2.4Ghz 802.11 b/g/n speed

rjcrowder

i have converted a wireless router into a WAP and i am using that atm but i just wanted to see if the internal card is better and reliable

I held out hope for a while that 2.2 (with Freebsd 10) would offer a more stable solution that would allow me to build an "all-in-one" box with an internal access point. I bought several "N" cards and played around with it for quite a while (you can search the forums for some of the threads). In the end I gave up… Problems ranged anywhere from core dumps and reboots, to the UI allowing you to pick options that didn't work, to not getting true "N" speed. I realize some of the problems were probably FreeBSD and driver related, but it still made for a very poor experience.

My take on it - I wouldn't use an internal card as my main wireless access point. It might be fine as a low volume interface for administration or something, but that's about as far as I'd trust it.

robina80

thanks for the heads up about the internal pcie card will be naff

but if my devices are all wireless n (2.4ghz) i should get LAN speed of 300mbps correct, thats if the card supports it which it says it does?

Derelict

Nobody gets 300Mbps on N. It's theoretical. Marketing hype.

johnpoz

All numbers for b, g, n, ac are all marking hype!! /2 for somewhat closer to what you might see.. You got to love those N300 device they sell, yeah they have a 10/100 interface - where exactly are you going to even see 150mbps to??

robina80

so roughly if all my devices can do wireless n (2.4ghz) and my card can do wireless n (2.4ghz), what real term speed do you think i can get from my LAN wireless

Derelict

It's wireless: "It depends"

robina80

johnpoz, i know what you mean, i have a tplink 500mbps powerline adapters but the ethernet interface is 10/100 so its complete BS

thank you everyone for your help

GruensFroeschli

Take a look at http://mcsindex.com
The column on the left shows the MCS rates and the resulting speed you can theoretically get in the middle.

300Mbit is MCS15, HT40, short GI.

MCS 8-15 mean there are two spatial streams.
You can only get multiple spatial streams if both the client and the AP have at least 2 antennas and corresponding chains behind it.
Additionally you need enough reflections in the environment to actually be able to get 2 streams.
In a LOS setup you usually don't have enough reflections, so you will get only a single spatial stream –> MCS 0-7.
The effect of switching between a single and two spatial streams is having double the bandwidth.

Guard intervals are used to make the signal more robust against distortions from echoes, propagation delay, and reflections.
So while you need these things to be able to have multiple spatial streams, they also destroy your signal if there is too much.
If the distortions are low enough, minstrel (the rate controller) will switch to shorter guard intervals (from 800ns to 400ns) to free up more time on the air for actual data.
Usage of SGI makes for about a 10% increase in throughput.

Originally 802.11 uses a channel width of 20MHz. On 2.4 when using channel 1(2412), this means that your signal is from 2402 to 2422.
802.11 introduces HT40+ and HT40-. This allows to increase the channel width from 20MHz to 40MHz. For channel 1 only HT40+ would be possible. The resulting signal would be 2402 to 2442. Main channel 2412(2402 to 2422), side channel 2432 (2422 to 2442).
As you can see this also means that you now have more overlapping channels.
Some implementations of APs refuse to start HT40 operation if there are other APs detected on 2.4GHz which would overlap.
Using HT40 about doubles the available bandwidth (a bit more actually).

So it really depends on your environment how much you actually will get.
Are you alone? Can you actually use HT40?
Especially on 2.4GHz, if you are not alone on the spectrum you hurt your throughput more by using HT40 than gain something from it.
You also hurt everyone around you.
Do your devices have actually 2 chains? Hint: mobile devices usually only have a single antenna and thus only one chain.
The maximum you can get with a single chain is 150Mbit.
What is the environment you are running in? How much metal is around? Lumber walls? Steel concrete?
This all influences how much streams you will get and if SGI can be used.
The last factor is RX sensitiviy and TX power.
Depending on the quality of the card you are going to use you will get different rates at different distances.

The datasheet to card you linked is here: http://pcengines.ch/pdf/wle200nx.pdf
(The card is actually a/b/g/b, and not b/g/n).

If you take a look at page 3 you will see the specifications of the TX power and RX sensitivity.
If you look at the TX part you will see that for MCS0/8 the specified power is +17dBm per chain.
For MCS7/15 it is only +10dBm per chain.
Using HT40 this goes even further down.

Looking at the RX sensitivity you can see that for MCS0 a sensitivity of -93dBm is specified.
This is with 2 chains. If you can use only a single chain this reduces the received power by another 3dB, so -90dBm.
Although if you are going to use this card as an AP i suppose you will connect both antennas.
Still -93 dBm is ok.
However if you want to use the higher rates like MCS7 you need a signal level of -72dBm.
If you want to use HT40 the sensitivity goes down further to -70dBm.
Since you are probably using consumer grade devices as clients as well, they wont be much better than this card.

In the end you won't get around setting up a test-bench and test it.
My basic expectation of n-rates on 2.4Ghz is that you won't get HT40 because of the overcrowded channel, but have enough reflections for two spatial streams. Usually too many reflections so SGI will switch on and off.
Depending on the distance/attenuation you will get something between MCS10 and 15 with drops to MCS 2 to 7, which will usually recover back to dual spatial streams.
In Mbit. ~80~100Mbit with drops to ~40~50Mbit.
This does not count in inefficiencies in the protocols used to transfer data.

Might I add -Speedwise there is still work being done on the ath(4)

See the todo:

802.11n TDMA support

In progress

802.11n rates work, A-MSDU support is needed to reach full 802.11n throughput

https://wiki.freebsd.org/dev/ath(4)

I have been very happy with pfsense wireless, Both WLAN and WWAN

robina80

thanks GruensFroeschili that was very useful reading

so basically i need to up the bandwith per stream to 40mhz and then i will get 300mbps instead of 144mbps if it was 20mhz per stream this takes into account i have 2 MIMO antennas

cmb

We've imported FreeBSD 11-CURRENT's Atheros driver and net80211 bits for 2.2.1. Thus far in our testing, things work better than ever. We'd appreciate additional testing. The latest snapshots available at snapshots.pfsense.org include that. I encourage those using wireless to try it out, especially if you see issues in 2.2 and/or earlier versions.