582n Wi-Fi Test Results

[#1]

Quote

Looks like an interesting router. It has 5 antennas where the 582n has two. It might also be set for bonding, can you check? Your speed increase is essentially double what you were getting before which makes me think it might be set for bonding whereas the 582n is not by default. G.

This is how it works, straight from the horses mouth. From my ex military background, it's what I would call beam forming.
"Five internal antennas built inside of the RT-N56U are tasked to provide the best stable wireless signal to users. ASUS Ai Radar‘s unique design increases wireless signal coverage and quality without consuming additional power. This technology simply adapts beam patterns to local conditions to enable higher-gain directional signal emulation. Ai Radar shapes signals to offer the strongest possible output, transforming what may be a weak omnidirectional signal to a stronger unidirectional one, improving throughput at same time.
ai radar optimized and reliable wireless coverage
Others blast signals everywhere, leading to signal loss. ASUS RT-N56U locks the wireless clients to provide dedicated and stable signals."

Nick

[#2]

I use an Orange Liveradio (wifi limited to G) to listen to streaming radio. I tend to listen to the same station all the time. Previously I could listen all day/night (I leave it on and switch on/off a "hifi" with sleep function) It only dropped when the adsl was dropped. (the stream is via oranges servers not direct)
In my bedroom I can "see" 4 other wifi networks. TT (next door) was the strongest, then PN, then Virgin from across the road (same level as the PN) and a BT from across the road as well. In the room where the router is I could see (using an old router with DD-WRT on it) 4 OTHER networks from the houses behind me - the "benefits" of living in a housing estate. (the bedroom ones were seen on an Android TV stick)
SO the router I sent back had difficulty in passing through 1 brick wall, a distance of ~ 15FEET, its signal was LOWER than the TT router next door that had to pass through 3 walls and further. It was placed in the identical place as my old speedtouch which it replaced when I got fibre.
I now use a netgear 3500 and have had NO Drop outs overnight, and apart from a lot of bad noise (loads of RED on my ping graph and no streaming radio at lunch time) a few days ago (solved by rebooting both modem & router) it has been rock solid.

[#3]

Hairy Biker, I think that we both suffered from the 582n's inability to cope with walls, ceilings etc. Agree?

[#4]

Aye. It might be fine in an open field, but in an older house (mine is 50+) with REAL walls it doesn't cut the mustard.

[#5]

Quote from: Graham21

All I was trying to do was compare routers to each other under the same conditions. The problem with doing this under *real* conditions is that the conditions change and nullify the test. We would have to build a house (or at least some rooms) inside and isolated chamber. Then we would be debating whether the walls were the right thickness ; )

It seems more like you were desperately trying to show that it's possible to get a decent wifi speed out of the TG582n. Or that some of the units must have been faulty. You might have thought that if you run the test for a long time, say 6 hours, and repeat the tests, that the changing conditions in the real world would average out.
It could just be that there's no "fault", instead it's just that the TG582n's wifi outputs a slightly less powerful signal than most other routers, and that results in poor performance in a congested environment. Imagine a scenario with a 582n and a neighbouring access point broadcasting a slightly more powerful signal. If they are a fair distance apart, then the 582n might be able to receive the signal from the other AP, but the other AP can't hear the 582n. Then it would go something like this:
The 582n is transmitting something.
The other AP wants to transmit. It checks, can't hear the 582n, so transmits, causing a collision, and both will have to re-send.
or
The other AP is transmitting something.
The 582n wants to transmit. It checks, can hear the other AP, so waits.
But I expect the neighbouring APs will get blamed, as it's been shown that it's possible to get a great wifi speed out of the 582n.

[#6]

Quote from: ejs

It seems more like you were desperately trying to show that it's possible to get a decent wifi speed out of the TG582n. Or that some of the units must have been faulty. You might have thought that if you run the test for a long time, say 6 hours, and repeat the tests, that the changing conditions in the real world would average out.

The intent was more to determine if 1) what the maximum possible throughput from the router was and 2) If the units returned by some of you guys were any different.  That's the first step.  Next is add in some distance (and walls and floors!) and see if there is a difference then.

