Welcome to the MacNN Forums.

If this is your first visit, be sure to check out the FAQ by clicking the link above. You may have to register before you can post: click the register link above to proceed. To start viewing messages, select the forum that you want to visit from the selection below.

You are here: MacNN Forums > Hardware - Troubleshooting and Discussion > iPhone, iPad & iPod > iPhone 3G (v 2.1) WiFi is 20% Slower on 802.11G than 802.11B

iPhone 3G (v 2.1) WiFi is 20% Slower on 802.11G than 802.11B
Thread Tools
Guy Kuo
Dedicated MacNNer
Join Date: Aug 2001
Status: Offline
Reply With Quote
Oct 25, 2008, 03:01 AM
 
My measurements are showing that the iPhone 3G (v2.1) WiFi implementation is slower when connected to an 802.11G network than an 802.11B network. The speeds should be similar or FASTER via a G network. This 802.11B slow down is NOT seen on the iPod Touch 2nd Gen (v2.1.1)

Repeated speed trials yield good evidence something is slowing the iPhone 3G when it connects to a 802.11G network.

Using...

iPhone 3G (version 2.1)
iPod Touch 2nd Gen (version 2.1.1)
Airport Express (version 6.3) in either B only or G only wifi mode

http://i.dslr.net/tinyspeedtest.html as speed measurement site.

Results with iPhone 3G connected via 802.11B
2668
1908
3026
2650
2874
3107
3076
2266
2668
3248
3033
3105
2447
2387
yielding an average of 2747 kbps

iPhone 3G connected via 802.11G
2537
1356
1533
1821
1559
1956
1659
1426
1166
2144
2699
1890
2445
2208
yielding an average 1885 kbps which is 20% slower than 2747 kbps via 802.11B

One might be tempted to conjecture that this is simply internet variability, but the difference between 2747 & 1885 is significant even if one throws out he highest and lowest measurements of each set. Also, a fair number of trials were performed reducing the probability that this is mere chance.

Further, measurements using the iPod Touch 2nd showed nearly IDENTICAL average speeds for both the 802.11B and 802.11G connection

iPod Touch 802.11B results
2564
2660
2694
3353
2392
2171
1769
3473
3559
2820
average = 2745 using 802.11B on iPod Touch

iPod Touch 802.11G results
2623
1549
2963
2801
2674
1759
1983
5378
1700
5172
average 2860 via 802.11G on iPod Touch

The iPod Touch runs had averages that were similar 2745 & 2860 between B and G usage. Notice that this figure also matches the iPhone 3G's 802.11B results (2747 kbps)

My conclusion is that the current software is slowing iPhone 3G WiFi performance when it connects to a 802.11G base station. Apple should be able to fix this problem as it does not happen with the iPod Touch 2nd Gen. For now, my solution at home is to keep a separate wifi 802.11b only network for my iPhone. That's not an onerous a solution since I need a separate network anyway because my main devices run on a wide channel 802.11n network with which the iPhone can't connect.

Guy Kuo
     
Tomchu
Mac Elite
Join Date: Sep 2005
Status: Offline
Reply With Quote
Oct 25, 2008, 03:41 AM
 
Try a different channel setting on the WiFi access point -- one of 1, 6, or 11, depending on whatever it's set to now. Then retest the iPhone 3G in 802.11g mode. :-P
     
moep
Senior User
Join Date: Nov 2003
Status: Offline
Reply With Quote
Oct 25, 2008, 03:49 AM
 
Not for me. I just pushed a 2GB file to my iPhone via SSH in G-only and B-only.
The router is a WRT54GL with Tomato Firmware, encryption is WPA2 Personal AES.
MBP was connected to the router via ethernet, of course.

802.11G: 974 KB/s average (7792 kbps)
802.11B: 589 KB/s average (4712 kbps)

Although something tells me that WLAN bandwidth might not the limiting factor for 802.11G, there’s still a noticeable difference between the two. My point is that you just can’t measure bandwidth of your LAN devices by pulling a bunch of 2MB files from a web-server through your WAN.
"The road to success is dotted with the most tempting parking spaces."
     
Tomchu
Mac Elite
Join Date: Sep 2005
Status: Offline
Reply With Quote
Oct 25, 2008, 04:02 AM
 
Originally Posted by moep View Post
My point is that you just can’t measure bandwidth of your LAN devices by pulling a bunch of 2MB files from a web-server through your WAN.
Oh ... he DID do that, didn't he. That's kind of dumb.

