn't economically viable based on the threshold of subscribers in that area. Imagine re-evaluating those business cases with potentially 1/10th of the equipment running costs (power, cooling) and 1/10th of the capital equipment costs.
For a small telco such as Telecom Niue (or perhaps yourself) the turning-over-the-tables of current build-out viability is probably quite attractive, and potentially as disruptive as COTS/open-source has been in the rest of the electronics industry. There are people open to the possibility out there
name a recent smartphone on the market that isn't now running a BSD/GPL Unixy kernel inside it and then project that same degree of confluence onto the infrastructure side.
Posted Sep 27, 2010 13:10 UTC (Mon) by Cyberax (✭ supporter ✭, #52523)
Not likely. AMPS/DAMPS died quickly with the advent of GSM/CDMA, for example (oh, sure there were few holdouts). Once we have viable 4G standards, they're going to spread quickly. There are just too many advantages: cheaper roaming, better data integration, more bandwidth, less power consumption, etc.
>Instead of thinking about the costs of initial hardware certification, consider instead every business case that you may have encountered (in designing networks) where a mast had been planned but wasn't economically viable based on the threshold of subscribers in that area.
Almost never happens. Even at $1 ARPU per month it's economically viable to install a tower even with 50 subscribers. You don't really get lower than that. And if you DO get lower, than most of your investment will be spent on getting uplink anyway (usually a microwave link, which requires, you guessed it, a tower for line-of-sight).
Posted Sep 30, 2010 7:13 UTC (Thu) by Cato (subscriber, #7643)
GSM also has good coverage for rural areas, as a GSM cell can reach up to 35 km (http://en.wikipedia.org/wiki/GSM#Cellular_radio_network ) vs. "over 10km" in theory for UMTS (http://books.google.co.uk/books?id=7m-MnwW_o7AC&lpg=P... ). Even today in a Western European country, the only way I can get UMTS 3G coverage at home is via femtocell, and I'm only 5 km from the nearest 3G base station. The economics of deploying a UMTS or LTE base station closer to me will only stack up if there's a very cheap wireless backhaul technology that can handle the required throughput for multiple 3G/4G subscribers.
Posted Sep 30, 2010 12:54 UTC (Thu) by Cyberax (✭ supporter ✭, #52523)
For GSM maximum distance is limited to about 40km (by lightspeed so you can't do anything about it). It's quite feasible that 4G technologies will allow 50-70km maximum distance to tower.
UMTS is quite far from ideal here. We've learned a lot since it was first designed.
Posted Sep 30, 2010 19:42 UTC (Thu) by Jan_Zerebecki (guest, #70319)
Usually the range is limited to the distance the signal can travel in one timeslot, but it seems you can do something about it if your BTS is modified to allow the signal to arrive in the next timeslot.
Posted Sep 30, 2010 19:50 UTC (Thu) by Cyberax (✭ supporter ✭, #52523)
Posted Oct 4, 2010 2:24 UTC (Mon) by showell (subscriber, #2929)
GSM will be with us for along time because the GSM chipsets are still 1/2 the price of the 3G / 4G ones (due mainly to IP payments). GSM handsets sell in poorer markets in massive quantities over 3G.
LTE and VoIP
Posted Sep 30, 2010 6:57 UTC (Thu) by Cato (subscriber, #7643)
Basically, 3G voice fallback has issues with the time to switch networks, while VoLTE is the long term solution (but has had to add SMS support to IMS), while VoLGA is the interim solution using GAN (Generic Access Network, aka UMA) - essentially circuit switch calls over an IP infrastructure, also used today by mobile phones that switch between GSM/3G and WiFi while on the same call (e.g. Blackberry and some Nokias), bypassing the need for femtocells in some cases.
Ultimately it looks like everyone will use VoIP with LTE, but it will take a few years given that some of the VoLTE standards aren't yet finalised.
Posted Sep 27, 2010 8:40 UTC (Mon) by rahvin (subscriber, #16953)
The kicker is that even when the patents expire, or in countries where they aren't valid the closed ecosystem prevents operators from offering service at locally affordable rates without paying the IP tax. The ultimate goal (at least from my perspective) of all these FOSS GSM projects is to break open the closed ecosystem and crack the security that enforces it so that software isn't an expense in the equation. This relegates the costs to construction, towers, power and general purpose equipment with FOSS on top. Without the "software" costs the systems can be erected cheaper, maybe even to the point where people making less than $2000 a year can afford service. At the least it's definitely going to lower the break point where more people can afford cellular service. Even if it takes 5 years or more to crack open, GSM is going to have a long lifespan in areas with low average incomes.
