Cellular Data Communications - Speeds

[AlGriefer]

We've had a lot of talk about cellular data communications and which type and carrier is best.? I'd like to build a table of the various carrier and speeds in different locations with differing options.? For instance T-Mobile currently support GPRS and EDGE and will soon support UMTS.? Speeds will vary in different locations.?

Append the info to this thread and I'll add it to the table below.? For the sake of comparing apples to apples, please use the speedtest at DSL Reports and use the site, Speakeasy for the test.

Carrier? ? ?	Service? ?	Location?? ? ? ?	Card? ? ?	Upload? ? ?	Download	Signal Strength	Date	Time? ? ?
T-Mobile	GPRS	Harrisburg, OR	Sony Ericsson GC-89	20 Kbps	41 Kbps	4 Bars	10-02-06	2100 PDT
T-Mobile	GPRS	Oyster Pt, CA	Sony Ericsson GC-89	53 Kbps	81 Kbps	-73 dBm	10-02-06	2255 PDT
Verizon	1X	Grand Junction, CO	Verizon Aircard 5740	1286 Kbps*	127 Kbps	NA	10-03-06	NA
T-Mobile	GPRS	Alameda, CA	Sony Ericsson GC-89	65 Kbps	108 Kbps	-71 dBm	10-04-06	1211 PDT
T-Mobile	EDGE	Lodi, CA	Sony Ericsson GC-89	80 Kbps	135 Kbps	4 bars	10-15-06	0730 PDT

* = Appears to be incorrect

Al

 

[Tom]

Al,

Carrier: t-mobile
Service: GPRS
Location: Oyster Point, CA
Upload speed : 53 kbps
Download speed : 81 kbps
Date: 10-2-2006
Time: 10.55pm
Signal strength: 74% (-73dBm)

I added time because I do see variations in performance at different times of the day and evening. Don't know if it's worth including signal strength; Some folks' devices may not provide that info. Your choice of course.

I ran only one test; Don't know if I should run several and report the average  ???

 

[Clay L]

Carrier is Verizon, Service is 1X,? Grand Junction, CO -? Download Speed is 127 kbps, Upload speed is 1286 kbps. Signal  strength is 5 bars.

I ran it three times and averaged the results.

Using a Verizon Aircard (5740 I think).

 

[Gary RV_Wizard]

Sorry, but I don't think this is a useful endeavor because data speeds on a cellular connection vary with the number of users concurrently online and what they are doing at the moment.  That's because the total bandwidth is a shared by all concurlent cellular data users on that tower at that time. You might have 10-15 kbps at one moment and 140 kbps a few minutes later.  About the only thing that is meaningful is the type of service available in each location, e.g. GPRS/EDGE/UMTS for GSM carriers and QNC/1Xrtt/EVDO for CDMA carriers. That sets the upper bounds of the potential performance

 

[AlGriefer]

Al,

Carrier: t-mobile
Service: GPRS
Location: Oyster Point, CA
Upload speed : 53 kbps
Download speed : 81 kbps
Date: 10-2-2006
Time: 10.55pm
Signal strength: 74% (-73dBm)

I added time because I do see variations in performance at different times of the day and evening. Don't know if it's worth including signal strength; Some folks' devices may not provide that info. Your choice of course.

I ran only one test; Don't know if I should run several and report the average? ???

Would be an interestring data point.  I'm also going to add the card type.  This may be an important data point.

Al 

 

[Tom]

I'm also going to add the card type.

My card is the same as yours Al - Sony Ericsson GC89. Some folks may be using a cell phone tethered to a PC rather than a card.

