In the documentation of the ar9220 chipset it does not show anything about spatial diversity support. However it supports spatial multiplexing. Does anyone know where I could get more detailed specs on the ar9220 chipset? I want to know all the features of it.

Hi ...

I hope this 'feature' still avialable, officialy or not, at MT ROS.

I'm using it at R52n (EOL card now?), R52Hn & R52nM equiped sites not exactly as spatial diversity harness yet but as sort of "multi beam" antenna array that feed customers at legacy 802.11G modes.

Next days I'll deploy a 433 with R52nM and two 90° sectors in place of a pair of 411ARs. This 433 will have a second R52nM @ 5GHz + 2 x 29dBi grids vertically spaced to solve a short link (1.6Km) over water that suffers some disruptions due to evaporation, airplane scattering & etc.

Regards;

The mikrotik card is ar9220. I see you using the card as if its supports spatial diversity. But I find only documentation of spatial multiplexing is supported. Please help me find documentation that the ar9220 supports spatial diversity.

Hi Dallas,
It's part of the 11n standard for chipsets to support spatial diversity & spatial multiplexing.

I agree with you but this chipset does not show spatial divirsity for support. I want to make sure everyone knows that. I know everyone is assuming it does. I try to go by facts. Does anyone have any documentation that they can show us that this chipset has support for spatial diversity? Thanks in advance.

Also I found in the 802.11n specs. The spatial diversity is not required. You can check to make sure I read this correctly.

http://www.wi-fi.org/files/kc/WFA_802_1 ... June07.pdf

I´m little bit busy this days

Well, I used to think that on "old fashion" spatial diversity I'll have 2 antennas, same polarization, vertically spaced, each one connected to a spatial diversity enabled ODU, e.g. a dual receiver, dual down converter, dual IF ... and some RF switch that will be driven by received signal level.

Then the RX IF signal will be routed to IDU, always the strongest one. Better the one that has better S/N.

May be for a single PSK modulated carrier it was ok. Some drops when switching and data errors recovered by some retries here and there.

I guess today it's possible to track bit error rate and chooses the smallest value and use this receiver. Add some buffers and no packet loss.

May be a step forward use both chains and "combine" correct data from each one to have the final packet.

I'm watching this "feature" simply as an aditional resource for - under certain circumstances - extend receiver operation withouth errors. Signals are already OFDM so destructive multipath affects some carriers but not all of them (as a single carrier PSK, BPSQ, QAM etc modulated).

I guess multiplexing on such chips extends rx threshold, on M mode increasing the troughput.

But on G only seems there's not multiplexing, may be sort of switch (mux) uses data from a single receiver, I don't know if RSL based, SNR based, BER based or what.

The "beauty" I intuitively observed is that the eventual switching from one antenna to the other is somewhat sinchronyzed so there's no detectable disruption on raw data stream when switching.

And at this test I'm doing - instead of use 2 cards, 2 antennas, 2 AP names, telling each CPE to lock to mac XX.XX or mac YY.YY, mainly those where the signal difference from AP1 to AP2 is small - this unofficial spatial (or mux) resource is helping a lot.

I'm ready to test a PTP with two R52nM configured the same way. May be 2 antennas vertically spaced, may be one V another H ...

This link is over the water (2Km) and both sides have dense vegetal covereage so lots of reflections, delays & etc. Early in the mornings there's lake evaporation that causes some deep fades (10 ... 30min around 6AM). Besides that, some airplanes cross by, 1 ... 1.5Km above, creating some scattering. Depending on the time of the day, one airplane each 4 minutes.

Another nice setup I finished this weekend: I have 2 distant APs, one of them using a 15dBi omni and a grid to a remote area, both APs at the same pole. One CPE 1.2Km away without rain/fog always connect (-70dBm) to the omni. But under rain/fog this signal starts to drop while it grows at the grid. Untill then it has two or three outages a day and netwatch let me know abt. Manually I switched the AP at this CPE. Now ... it's connected since last saturday evening, zero drops.