[#7]

Sorry to come late to the party on this topic. As we all know other WiFi networks near by can disrupt the performance of one's own network.
So the cellar testing (the nearest that can be achieved to a Faraday cage, (rather than an Anechoic chamber) is a good idea to identify any obvious issues with throughput on a particularv device and maybe signal strength if there's a major fault.
It's probably unwise to look for exactness in any testing because you haven't got control over the precise environment and also at these sort of frequencies a centimetre or two difference in position and orientation could make a big difference because of internal aerials on the modem/router as well as the laptop. Also there aren't too many Anechoic Chambers around that have custom built walls and floors in them that you could use, never mind for free, so looking for exactness is a bit of a non-starter as you need some real building scenario testing.
The environment can't be controlled because the strength and/or presence of the other networks may vary, they may not be on all the time, the strength of the signals from across the road will probably vary with the weather, passing vehicles may affect the signals as a result of reflections and so on.
With regard to signals coming from next door, these can be affected by individuals moving around either house dependant on their positions in relation the modem/routers and laptop. There's also the problem of other devices using similar frequencies and possible harmonics from them. The most obvious examples are Microwave ovens, DECT phones and Video transponders. So controlled tests are not an easy thing to achieve in practice.
So what can I suggest. Two things come to mind.
1) Longer tests, at least 24 hours, a few days might be good but perhaps I'll comeback to that.
2) In the real world testing, reverse the equipment locations as well. The Laptop in the cellar, so it doesn't receive any other networks or at least the signals from them are greatly reduced. Move the modem/router under test to different floors, different positions etc. for similar test periods.
The thing to look for is patterns, similarities etc comparing the results with the equipment positions reversed as well as the results from cellar testing.
Test durations is a difficult one, because you never know what those uncontrolled aspects may do. The longer one tests the more likely a representative result may be seen. Hopefully the local networks will be on 24/7 so that will at least mean some consistency.
You really need to test over a least 24hrs because things can vary between day and night.
Not in the same way as with ADSL where interference from MW propagation can have an effect, but there may likely be less neighbours/people at home during the day, so less use of DECTs, Microwave ovens and the like. The moisture content of the air may be dryer in the day, depending on the weather and time of year. The equipment temperature may be different, all sorts of things that could have an effect, and I have to say, could be different the next week/month that you test. I have also read that inversion effects where the cooling of the ground and air close to the ground at night will adversely affect propagation at these frequencies.
As I've not done any testing at 2.4Ghz myself, other than with Microwave ovens interfering with Video transponders, it would be interesting to see if there is any consistent pattern. Behaviour at these sorts of frequencies in a building can be a bit unpredictable most of the time.
I think I'd try testing all the 582n's for 24 hours each in each scenario, and then pick one or two of them and test for several days to see if any patterns remained the same.
Having said all that above there will clearly be cases where you could take another brand of WiFi kit and the performance will be obviously better immediately. In these situations, you really need to check whether they are pushing out more power, or whether it's purely antenna configuration that's making the difference.
To finish, a nice quote for you -
Radio propagation is seldom 100% predictable, and one should never hesitate to experiment. It's very useful, though, to be equipped with enough knowledge to know what techniques to try, and when there is little probability of success.
HTH in some way.

[#8]

Just an FYI, Graham has been doing some further tests to add range and walls into the mix. It's interesting.  Hopefully he'll get the chance to put out some stuff about it at some point.

[#9]

Perhaps the test should be run for at least 7 days, to take into account differences between weekdays and weekends. That should also guarantee the testing will take so long it won't be completed before these routers are obsolete and replaced with some newer model. Oh well, at least the problem probably isn't the wifi chip overheating.
I think it would be more useful to capture the 802.11 level packets with something like wireshark running on Linux with a monitor mode interface. Then we might be able to see what's going on when these routers perform badly, can compare them with some captures from where the routers are working better.
If you want to test the TG582n against other routers or other wifi networks, I would run iperf over the other networks at the same time.

[#10]

Hi Anotherone
This is very useful and interesting. When I tested a bunch of routers that were on the ground floor from my 2nd floor office, I did find that the best test results seemed to be achieved during the day. Always wondered about that. At the same time, the inconsistency of any kind of real world is very frustrating.
Just as an update, Technicolor found no difference between the test units I sent them and two random units. I found that the random units performed better under a few different scenarios. At least that's how it appeared to me. I was really hoping that they would find something, even a config change that I had missed.
My plan is to go back to the cellar ; )
The cellar is L-shaped so I can put the router near one end of the L and the laptop at the other end. This will get them farther apart and will also get a wall in the way. We've also got a spectrum analyser now which will help to make sure that the conditions remain the same.
I'm not sure how I am going to convince my wife to leave our other 582n turned off for 24 hours at a time (not to mention the two baby monitors) in order to leave a test going for 24 hours or longer. Does anyone have an unused house in the Sheffield area they can lend us?? ; )

Graham

[#11]

A couple of good suggestions there ejs, but

Quote from: ejs

...........testing will take so long it won't be completed before these routers are obsolete........

