You are here

SISO degrades MIMO performance?

15 posts / 0 new
Last post
ko0ooo
SISO degrades MIMO performance?

one of our node owners has requested only MIMO devices connect to his 120 degree MIMO sector on 2GHZ, no Bullets, AirGrids etc...

a SISO device degrades a MIMO performance that much?     I believe there was some discussion about this? 

how does reflection factor into this?   Las Vegas is surrounded by mountains.    it's 26 miles direct from me to the mountain node.   I can rotate my Bullet so the mountain node is behind me...  and the mountain node still hears me and adds me to his node list.   

thanks...

Richard    ko0ooo

AA7AU
AA7AU's picture
+1

I would also be *very* interested in what the AREDN technical position is regarding Richard's question. I still have/use several M2 Bullets on -2, both in a remote Idaho [mixed] mesh island as well in the LV valley.

Thanks,
- Don - AA7AU

K9CQB
K9CQB's picture
It only matters if MIMO nodes are connecting through you.

Richard,
Anybody's MIMO node is virtually unaffected by a SISO node connection. The problem is if other MIMO nodes are using your SISO node to connect to the rest of the 'MIMO only' network. That can slow things down and cause a bottleneck.
I can see how many clubs will want a 'MIMO only' core network so that it is faster, better SNR, and responds to weather and noise better. But if you're an edge node, it won't really matter that much. We still have guys with the SISO AirGrid devices connecting to MIMO panels and dishes, but those SISO nodes are on the edge of the network, not part of the core and have no real impact on the rest of the network. They are also the gray-beards that started this whole Northern Virginia AREDN network and so we will always bend over backwards for them.
I don't control our network, but I'm still very particular. I recommend folks stop buying SISO equipment (with the exception of the uber-convenient GL-USB150) because they can always get far superior MIMO equipment for a similar price. The only thing that is more expensive is the dual polarity antenna that you now have to buy or configure, but it is well worth it.

-Damon K9CQB

ko0ooo
kinda what I thought...

thanks Damon...

looking at the mountain node stats..  one station has a TxMbps of 60 (and 10 miles closer to the node than I am), my Bullet is second with 30.     the other 6 are in the teens or single digits.    one of the 6 is another bullet, the remaining MIMO devices.  

I've stopped buying SISO devices..   but when there's 3 bullets with high gain open grid dishes sitting in my backyard, it's hard to justify the spending

Richard     ko0ooo

AE6XE
AE6XE's picture
Let's define the combinations

Let's define the combinations first.  While higher xmit power or gain might compensate, If we compare MIMO to SISO for 2 devices with the same total xmit power and antenna gain:.

MIMO <-> MIMO device:   Best overall performance. Better mitigation of fading as power is transmitted on 2 H and V Polarities (fading can affect one polarization more than the other.)   Max link rate or 65Mbps in 10MHz channel at MCS15 (and at long distances generally always means Long Guard Interval LGI).  Part 15 certified devices split the limited (<=1W ) power on the # of antenna-polarizations-chains.   (Note, the old linksys 54G with 2 antennas is NOT MIMO, rather 2 antennas on 1 xmit/rec.  MIMO is 2 xmit/rec and can double up different data across both antennas.)   We see 65Mbps on 40 mi 5GHz links, only achieved by this combination. 

MIMO <-> SISO device:   Lowest performance.  Max link rate at 32.5Mbps @ 10MHz channel.  MIMO device has 2 transmitter antenna chains.   Thus half the power is transmitted on a 90deg opposite-cross polarity and the SISO device receives very little of this energy.  Lowest SNR signal of MIMO -> SISO of the 3 combinations. (This means lower data thoughput.)

SISO <-> SISO device:   Highest SNR for a single polarization.   Max 32.5Mbps @ 10MHz channel.  Can work at edge of range when every bit of SNR is needed on a signal to receive and decode.

Note, if there is a tower site MIMO device, e.g. Rocket M5 with 120deg sector,   this node will communicate uniquely with each remote device.  A statistical table is kept of the success rate of the xmited link rates attempted with every neighbor.  ~10% of the xmitted data is done at exploratory rates to track with the optimal rate to xmit to a given neighbor.     The tower site, could transmit data to a close in neighbor at 65Mbps, and a far out neighbor only ~3Mbps.  Every other packet transmitted could be at a different modulation and coding rate.   

Joe AE6XE

ko0ooo
thanks...

you're awesome Joe.     

Richard    ko0ooo

AE6XE
AE6XE's picture
Avoid marginal links everywhere

If a far out node has a marginal link, and suppose it is only achieving an MCS0 rate at ~3Mbps on 10MHz channel.    Where it could impact the tower node, is when it has active data transmitted to the tower site.   Consider that sending a 1055 byte frame of data at 3Mbps would consume 20x the channel time compared to other nodes exchanging the same data at 65Mbps.   This marginal link node is consuming a disproportionate time on the channel.  No one else on the channel can transmit at the same time with the half-duplex protocol.   If there's a lot of data loss, then more retries and protocol overhead at this slow rate.   It could tie up the channel for the tower node to service other nodes if this far out node is trying to transmit data for any significant amount of time.   
 
