What data speeds are you getting?

Kevin OQuinn

AC Team Emeritus
May 17, 2010
9,267
496
0
Visit site
Inconsistent speeds in Chicago near O'Hare Airport.

I ran these back to back about a mile away from the airport. (notice the time in the top right corner)

snap20110101_175102-1.png


snap20110101_175119.png


snap20110101_175138-1.png
 

MattMJB0188

Well-known member
Dec 8, 2010
956
101
0
Visit site
Ok I take back all of NWI not having 3G coverage. I'm in Highland, In getting download speeds of 5mb. I'd post a screenshot of speedtest but I don't know how lol.
 
Last edited:

Wookie Claws

Member
Feb 13, 2010
19
0
0
Visit site
I am in the Portland, OR area: I get about 3.4m to 6m down, typical is about 5 or 6 though. Up is pretty consistent at .65m. The only time I got 1.5m up was when I was in Seattle for a weekend (and I got an 8m down once there too).

Practical use though... things load and download pretty darn fast so I'm not complaining. I switched to a new SIM about a month or so into my phone to see if that speeds things up.. it didn't.
 

Kevin OQuinn

AC Team Emeritus
May 17, 2010
9,267
496
0
Visit site
Ok I take back all of NWI not having 3G coverage. I'm in Highland, In getting download speeds of 5mb. I'd post a screenshot of speedtest but I don't know how lol.

The easiest way is to use shootme to take the screen shot and photobucket to upload them to the website then login here to link them.

I tried many different ways and this is the easiest. Photobucket is free and the app works great.
 

MyTouch 420G

Member
Nov 30, 2010
24
1
0
Visit site
Beaverton, Oregon (Next to 1/9 Intel Campuses) at 11:00am:

snap20110104_105844.png




I usually get upwards of 2mb/sec upload, but midday is a little choppy (damn you tech sector!!!)
 

Kevin OQuinn

AC Team Emeritus
May 17, 2010
9,267
496
0
Visit site
I think that's what we're establishing here. ;) I think 1-1.5 is good for 3G. HSPA+ I'm guessing 3-5 would be average.

Sent from my myTouch 4G
 

jbjtkbw007

Well-known member
Oct 26, 2009
347
49
28
Visit site
I think that's what we're establishing here. ;) I think 1-1.5 is good for 3G. HSPA+ I'm guessing 3-5 would be average.

Sent from my myTouch 4G
I'd agree with this based upon CURRENT speeds. That won't be the case with their 42 mb/s+ announced this week at CES. :)
 

anon62607

Well-known member
Sep 23, 2010
436
27
0
Visit site
From my experience:

EDGE <1 Mbps

Average 3G I would say 1-1.5 Mbps.

Average fast 3G seems to be ~3 Mbps.

Average HSPA+ varies on your location, but should be >3 Mbps.

EDGE should be no faster than 236 kbps unless it's evolved edge, which on tmobile it's not.
 

jbjtkbw007

Well-known member
Oct 26, 2009
347
49
28
Visit site
For current devices, the upgrade to 42 mbps won't help.

That's not entirely true. Some of the regular 3G phones will receive a bit of a boost, but they won't reach the speeds of the 4G phones (that part goes with your thinking). Some users with just 3G phones (e.g. Nexus S) have reported speeds between 5-7 meg. That could indeed get bumped just a bit.
 

anon62607

Well-known member
Sep 23, 2010
436
27
0
Visit site
That's not entirely true. Some of the regular 3G phones will receive a bit of a boost, but they won't reach the speeds of the 4G phones (that part goes with your thinking). Some users with just 3G phones (e.g. Nexus S) have reported speeds between 5-7 meg. That could indeed get bumped just a bit.

Why would it be increased? From what I understand about the upgrade to 42 mbit from 21, it's the addition of another carrier with another 5 Mhz spectrum, but all other modes remaining the same. That is to say, the number of spreading codes used would be the same (and remember, that Nexus S can only despread 10 WCDMA spreading codes and no change to the tower would change that) and the code rate would remain the same. So if the code rate is the same, symbol rate is the same, and the number of spreading codes used to transmit to the user equipment do not change, the modulation remains the same (which is limited on the Nexus S to 16QAM) and obviously the user equipment doesn't get an extra receiver to handle another carrier - how would the bandwidth to the phone increase?
 

Kevin OQuinn

AC Team Emeritus
May 17, 2010
9,267
496
0
Visit site
Why would it be increased? From what I understand about the upgrade to 42 mbit from 21, it's the addition of another carrier with another 5 Mhz spectrum, but all other modes remaining the same. That is to say, the number of spreading codes used would be the same (and remember, that Nexus S can only despread 10 WCDMA spreading codes and no change to the tower would change that) and the code rate would remain the same. So if the code rate is the same, symbol rate is the same, and the number of spreading codes used to transmit to the user equipment do not change, the modulation remains the same (which is limited on the Nexus S to 16QAM) and obviously the user equipment doesn't get an extra receiver to handle another carrier - how would the bandwidth to the phone increase?

