[NCLUG] Data Through Cellular Technology?

Michael Dwyer mdwyer at sixthdimension.com
Mon Aug 12 10:35:58 MDT 2002


John L. Bass wrote:
> 	AT&T's data offerings don't seem that good to me.  I think the three you 
> 	need to look at are Verizon, SprintPCS, and Voicestream->Tmobile.
> 
> 	AT&T uses a CDPD analog system, last I knew.  I think they are 
> 	transitioning to a GSM interface, though.  I think this is the system 
> 	that Sean et al at Tummy have been using.  It is kinda slow, but it 
> 	works great for them.
> 
> 	Verizon and Sprint both use CDMA systems.  They are currently limited to 
> 	14.4k.  Both are supported by Sierra Wireless pcmcia cards, though only 
> 	the older models support Linux.  The rest are akin to winmodems. :( 
> 
> Technically, all wireless communications are analog at some point
> when modulated on to the RF ... but CDPD service is really IP packet
> oriented digital service at it's descriptive hardware interface.
> 
> The CDPD technology that both AT&T and Verizon support today is
> the same - and the CDPD modem cards are usable interchangably on
> both services. The fact that AT&T uses TDMA for digital voice,
> and Verizon/Sprint use CDMA is a choice of how much range and
> buildout you need to service a coverage area. There are small
> advantages to the CDMA approach in high density metropolitian
> areas, and huge advantages to TDMA in rural and suburban areas.
> As a result, AT&T offers a much broader coverage area outside metro
> areas.

Verizon has an analog AMPS network that (presumably) supports CDPD, but 
their newer systems are CDMA. Verizon is a bundle of a whole bunch of 
smaller systems, though, so the network is often different depending on 
where in the US you are.

  > Even after G3 services are deployed, CDPD will be around for a
> long time ... simply because the lower bandwidth gives the technology
> 2-4x the range/building penitration.  As you double the data rate,

...same reason I expect AMPS to be around for a long time. An AMPS tower 
covers something like 10x the range of a similar digital tower.

> Another advantage is that CDPD is implemented in the 800mhz band,
> while G3 is implemented in the 1,900mhz bands (and higher). Similar
> to the power per bit problem we have a problem that as you double the
> carrier frequency you lose half (3db) of your usable signal to noise
> resulting in reduced range and building penitration,

Actually, Verizon's 3G CDMA network is at 800MHz.  So it has an 
advantage over the 1900MHz SprintPCS CDMA system.  It is a common 
mistake to think that PCS == Digital.  Actually, PCS == 1900MHz.  The 
800MHz band is often called 'Cellular', even though there are PCS-type 
services on it.

> Additionally, 800mhz CPDP has much less problems with phase noise
> caused by multipath - the higher data rate 1.9Ghz systems are much
> more sensitive to multipath (as are 802.11 systems).

CDMA systems /depend/ on multipath.  Their rake receivers listen to 
multiple tower and multiple paths at the same time, making the signal 
/more/ reliable instead of less.

http://denbeste.nu/cdmafaq/index.shtml

Oh, and just so Matt doesn't tear me a new one:
ObLinux: I just got back from the Leadville Trail 100 Mountain Bike 
Race, where unknown to most people, Linux was running the whole show.  A 
linux box connected to a network of AX.25 radio modems takes data from 
remote aid stations and puts it all into a MySQL database.  The results 
are then displayed with a PHP web application.  You can view the results 
at http://www.trail100.com/

ObForPeopleWhoGiveATossAboutISPs:  One of the Trail 100 riders works for 
AT&T Broadband.  I explained my concerns with their AUP with respect to 
Linux servers -- like the one Trail100.com is currently sitting on.  He 
said he would 'see what he could do'.  I don't expect much, but I 
certainly felt better about putting a message in somewhere that might 
get heard as opposed to all my other complaints...




More information about the NCLUG mailing list