Most people in the west don't realize it, but cellular phones are the primary means of access to information in large areas of the world. Not just voice, but the primary internet access as well. These efforts to crack this closed ecosystem and the security that enforces it are vital to free access to information.
Harold's role in this is critical as the security is a key component to building a cellular service. Not only that but the benefits to those of us in the west as he could find security holes that right now only the government and criminals are aware of.
Posted Sep 27, 2010 8:50 UTC (Mon) by Cyberax (✭ supporter ✭, #52523)
We have like 80% of coverage for the whole country, with many competing independent cell networks. I can literally travel everywhere and get a good phone signal (even in Chernobyl).
Posted Sep 29, 2010 8:31 UTC (Wed) by mfedyk (guest, #55303)
Posted Oct 7, 2010 7:46 UTC (Thu) by gat3way (guest, #47864)
3G did develop practically in the last 3-4 years and it was incredibly expensive in the beginning, now you can get postpaid plans with unlimited traffic for about $15-$20/month. Lots of people switched off from ADSL to the mobile telcos in the rural areas as it costs almost the same (and the national ADSL provider is notorious for its poor support).
Yet as the nice new Android smartphones and the iphones are quite expensive here, GSM is here to stay for at least a couple of years.
Posted Sep 27, 2010 11:28 UTC (Mon) by marcH (subscriber, #57642)
They can easily transfer credit from one account to another, sometimes using phone credit as mere pocket money.
Price of GSM/3G base stations
Posted Sep 30, 2010 6:44 UTC (Thu) by Cato (subscriber, #7643)
The test would be whether WiMAX base stations are cheaper, as they are built in a more "commodity" way - if GSM becomes more open it would be possible to build GSM base stations like that.
UMTS femtocells are very cheap (there are few if any GSM femtocells) with a consumer price of $70-$150 - I bought mine for $70 new off eBay.
Posted Sep 30, 2010 12:26 UTC (Thu) by Cyberax (✭ supporter ✭, #52523)
Posted Sep 27, 2010 8:48 UTC (Mon) by nhippi (subscriber, #34640)
Posted Sep 27, 2010 11:31 UTC (Mon) by marcH (subscriber, #57642)
Posted Sep 27, 2010 11:59 UTC (Mon) by Cyberax (✭ supporter ✭, #52523)
As for pausing bittorrents - that's what QoS is for.
Posted Sep 27, 2010 15:50 UTC (Mon) by i3839 (guest, #31386)
I hope the security part is simple and separate from the complex rest.
Posted Sep 27, 2010 22:10 UTC (Mon) by marcH (subscriber, #57642)
Yes designing a brand new network is a nice opportunity but... seeing is believing.
Posted Sep 28, 2010 11:40 UTC (Tue) by Cyberax (✭ supporter ✭, #52523)
On the wide Internet it's probably not going to make much difference, anyway. Voice traffic is fairly narrow-band, even high-quality codecs do not require more than 64Kbit.
Posted Sep 28, 2010 15:44 UTC (Tue) by marcH (subscriber, #57642)
Narrow band does not grant immunity from occasional packet loss and high latencies. Only QoS helps in such cases.
Posted Sep 30, 2010 7:18 UTC (Thu) by Cato (subscriber, #7643)
Posted Sep 30, 2010 13:03 UTC (Thu) by Cyberax (✭ supporter ✭, #52523)
Quite often voice and data backbones also use separate circuits, so QoS is not used even there.
Posted Oct 6, 2010 9:39 UTC (Wed) by marcH (subscriber, #57642)
This is actually some simple, manual, very coarse form of QoS when you think about it. Simple => works.
Posted Sep 29, 2010 8:43 UTC (Wed) by mfedyk (guest, #55303)
Posted Sep 29, 2010 9:04 UTC (Wed) by Cyberax (✭ supporter ✭, #52523)
Posted Sep 29, 2010 19:25 UTC (Wed) by nhippi (subscriber, #34640)
LTE is going to use H.323 family of standards for audio/video data
Posted Sep 30, 2010 19:10 UTC (Thu) by Cyberax (✭ supporter ✭, #52523)
H.323 is not going to be used on end-user devices. Which is a saner decision then what I expected from committees. It was still on the table last time I inquired about it, so I'm sorry for misinformation.
Posted Sep 27, 2010 13:15 UTC (Mon) by tzafrir (subscriber, #11501)
Posted Oct 2, 2010 16:36 UTC (Sat) by job (guest, #670)
There's a lot of guesswork going on in this discussion thread.
Posted Oct 7, 2010 9:08 UTC (Thu) by marcH (subscriber, #57642)
Copyright © 2013, Eklektix, Inc.
Comments and public postings are copyrighted by their creators.
Linux is a registered trademark of Linus Torvalds