it would be interesting to hear, are there any tests for backhaul links 5+ km range in 802.11n mode and what speeds are(could be acheived) i.e. with 2 dish antenas on both sides.
I wouldnt exactly call this a success story,but maybe this ought to answer alot of people’s questions with regards to the SR71A made by Ubiquiti and RouterOS.This morning i upgraded my two RB433AH routerboards to the lates version 4 beta version as well as upgraded the BIOS,clicked “update license key” and pronto,after a quick reboot Mikrotik detected the two cards.Enabled them and got them wirelessly associated to each other.Although i only got 35mbps through them on a throughput test,i didnt exactly spend much more time trying to fine tune it as i ran out of time.But both boards were simply sitting next to eachother on the workbench and noise levels were quite high which affected my CCQ,which in turn affects the total throughput.Perhaps when i get some time i’ll do an outdoor link and post test results.
yes i used two antennas on each card,i did an outdoor test over a couple hundred meters using a dual polarized antenna with RB433AH’s either side using the SR71A’s.I also had two pc’s connected to each board on both ends and used the mikrotik bandwidth test to do the tests through the boards for accurate results.I Suspect i might have been able to get faster speeds as the 433AH’s only have 10/100 ethernet ports as compared to the RB600A’s with Gigabit.Despite that i managed to consistently get 95.7Mbps without any drops.In my opinion i think the new R52N would perform a bit better though
HT Extention Channel - none, above control, below control
User selects a control (primary) channel (frequency) and sets if the bundled(bonded?) channel is above or below the selected channel.
HT AMPDU Priorities
AMPDU allows bursting 802.11 frames up to 64KB and is performed in the hardware. Not quite sure what these settings do yet
HT Guard Interval - any, long
Short GI is 400 nanoseconds vs. the traditional GI of 800 nanoseconds. Short GI reduces the symbol time from 4 microseconds to 3.6 microseconds to improve throughput by 10%.
HT MCS - Currently unchangable
MCS selects, based on RF channel conditions, the best combination of 8 data rates, bonded channels, multiple spatial streams, different guard intervals and modulation types (noting that 802.11 b/g adapts to channel conditions by selecting the highest of 12 possible rates from 1 to 54 Mbps)
Antenna Selection - Not sure if this should be enabled for use with 802.11n Given the Tx/Rx Chain settings?
HT TX/RX Settings - I'm still not entirely sure how the Tx and Rx chain pairing works, but I know that the card I have has 2 antenna sockets although I see only a small increase between having 1 or more selected.
Improvements we might be seeing available in Router OS:
Preamble Mode: long, short, both, greenfield (new)
Greenfield preamble cannot be interpreted by legacy stations. It is shorter than the mixed mode or legacy mode preamble and improves efficiency of the 802.11n networks with no legacy devices.
Would any mikrotik reps care to correct me here or add any further comments?
After a day or so of messing about I’ve gotten the following results…(thanks to omega-00 for the excellent information)
[w/ conn. track]
TCP: 60.6MBps (peak) 60.1MBps (average) Very stable (w/ conn. track).
UDP: 179.8MBps (peak) Very unstable.
Ping Times - 1ms/2ms/22ms (min,avg,max)
[wo/ conn. track]
TCP: 64.5MBps (peak) 63.2MBps (average) Very stable (wo/ conn. track).
UDP: 196.7MBps (peak) approx. 165MBps (average) Stable.
CPU Usage approx 30-35% on Master Draft-N unit (bridge unit).
Ping Times - 1ms/1ms/8ms (min,avg,max)
Using 2x RB600A boards and TP-Link Draft-N cards.
Using 2x RB600A boards to generate traffic.
Very noisy environment. Scanning detects 5x APs in the building. One AP running 100mW about 10 inches from the one test board. Turning all this off makes no difference.
Channel: 2412 (other frequencies perform more poorly)
AMPDU: All selected on both units (definite bandwidth improvement)
HT Extension Channel: Above Control (definite bandwidth improvement)
HT TX/RX Chains: All selected on both units (definite bandwidth improvement).
These three changes upped the UDP bandwidth from 95MBps peak to 179.8MBps peak on UDP and from about 44MBps to 60MBps on TCP (stable). Disabling connection tracking boosts the TCP average by about 3MBps and makes UDP more stable with the peak improving by about 20MBps.
Tried various other settings, but none made a difference (except to possibly worsen it)
A> does nstream (single and/or dual) work with the 11n adapters yet?
B> Does the MIMO work in b/g mode (support 2x2? 3x3?)?
C> Can you control the channel usage (ie - using it on sectored towers, I’d rather avoid as much crosstalk as possible)
D> 5/10 mhz rates? Turbo-a/g compatibility? Is there a such creature as HT5 and HT10?
E> Can I use a PCI Express type card (ie - in a x86 motherboard’s PCI-Ex 1x slot?) with MT?
F> My logic on MIMO applications is the antennas should be in the same polarity (idea being that you’re more likely to get a narrow obstruction [tree limb, light pole, whatever] than a cleanly spun signal). Any input on this?
I’m hoping to test b/g or g-only MIMO on customers that have issues with low data rates due to tree multipath.
Uldis also updated some missing info in your tutorial, more specifically the Chains. I see that you used an old generation card with three chains, new ones use only two to get the same performance. Chains - Antennas.