Regards;

in your 2x2 both vertical setup. If you turn one antenna off(disable 1 tx chain from each side). What db loss do you get? I want to know what dbm value gain we get for having 2 vertical antennas on each side.

in your 2x2 both vertical setup. If you turn one antenna off(disable 1 tx chain from each side). What db loss do you get? I want to know what dbm value gain we get for having 2 vertical antennas on each side.

Hi Dallas ...

I does not detect any difference on signal levels (cpes are ubnt).

I think it's because one chain normally have signal > 10dB than the other, so adding lets say -65dB + (-75dBm) = -63.5dBm ... but rsl floats a bit, at least last digits so uncertain readings to detect 3dB changes.

But - I have a setup at my backyard, with a R52nM as AP. I'll add a pair of omnis (those small 3 dBi ones) side by side, one for each chain, and measure the rsl with a Bullet2 + 9dBi patch. Will left ap at a fixed data rate and collect rsl data on Dude ... and then use chain 0, chain 1, both.

I'll post the results here.

Regards;

From documentation I have read. I believe if you have 2 chains(2antennas) and one spatial stream. The 2nd chain is used for spatial diversity. What this does is send the same bits over each chains. If you have 2 spatial streams. Then it will some some bits over 1 chain and the other chains gets the other half of the bits. One spatial stream enables multipath. Keep in mind all MSC rates haves redundant parity bits. But we are talking about diversity for multipath redundancy. I need to do more testing to prove this information. Also diversity is suppose to give you 1-3db more power overall. Because it gets same signal from both antennas. There is a formula for that too.

Here is documentation that helps support my research.

http://s2n.merunetworks.com/2009/08/wha ... 11n-works/

Have you guys checked out some of my videos? I will make better ones.

Agree.

On legacy modes more test are needed.

If total bits are shared by both chains ... what will happen when - for a single chain - multipath became destructive (e.g. direct signal is nulled out by out of phase reflected one)? The second half of bits will be lost? So several retries at the good chain to recover packets in a healthy way?

I'm not used to OFDM x multipath. I think it's more detectable on narrow band modes because the notches can be wide enough to kill the signal. But at a 20MHz wide OFDM signal may be we lose part of those 20MHz (some carriers). But I don't know how many carriers we can "discard" before an 802.11 A/G signal became useless. 10%? 20%? Still a little bit more than 50 carriers anyway.

Some years ago I did some test with single carrier QAM (32/64/128) on 8GHz (28MHz channel, up to 155Mbps). It was possible to keep it working with a 18dB notch inside the channel bw, the modem equalizer compensate it (I guess it has 40dB range per symbol). Looking that numbers OFDM seems the way to go (if there is bw for that) but our target was max bit per hertz and a power amp for OFDM was too expensive these days (and I guess still expensive).

Funny to remember a 76Km 8GHz link, 100Mbps, 32QAM ... 1.2m dish one side due to wind loading at one tower, a 2.4m dish another end. SLA (measured): 99,998%

Regards;

You can also use spatial diversity like this. Use 2 chains for rx and tx on each side and 2 spatial streams. Doing this, you will have diversity. Each spatial stream will choose which antenna to receive. The diversity picks the best signal that is received on each antenna. But you will not have redundant signal. Which is what I would want. However doing it like this you would get double the throughput. Whats your opinion?

This is spatial multiplexing and diversity by best signal.

I also found this:

Diversity leads to improved link reliability
by rendering the channel “less fading” and by
increasing the robustness to co-channel inter-
ference. Diversity gain is obtained by trans-
mitting the data signal over multiple (ideally)
independently fading dimensions in time, fre-
quency, and space and by performing proper
combining in the receiver. Spatial (i.e., anten-
na) diversity is particularly attractive when
compared to time or frequency diversity, as it
does not incur an expenditure in transmission
time or bandwidth, respectively. Space-time
coding [5] realizes spatial diversity gain in
systems with multiple transmit antennas with-
out requiring channel knowledge at the trans-
mitter.