The general best practice is to avoid marginal links, where ever they are (SISO or MIMO).   With enough money dropping from trees, one could always do this :) .

Joe AE6XE 

K9CQB
K9CQB's picture
Never thought of resource use, especially w/retry

Joe,
You really schooled me here. I didn't figure in the math for weak links at all. I stand corrected. It really would suck a lot of the distant node's resources, especially with numerous retries. 
I just wish there were cheaper options for high-gain, omni-directional 2x2 MIMO antennas out there. That would help a lot of these folks who keep buying Bullets. Actually, I think folks keep using Bullets because they already have them lying around, or because the word just hasn't gotten out there yet that they should opt for MIMO gear, not SISO. 

-Damon K9CQB
 

AA7AU
AA7AU's picture
Good for go-kits?

Bullet M2 with a cheap 2.4G Yagi-Uda mounted on a cheap camera tripod makes a very inexpensive alternative for quick stand-up go-kits, especially when one doesn't have deep pockets to fund new gear. It's difficult to simply discard a Bullet simply because it's only SISO. The best go-kit sometimes is what you can afford in the circumstance.

Problem is with M2 Bullets that there's only one good channel (-2) on 2.4G, and when someone on a dominating mountain top says "my 2.4G node only wants MIMO connections" it puts you into conflict.

Up here in rural Idaho we are slowly cycling out our old Bullets (in fixed locations) for MIMO gear as we can afford it, but being on a shoestring budget we just can't keep up with the rich urban crowd.

Just a different perspective,
- Don - AA7AU

AE6XE
AE6XE's picture
I suspect generally, you're

I suspect generally, you're right, these weak link nodes don't often affect the tower site.    They'd have to have, e.g. an ipCam, and someone remotely trying to view the video to tie up the tower node. Because it's a weak link, it may not be that usable and thus no one bothers to try and access the ipCam.   Another use case, the owner might be uploading large data files through the tower site periodically to have this noticeable effect.   If only using meshchat, that may not be noticeable.   The usability of the weak leak may inheritably choke off anyone trying to use it.  

Joe AE6XE

AA7AU
AA7AU's picture
It might be nice ...

It might be nice to have an option in setup to tell *your own* marginal/SISO node to stop trying to achieve a link (and thereby stop tying up distant/etc nodes) when the signal quality (over time?) just doesn't warrant it. And/or a do-not-link-to list of MAC addresses to just ignore (and even perhaps xmit "ignore me" handshakes). Problem is with a blacklist is that it can be used in less-than-ethical and/or very non-colaborative ways ... better off with "do not disturb them" rather than "do no disturb me".

- Don - AA7AU

ko0ooo
limiting links

several months ago I asked about a way to limit links with MESH nodes.    

we now have a node operator asking for only MIMO links at his Las Vegas node at 8k feet..        but there's little that can be done about reflection and the mountains that surround Las Vegas.

example..   node A on channel -2 in north central Las Vegas has a high gain SISO system pointed SE.    node B, the one asking for MIMO only, is almost 30 miles SW of node A, off the side of node A's high gain dish.    maximum attenuation of signal going SW.    yet node B still hears node A and adds it to his routing table with a single digit TxMbps     node A has no desire to link with node B, his goal is linking with node C to his SE     oh..  nodes A and C were on the air years before node B came along

nodes A and C could scrap their 2GHZ stuff and buy 5GHZ

or... node B could increase the downward tilt of his sector limiting coverage to about 15 miles...   that should help eliminate weak signals

or...   if ARDEN offered a way to limit links...    maybe using TxMbps?    ETX quality?    LQ/NLQ quality?   node operators can now set a minimum level for links weeding out those nodes with marginal signals and degrading the performance of their node.

thanks..

Richard    ko0ooo

AA7AU
AA7AU's picture
Almost a month ago you posted

Almost a month ago AE6XE posted: "Note, if there is a tower site MIMO device, e.g. Rocket M5 with 120deg sector, this node will communicate uniquely with each remote device."

Wonder why this "only connect" notice is still in effect today for a node which is setup on the only really usable 2.4G channel (-2/10) and located on top of a very dominating peak overlooking the entire southern end of the Las Vegas valley (see image captured today; call signs cropped out). Funny, it does claim to be "serving Las Vegas NV."

Aside from crippling my antenna direction, how can I tell my 2.4 nodes to not connect there?

- Don - AA7AU

Image Attachments: 
nc8q
nc8q's picture
a way to limit links with MESH nodes

In these parts, we have defined 2397 MHz LANs with unique SSIDs.
We still have channel contention, but linking is via chosen valid links on 3/5 GHz.
Then again, on our network of 35 nodes, on 2397 MHz we have 5 high profile nodes and
only 2 ham-terminal nodes.

Chuck
 

AE6XE
AE6XE's picture
As long as the 2 SSIDs don't

As long as the 2 SSIDs don't have people sending traffic to each other, this works.   If someone on one SSID views an ipCam on the other SSID, then this means these networks are connected through a tower site somewhere and now the traffic on the shared frequency is double.  If this happens very often, it would be better to share the same SSID. 

Joe AE6XE

Theme by Danetsoft and Danang Probo Sayekti inspired by Maksimer