OpenWrt Forum Archive

Topic: Extending WLAN range by relay with 2 APs

The content of this topic has been archived on 2 May 2018. There are no obvious gaps in this topic, but there may still be some posts missing at the end.

hi

I've set up a small neighbourhood wifi network that uses relay-stations and I have a few questions about that..
I thought a good idea would be to use 2 APs instead of a single WDS unit to relay the signal and avoid performance loss.
I think I already asked this some time ago, but I'm having some trouble with this and I need to be sure that this is a correct configuration.



That's the idea:
~~~ [wifi-connection]
----- [cable-connection]

Client ~~{wifi1}~~~~  [AP---APsta]~~~{wifi2}~~~MAIN-AP-STATION~~~{wifi2}~~~[APsta--AP]~~~{wifi3}~~~Client




Instead of:

Client ~~{wifi1}~~~~ APwds ~~{wifi1]~~~ MAIN-AP-STATION ~~~ {wifi1}~~~APwds~~~{wifi1}~~~Client




Most interesting to me is how the 3 different wifi networks have to be configurated or if there are any major reasons why my idea is totally wrong.... Currently I have the following configuration:

Channels: I'm quite sure about that, they have to be set different, because these are 3 "independent" wifi-networks
ssid & encryption: I'm using the same SSID and WPA-PSK TKIP encryption for all three wifi-networks to make it possible for everyone to log in everywhere.

Again: I've already set this up that way, but sometimes I'm getting lots of lost packets even with a good link and I'm yet not sure why... I already tried to use WDS instead of the APsta units, switched trough the channels and and and ... but this shows the same behaviour, so I need to be sure I didn't make a networkdesign-failure and my relay-idea is right....

May be there is a problem with different networks sharing the same SSID?
What about the distance between the AP and the APsta at the relaypoint. Do APs that run on completely different channels (at least a difference of 3 channels) disturb each other if they are placed right next to each other?

thx a lot

the topology that you think of is one that i am successfully using, be it that in my case there is no roaming and the wifi1, and wifi2 have different ssids.
An aspect that I do not know or have tested is whether the radio driver, in case it while scanning finds more of the same ssid, will log into the strongest one. In any case, going form one AP to another by a particular client, will cause a short interruption of service as this is not in 802.11 in a seamless way the way it is in cellular telephony.

Other than that a few suggestions that will work and may help you:
the point-to-point links (what you call AP-sta <> Main-AP) are not intended for clients to log into; so give them other ssids and passwords. They are also between them different, so in fact you have:
2 x omni-directional wifi for client login, and
2 x point-to-point for your backbone

Use vertical polarization for the omni-directional, e.g. using collinear vertical array antennas, and horizontal for the point-to-point, e.g. with yagi or parabolic antennas. This will improve receiver performance as it reduces the otherwise extremely strong local signal from the other radio. Even though the signal is three channels or more away it is so strong that it can reduce receiver performance on weak signals.

The rest is really doing proper routing. Depending on which AP the clients login to they should get proper IP addresses and gateway information sent per dhcp.
As the main AP will have to decide to send answers via either one AP-sta or the other it will help to assign different subnets to the client wifi links and to inform the Main-AP  through routing table entries which subnet is reached via which AP-sta.

hi doddel,

thanks for your response, I've already set this up and it's working here too, I'm just trying to figure out where my packet-losses come from...
(I already mentioned the things about same SSID, APs located next to each other etc..) it's really annnoying that the connection gets lost from time to time... maybe the same SSID for multiple networks is still not a good idea, you said you're not using it...

But actually I think it's the only way to have a great wlan coverage for everyone in my network, setting them to different SSIDs would need multiple configuration-settings on the clients sad May be I'm gonna look for a disable-background scanning or something like that at my AP units... I already know that roaming is not "live", so you'll have a small connection lost, but thats no problem for me...

Finally I also thought about dedicating the "Backbone" to the STA units, but as the network will not grow any more and it's still is a small neighbourhood network, I decided to also enable clients to log in directly there (there are about 2 clients with good direct links up to toe main-ap)

I'm using MAC-filter to prevent the STA-units to log in on the wrong APs (instead of the MAIN-AP).. may be also not the best idea, but I do not know any other.. as I wrote I also tried out to use WDS at the STA units for the uplink, but that does show the same nasty behaviour. The connection (APsta --> MAINAP and Client --> MAINAP) sometimes is great, and sometimes there is nearly not one ping that gets through...

I noticed that I'm getting a signal from up to 12 other/foreign APs at my MAIN-AP simultaenously. May these ones are just jamming the wifi-card?
I think there are products like wifi-signal filter, which filter any other signal than the one that is set... does anyone have some experiences with this kind of hardware?

Yet I'm also bridgeing not routing (wifi1,2,3 merged into a 192.168.14.x subnet). another possible failure?

So what you write about the antennas is really interesting.
I currently have 3 antennas at the top of my roof.
1) 9db omni (ch1 - neighbourhood network)
2) 12db biquad (ch14 - p2p connection to my office)
3) 7db omni (ch4 - my private wifi for home use)

I think most interferences here would be created between the both omnis, as the distance between them is not more than about 4m.
the biquad is right next to the 9db omni (less than 1m), but that shouldn't make too much problems because of different polarization?

Just some thoughts...thx

not only should pings get through, also the delay times should be low and quite constant and the modulation data rate (check 'wl rate') at a high level, e.g. 48 or 54 Mbps.
It sounds as if the spectrum is rather full there. Often a bigger problem than WLAN signals are wide-band FM modulated signals as produced by AV link systems, that send picture and sound from one room to another. They tend to destroy proper de-modulation of a high percentage of data packets as a strong frequency modulated carrier sweeps through the wlan spectrum. Microwave ovens are the other common noise source.

That is why in general it is better to use point-to-point links with directional antennas in the backbone and only omni-directional ones in the vicinity of the clients where you can produce a strong signal.  Also experiment with the channel settings to find the least polluted spot. Directional antennas at both ends have the great effect of much improving the wanted signals and weakening the signals that come from other directions. V.v. that we you also produce less 'noise' to others.  Note that in case in your jurisdiction you could use channels 13 and 14 that their spacing is different (bigger) than the 1-12 ones; that way you can push 4 radios into the spectrum, respecting minimum frequency spacing.

Regarding your antennas: the bi-quad's polarization depends on its orienation; by rotating it 90 degrees you can switch H/V.

Well, I finally found the reason for my connection-losses...

it was the 100mw booster!!... damn.... smile
everything working very well now .. with same SSIDs on differenct channels
yeah! smile

The discussion might have continued from here.