Backhaul with 5.4 MHz AP and two clients

RouterOS General
1

Hello

I have a setup with 3 5.4 MHz Atheros/Routerboard radios for backhaul.

It looks like this:

Station1 ------ 6 km ------ AP ------ 9 km ------ Station2 — Gateway

Station1:
30 dBi MT dish
ack-timeout=43 (dynamic)
signal=72 dBm
Tx power=20 dBm
Noise floor=100
Tx/Rx rate 24 - 36 Mbps

AP:
13 dBi 5.4 GHz MT omni
signal=72 - 74 dBm
Tx power=20 dBm
Noise floor=100
Tx/Rx rate 24 - 36 Mbps

Station2:
30 dBi MT dish
ack-timeout=65 (dynamic)
signal=75 dBm
Tx power=20 dBm
Noise floor=100
Tx/Rx rate 24 - 36 Mbps


No bridges are applied - the wlan interfaces are in one subnet, and the three lan interfaces are in different subnets.

Bandwidth tests with unidirectional UDP (not on the primary CPUs):

Station1 - AP: 20 - 22 Mbit/sec
Station2 - AP: 20 - 22 Mbit/sec
Station1 - Station2: 6 - 8 Mbit/sec and very unstable.


I have two questions:

  • Are the radio-links as you should expect, when using standard MikroTik packages right-out-of-the-box ???

  • Can the setup be improved by using bridge interfaces, and if you answer yes to this question, then how would you suggest to do it ???

Thanks in advance…
KimC at gjerrilds.net

2

signals looks to be within levels.

I assume your doing routing with your setup since you say your not doing any bridging.

Would help to figure out what coudl be causing the problem knowing more about your ip setup. Also when your running the test check the CPU load on the AP unit.

Generally bridging IMO will not help you except maybe resolve some IP issues that could be easy to fall into trap. Either you have to use WDS and bridge or create EOIP between the two CPE units.

3

If your report for the AP

AP:
13 dBi 5.4 GHz MT omni
signal=72 - 74 dBm
Tx power=20 dBm
Noise floor=100
Tx/Rx rate 24 - 36 Mbps

Means that you have the TX/RX rate set on the wireless interfaces to these rates, then you should change that to allow lower rates – perhaps you lose a packet then a client/station has to reregister – which takes some time and slows the link down more.

John

4

To Tully I can say, that the tx/rx is the negotiated speed. I have found, that reducing the tx-power to 17 - 18 dBm gives a better connection. Now I get clean 10 Mbps over the two hops.


The ip setup is:

station 1, wlan: 10.72.1.200
station 1, lan: 10.76.1.10

AP, wlan: 10.72.1.100
AP, lan: 10.74.1.10

station 2, wlan: 10.72.1.1
station 2, lan: 10.67.2.10


The setup is based on static routes - and normally this kind of job is not a problem for me :wink: Furthermore, all routing works as it is supposed to.

At the max. throughput, the AP cpu load is at 85 - 90 %, which means that we are close but not quite there…

Regards
KimC

5

I don’t know how you are doing your testing, so I can’t answer.

John

6

I test both situations between two directly attached mikrotik routers with the included bandwidth testing tools. This means, that the bandwidtht testing software runs on different cpu´s (2.8 GHz P4) than the routerboard based radios. The reason for this is, that the testing tools are consuming a very high amount of cpu-power.

I use UDP testing one-way. The numeric results with udp-testing are generally higher than a real ftp-transfer between the two routers.