 

[AlGriefer]

Sorry, but I don't think this is a useful endeavor because data speeds on a cellular connection vary with the number of users concurrently online and what they are doing at the moment.? That's because the total bandwidth is a shared by all concurlent cellular data users on that tower at that time. You might have 10-15 kbps at one moment and 140 kbps a few minutes later.? ?About the only thing that is meaningful is the type of service available in each location, e.g. GPRS/EDGE/UMTS for GSM carriers and QNC/1Xrtt/EVDO for CDMA carriers. That sets the upper bounds of the potential performance

Gary,

I think it gives you useful data.? One bit of information is the realistic speed you can expect in a certain location at that time of day.? Sure, if the cell site is overloaded, you won't be able to get maximum throughput, but if you get reasonably consistent readings, you know what to expect.

For instance, I've found areas where there were consistently low data rates regardless of time of day.? After checking with the carrier, I was told that they limited the number of channels that a user could concurrently use in that area.? He didn't know whether it was becaus of equipment limitations or what, but he did know that there were no other data users on that cell site at the time.? He stated that, on GPRS a user could be assigned one to four channels for data transfer and that was settable by site, among other things.

So, while these speeds are no more "accurate" than DSL or cable speeds, you still get a feeling for what is reasonable and how good the general service appears.

Al

BTW: I just reran the test here at Harrisburg and got the same speeds +/- 5%

 

[AlGriefer]

My card is the same as yours Al - Sony Ericsson GC89. Some folks may be using a cell phone tethered to a PC rather than a card.

Hmm... Does your software show whether your using EDGE or GPRS?? There's two different programs you can use to manage the card; the one that T-mobile sends you and one that's available on the Sony ericsson Web site.? The Sony program shows whether you're attached on GPRS or EDGE.? It's available here.

 

[Tom]

Al,

I previously downloaded the Ericsson software, but haven't been able to get it to work. So I reverted back to t-mobile's software. It appears to tell you which service you're on - click Tools|Network Info|Network, although the few times I've checked, it's always said GPRS.

 

[Tom]

Al,

Carrier: t-mobile
Service: GPRS
Location: Alameda, CA (Oakland Estuary)
Upload speed : 65 kbps
Download speed : 108 kbps
Date: 10-4-2006
Time: 12.11pm PST
Signal strength: 78% (-71dBm)

 

[Clay L]

I noticed that my download speed is much slower than my upload speed. You reversed what I reported because everyone else has the reverse.
I went back to speakeasy.net and re-ran the test six times using two different servers and got an average of: 130 kbps download and 1164 kbps upload. 5:30pm, 4 Oct. 5 bars

That doesn't make sense to me because the Verizon Venturi accelerator software/server  compresses the download data and the Venturi software in my computer de-compresses it. The download should be faster but it isn't.

Even if the download compression isn't very effective because of the file type speakeasy uses for the test, it should still be faster than my upload speed.
That's a puzzlement...

 

[AlGriefer]

I noticed that my download speed is much slower than my upload speed. You reversed what I reported because everyone else has the reverse.
I went back to speakeasy.net and re-ran the test six times using two different servers and got an average of: 130 kbps download and 1164 kbps upload. 5:30pm, 4 Oct. 5 bars

That doesn't make sense to me because the Verizon Venturi accelerator software/server? compresses the download data and the Venturi software in my computer de-compresses it. The download should be faster but it isn't.

Even if the download compression isn't very effective because of the file type speakeasy uses for the test, it should still be faster than my upload speed.
That's a puzzlement...

I would expect that the upload speed is incorrect.? I've seen that sort of thing in the past, especially when compression software is used.? You might try this site and see if it reports significantly different results.

Al

 

[Clay L]

I would expect that the upload speed is incorrect.? I've seen that sort of thing in the past, especially when compression software is used.? You might try this site and see if it reports significantly different results.

Al

Okay, did that.
Got a more reasonable upload speed - 114 kbps, but now the download speed - 18,699 kbps really (wrongly) reflects the Venturi acceleration.
Probably an Upload speed = 114 kbps and a download speed = 130 kbps is close to the real speeds.

Right now I am connecting to the slower default - National Access - data net. The faster net - EVDO? -or Broadband Access is a lot faster. It's available in a lot of places but not out here in the sticks.