Never test wireless performance by doing something over the Internet. :-|
     
Guy Kuo  (op)
Dedicated MacNNer
Join Date: Aug 2001
Status: Offline
Reply With Quote
Oct 25, 2008, 04:54 AM
 
But.... you CAN measure the response of the system as a whole by that methodology. There is a difference in how well the iPhone 3G handles actual web accesses when connected by 802.11b or 802.11g. It may not be a raw data rate issue, but a problem with the 3G dealing with longer latency servers. The longer latencies are not duplicated in LAN only testing.

Because most of the websites I'm interested in visiting are not hosted on my LAN (and I suspect nor are the sites visited by most users), performance metrics including the WAN are useful. Of course, metrics over a WAN connection are be affected by internet traffic. So, multiple test runs are needed to reduce those confounding effects. That's why I bothered to do multiple measurements.

Also, this issue seems to be limited to only the 3G version of the iPhone. Others have noticed this difference between the 3G iPhone and the original iPhone. That's why I decided to measure my own phone.
     
Tomchu
Mac Elite
Join Date: Sep 2005
Status: Offline
Reply With Quote
Oct 25, 2008, 02:30 PM
 
I do networking.

Typical WiFi latencies are 1-2 ms. Typical cable/DSL latencies to servers within your ISP's immediate network are 20-30 ms. That's over one order of magnitude difference. Any slight variation you see in WiFi latencies is going to be completely overridden by the variance in latencies going out to the WAN -- to the point where your margin of error is large enough that you can't draw *any* conclusions about the wireless performance.

I'm not even going to talk about throughput and the fallacy in testing WiFi through via a typical cable/DSL link to the outside world. Unless your Internet is capable of pushing >4 MB/s, you're not going to get any real WiFi throughput results.

Once something leaves your LAN and goes out to the WAN, no host on the other end is going to care whether you used 802.11b or 802.11g for the initial hop. Not only that, but it's not going to know.

Whatever conclusions you drew about WiFi by testing your Internet's download speed are fallacious. Do a real (aka. local) test, and then get back to us. If the local testing shows no differences between 802.11b and g on the iPhone, or differences that are in line with what you'd expect, then that's because there's nothing wrong.

If local testing shows that g is in fact slower than b, then you're on to something.

In the scientific process, you change one variable while keeping all the others constant. That's impossible to do if you're throwing the Internet and bandwidth tests into the mix.
     
Guy Kuo  (op)
Dedicated MacNNer
Join Date: Aug 2001
Status: Offline
Reply With Quote
Oct 25, 2008, 05:02 PM
 
By your reasoning, it would be impossible to ever measure speed over the internet. Any measured difference in accessing the internet between two devices should be ignored even if there is a measurable difference across multiple trials with only a single control variable altered? No. That isn't entirely true. When one attempts to measure something that has noise in the data, one takes multiple readings to allow the noise to cancel itself out.

I understand your wanting to remove the WAN because it is difficult to hold that constant, but the WAN IS where the servers live. I do measure a difference on the --->3G <-- iPhone when it connects to an Airport express via b or g protocol and talks to web servers. I change ONLY the airport B/G status between trial runs and I minimize the WAN variability by doing multiple trials to reduce that noise. Also, an iPod Touch 2nd Gen undergoing the same type of testing does NOT have the same slow down effect. In fact its results match that the iPhone 3G in 802.11b mode and corroborates my conclusion. Were the WAN effects to swamp out the difference being measured, the touch's results would not be so equal between b and g modes.

Read this very carefully ----> I agree that G may well be faster than B for LAN performance of the iPhone. However, I also measure a repeatable difference when the device communicates across the internet. That isn't opinion. It's simply what I measure.
     
turtle777
Clinically Insane
Join Date: Jun 2001
Location: planning a comeback !
Status: Offline
Reply With Quote
Oct 25, 2008, 05:34 PM
 
Originally Posted by Tomchu View Post
I do networking.

Typical WiFi latencies are 1-2 ms. Typical cable/DSL latencies to servers within your ISP's immediate network are 20-30 ms. That's over one order of magnitude difference. Any slight variation you see in WiFi latencies is going to be completely overridden by the variance in latencies going out to the WAN -- to the point where your margin of error is large enough that you can't draw *any* conclusions about the wireless performance.

I'm not even going to talk about throughput and the fallacy in testing WiFi through via a typical cable/DSL link to the outside world. Unless your Internet is capable of pushing >4 MB/s, you're not going to get any real WiFi throughput results.

Once something leaves your LAN and goes out to the WAN, no host on the other end is going to care whether you used 802.11b or 802.11g for the initial hop. Not only that, but it's not going to know.

Whatever conclusions you drew about WiFi by testing your Internet's download speed are fallacious. Do a real (aka. local) test, and then get back to us. If the local testing shows no differences between 802.11b and g on the iPhone, or differences that are in line with what you'd expect, then that's because there's nothing wrong.

If local testing shows that g is in fact slower than b, then you're on to something.