Because with increased speed comes increased bandwidth. Increase the bandwidth and you increase the speed for everyone. Ever wonder why at night your home internet connection is faster even though the rated speed stays the same? It's because less people are using it. So, more bandwidth = more speed. Same is true for mobile connections.
 

anon62607

Well-known member
Sep 23, 2010
436
27
0
Visit site
Because with increased speed comes increased bandwidth. Increase the bandwidth and you increase the speed for everyone. Ever wonder why at night your home internet connection is faster even though the rated speed stays the same? It's because less people are using it. So, more bandwidth = more speed. Same is true for mobile connections.

Ok, I don't know much about this but I want to understand the mechanic behind how it's actually working.

In HSPA, a subframe consists of 2560 chips per timeslot, which is 2/3rds of a millisecond. With a TTI of 2 ms that gives us 3 subframes per TTI (2/ (2/3)) for 7680 chips per 2 milliseconds or 3.84 million chips per second. Presuming the very best error correction rate (ie code rate = 1) - that is to say radio conditions are so good that we're not using error correction at all - something that will never happen in reality lets look at what we have.

Every 2 milliseconds the tower decides how to divide up the bandwidth it has available and it does this via a mechanism called CDMA. You can take any channel and code it with a rate higher than the symbol rate to divide that channel into n number of equal bandwidth, where n is the spreading factor. That is to say, with a spreading factor of 1, all of the bandwidth in the channel goes to a single code. HSPA always uses SF=16 but a given receiver can use a varying number of codes to despread a given transmission.

Basically, that 2 millisecond block that the tower (and handset) are working with is divided up into 16 parts in the code domain. Those 16 parts (codes) are divided up among handsets that are going to be receiving information in that 2 ms slice, if I have 3 handsets that I want to divide bandwidth equally to, I would transmit for 5 codes to handset 1, 5 to handset 2, and 5 to handset 3.

Ok, so given that we have a 2 millisecond time block to work with and there are 7680 chips within that block, and that can be broken up 16 ways via spreading codes, how many bits are being transmitted in that 2 milliseconds within a single code?

Using 16QAM you get 4 bits per chip - there are four possible positions in the constellation (the combination of phase and amplitude) that the signal can be in. So, 7680 chips * 4 = 30720 bits per 2 milliseconds and that is broken up into 16 equal parts for 1920 bits per code per 2 milliseconds. Since there are 500 of those 2 millisecond blocks in a second that's 960,000 bits per second per code.

Early on, a handset could receive (despread) 5 codes of the 16 so that made the maximum possible bandwidth 4.8 megabits / second, then later on 10 codes could be despread for a maximum bandwidth of 9.6 megabits per second and then finally 15 codes could be despread (on the G2 and mytouch 4G) for 14.4 megabits per second.

Note that in reality there's another factor that's being left out, and that's the code rate - how much of a given block of bandwidth goes to error prevention - the lower the code rate (more bits used for error correction) the more immune the transmission is to noise and other problems.

So HSDPA defines several different categories for the user equipment. One of these categories can despread 5 codes and uses a 3/4 code rate which gives the very highest bandwidth that the handset can actually receive (not counting error correction) as 4.8 mbps * 3/4 = 3.6 megabits per second (sound familiar)?

A higher up category can despread 10 codes and uses the same 3/4 code rate and gives 9.6 mbps * 3/4 = 7.2 megabits per second (still sounding familiar)?

And finally we get to where the mt4g and g2 are, despreading 15 codes but with a much higher code rate (35/36 I believe) for 14.4 mbps * 35/36 = 14.0 megabits per second.

A given handset's chipset is pretty fixed in it's capabilities - it can only despread so many codes for example. So for example if you have a 3.6 mbps 3G HSDPA network what that's really saying is that the tower will never try to transmit to any handset using more than 5 spreading codes. If the tower's software is upgraded to allow 10 per handset that doesn't matter for a given handset that is already chipset limited to despreading 5 codes.

So going along with all that, my question is given that a given handset can only despread a certain number of codes at a certain maximum chip rate, how is increasing the tower capability going to increase the performance of that same handset? I'm looking for what part of the handset (say a Nexus S, able to despread 10 codes at 7680 chips / 2 milliseconds with a 3/4 code rate) is being changed to allow for increased download performance?
 

Trending Posts

Forum statistics

Threads
942,921
Messages
6,916,503
Members
3,158,737
Latest member
Jan