Quote from: Anotherone

............and then pick one or two of them and test for several days to see if any patterns remained the same.

Hopefully you may get some results that way before they go obsolete  
Perhaps pick the two worst from the previous tests
Graham, if you dig a big enough hole in your cellar ......... no, that's a daft idea, forget that  
Edit: sorry about that, feeling in a humorous mood this morning!

[#12]

Hi EJS
You must have posted while I was writing my reply to Anotherone.
Interesting idea about the really long test period, perhaps I could start a test down in the cellar, leaving it for 7 days and not worry about interference from my own house (the neighbous' interference doesn't seem to reach to my cellar). You would expect the various factors to even out over that kind of time frame. Except the temperature of course, which will hopefully get warmer!

Graham

[#13]

The following known wireless issue has been reported in Technicolor's release notes documents for all the firmware versions I've seen.

Quote

When using WEP encryption or WPA-TKIP wireless security the throughput performance of the 802.11n wireless interface is lower than when using no wireless security or WPA2. http://download.modem-help.co.uk/mfcs-A/Alcatel/Modems/TG582n/v1/Manuals/Release-Notes/EN-r8.4.5.I.p...

When creating a new wireless connection in windows you don't get the option to specify the encryption type. Mostly it seems to select AES but does sometimes select TKIP.
I've got the impression that if one wireless device uses TKIP encryption then that slows things down for other devices regardless of the encryption they use.
Thought I'd mention this incase it's relevant to the problems in this thread?

[#14]

That sounds like the general "you can't use WEP nor TKIP with 11n" which applies to everything.

[#15]

I'll follow this thread with interest.
I've done numerous tests in the old house, a three storey Victorian terrace, and need to do some more in the new house - a four storey Victorian terrace with brick wall "cells" in the cellar. In both cases the first and second floor signals have to pass through one load bearing brick wall. Used the same channels in all tests with the destination devices sat on desks in the bedrooms and study in the cellar, and using some metering software on my laptop.

• Billion Bipac 5400G: G only. Signal extends all through the cellar but with only ~30% signal in the ones furthest from the router. First floor signal strength is awful - around 20%. Signal does not reach the second floor.


• Buffalo WHR-HP-G54, Tomato firmware. G only. Signal extends much further than the BIPAC's - first floor signal strength ~60%, second floor ~30%.


• BT Home Hub: G and N. Better than the Bipac in the cellar on G and N on cellar, ground and first floor only. G does reach the second floor but keeps dropping.


• Technicolor 582n: G and N. G performance on first floor, ground floor and cellar is not as good as the Bipac. N performance drops off in the cellar's "cells" and doesn't even penetrate the non-loadbearing walls of the the first floor.


• D-Link DSL-2740R, dd-wrt firmware: N only in bridged mode. Three aerials. This is the only N router with an N signal that gives >130mbits/sec consistently everywhere else in the house except the second floor.

By way of an aside, we can also see a SSID from TalkTalk (Wireless G) in our house and it's coming from the house opposite. Specifically, that signal is passing through their triple glazed window, their very thick 10' hedge, a wide main road, my tall fence, my garden shed and my bay window and after all that its signal strength on the second floor still outperforms that of the Technicolor  which is only the other side of two lathe and plaster ceilings + 1 foot crawl space with no pipes or electric cables in it.
The D-Link is something I paid a fiver for at the Reclaim shop at our municipal recycling centre. The ADSL circuitry is FUBAR so I'm using it primarily as a switch and wi-fi range extender because it spanks everything else. Even when it's sat only 20cm away from the BT Home Hub, the wireless performance difference between the two boxes is astonishing.
So I'd say the key factor for the wifi issues with this router is the builtin aerial in these routers. Unlike the D-Link, you can't position the aerials so even if it was more powerful you are a bit stuffed if you've got devices above, below, to the left and to the right of the Technicolor.
My in-laws had this router before me and had no complaints but then their router was in the front room, so were their laptops, and it's a 1980s built house with Not Very Solid interior walls.

Edit: In other words, the blind spots in a four storey town house are likely to be above, below, beside AND behind the router with brick walls, wire mesh, pipes, cables and lord knows what else creating interference. Sticking the test router in the far corner of a basement only allows you to check the blind spots in front of and above the router. Dumb suggestion - turn the router upside down and back to front, then repeat the test.