EDIT:
Just for grins I ckecked the Verizon site and found the following speed info:
Download only, no compression;
National Access, 60 to 80 kbps with bursts up to 144 kbps.
Broadband Access, 400 to 700 kbps with bursts up to 2000 kbps.

Sure be glad when they get Broadband deployed all over!!

 

[Tom]

Carrier: t-mobile
Service: GPRS
Location: Alameda, CA (Oakland Estuary)
Upload speed : 65 kbps
Download speed : 83 kbps
Date: 10-5-2006
Time: 8.21am PST
Signal strength: 58% (-81dBm)*

* Note: The signal strength is varying quite a bit.

 

[caltex]

Al, I tried the site you recommended and got an download speed of 178k. My dial up modem says I am connected at 26k.  What an I doing wrong?

 

[AlGriefer]

Al, I tried the site you recommended and got an download speed of 178k. My dial up modem says I am connected at 26k.? What an I doing wrong?

My guess is that ypu're using compression software, either in your computer or in the modem.  What modem are you using?  Are you sure the speed measurement wasn't done by your fishing scale?  ;D

Al

 

[caltex]

Al, I am sitting out in the country in Oklahoma.  Using earthlink dial up with the modem built into my Apple laptop.
I think the telco lines out here are not too good and the modem won't support 56k.  I wish I was fishing.

 

[Gary RV_Wizard]

You guys are going to drive yourselves batty trying to make sense out of a thing that has so many variables it is all but imponderable!  :

First there is the raw data speed of the hardware at the ends of the link and the raw speed of the link itself.  The hardware has a fixed max and so does the link, but the link is typically degraded below its max for  a variety of reasons and may (in the case of wireless) be shared as well. The raw speed is the rate at which bits move, but not all bits are data bits - a lot of them are for managing the flow of bits across the link. Therefore a 56 kbps link doesn't actually move 56 kilobits of pure data bits. However...

Then there is compression.  There are several levels of compression and just about every transmission is compressed to some degree or another. High level functions like Venturi is just one method and there is still data compression in effect  on the physical links when you turn Venturi off. Compression can make the amount of daa which actually reaches your compter actualy exceed the raw data speed of the link and your hardware. This typically occurs when the data is highly compressable (e.g. text or bit mapped graphics).

Those speed measuring sites generally try to evaluate net data flow over a link by transferring a known amount of "end use" data and measuring the net time to completion.  For example, transfer a file that Windows shows as 10,000 bytes long and see how long it takes to arrive.  Different data content may get compressed more - or less - depending on the compression algorithm.    No two measurement sites uses the same data or the same measurement procedure, so you cannot compare numbers from different sites very well. And the net time also depends on the path in use at the time, i.e. where & how far the data has to travel to reach you. That is not a constant factor on the net, even when you make a connection several times on the same day, because the path across the internet backbone is somewhat dynamic.  Internet loading is also a factor - a minor hiccup in the data routing - a delay of a half a second on a single data packet - can distort the net data transfer measurement badly.  You may get a singularly speedy transmission for a few seconds - or a singularly slow one.

 

[Tom]

Al,

Here's my latest results. Please ignore or use at your peril:

Carrier: t-mobile
Service: GPRS
Location: San Rafael, CA
Upload speed : 65 kbps
Download speed : 115 kbps
Date: 10-6-2006
Time: 4.32pm PST
Signal strength: 82% (-69dBm)

 

[AlGriefer]

Tom,

You're getting more or less the same speeds at the locations you've been to.  Of course, they're all in the SF metro area so I won't put all of them on the list, but I'll be interested in what you get in other locations, especially the Delta.

Robert,

As Gary said, the speeds you get will vary.  The modem connected speed (26k) is what your modem reported to the operatring system when it connected.  The Download speed is what you got from the test site, end to end, while going through any compression software on your system.  I know Earthlink has compression software as does many modems. 

While speeds will vary do to all of the factors Gary mentioned, we can get at least an order of magnitude feel for the different services and locations.