In the scientific process, you change one variable while keeping all the others constant. That's impossible to do if you're throwing the Internet and bandwidth tests into the mix.
All your blabla does NOT explain his results, download speed of B > G when downloading over the internet.

Why the heck would he ONLY be ontu something if he can verify the same results over a local network ? Over the local network, you introduce other variables, which again, might lead someone to claim "it's not comparable".

I'd like to hear a real good explanation how his results could happen w/o anything being wrong with the iPhone software.

-t
     
Tomchu
Mac Elite
Join Date: Sep 2005
Status: Offline
Reply With Quote
Oct 25, 2008, 06:51 PM
 
Originally Posted by Guy Kuo View Post
By your reasoning, it would be impossible to ever measure speed over the internet. Any measured difference in accessing the internet between two devices should be ignored even if there is a measurable difference across multiple trials with only a single control variable altered?
How do you know that was the only single control variable altered? Maybe a B client was on the G network while you were testing? Maybe the channel you're using is the same as a neighbour's channel, and it was being used during the iPhone tests? You can't definitely say that no other control variables were altered while testing WiFi in an area where there are most likely many other WiFI connections ...

Originally Posted by Guy Kuo View Post
When one attempts to measure something that has noise in the data, one takes multiple readings to allow the noise to cancel itself out.
Absolutely.

Originally Posted by Guy Kuo View Post
I understand your wanting to remove the WAN because it is difficult to hold that constant, but the WAN IS where the servers live. I do measure a difference on the --->3G <-- iPhone when it connects to an Airport express via b or g protocol and talks to web servers. I change ONLY the airport B/G status between trial runs and I minimize the WAN variability by doing multiple trials to reduce that noise. Also, an iPod Touch 2nd Gen undergoing the same type of testing does NOT have the same slow down effect. In fact its results match that the iPhone 3G in 802.11b mode and corroborates my conclusion. Were the WAN effects to swamp out the difference being measured, the touch's results would not be so equal between b and g modes.
It is indeed unexplainable. I'm not doubting the results. I'm doubting your methods and the conclusion you've arrived at.

I'll explain again, in simpler terms:

As long as your Internet connection's download bandwidth is LOWER than the downlink bandwidth of your 802.11b network, measuring WiFi performance using your Internet connection is NOT going to yield any useful results. If your Internet connection's download is somewhere between the 802.11b and 802.11g speeds, then you'll get only somewhat-useful results -- the measured 802.11g performance will be below what it's actually capable of. The Internet connection is the bottleneck. (What's your maximum download bandwidth? Looks like 3 Mbit to me.)

It's kind of like testing the maximum water-carrying capacity of two different-sized sewer pipes by testing it with a single running tap.

Originally Posted by turtle777
Why the heck would he ONLY be ontu something if he can verify the same results over a local network?
See above.

Originally Posted by turtle777
Over the local network, you introduce other variables ...
No you don't. Those variables have already been introduced simply by using that same network to access the Internet. By testing just a local server, he MINIMIZES the amount of extraneous variables. You can't control who's doing what on the Internet. You CAN control a local server.
     
ghporter
Administrator
Join Date: Apr 2001
Location: San Antonio TX USA
Status: Offline
Reply With Quote
Oct 25, 2008, 07:33 PM
 
Why not try using a standard speed test site like www.SpeedTest.net? Of course there IS the possibility of introducing extra variables by measuring speed over your LAN, because multiple concurrent downloads can cause INDIVIDUAL COMPUTER'S bandwidth to fall. So use www.SpeedTest.net and have NO OTHER computers on your network doing anything online.

Glenn -----OTR/L, MOT, Tx
     
Guy Kuo  (op)
Dedicated MacNNer
Join Date: Aug 2001
Status: Offline
Reply With Quote
Oct 26, 2008, 04:10 AM
 
Originally Posted by ghporter View Post
Why not try using a standard speed test site like www.SpeedTest.net? ......
No Flash support
     
ghporter
Administrator
Join Date: Apr 2001
Location: San Antonio TX USA
Status: Offline
Reply With Quote
Oct 26, 2008, 06:13 PM
 
Originally Posted by Guy Kuo View Post
No Flash support
Crap, I forgot that. Try AT&T's speed test then. It's not Flash based, and it's fairly comparable in its results to SpeedTest's.

Glenn -----OTR/L, MOT, Tx
     
   
Thread Tools
 
Forum Links
Forum Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts
BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Top
Privacy Policy
All times are GMT -4. The time now is 05:23 AM.
All contents of these forums © 1995-2017 MacNN. All rights reserved.
Branding + Design: www.gesamtbild.com
vBulletin v.3.8.8 © 2000-2017, Jelsoft Enterprises Ltd.,