You are here

5.8 GHz Point to Point Link info wanted...

10 posts / 0 new
Last post
K6OQK
K6OQK's picture
5.8 GHz Point to Point Link info wanted...

Gang,

I want to create a point-to-point link from my home in Glendale, CA to Mt. Wilson for my Amateur Radio repeater.  The distance is 14.2 miles and optical between the proposed ends.  I would like to use 4' dishes at both ends, not so much for gain, but for a narrow beam width (view) from Mt. Wilson, and extra gain at my home to hopefully be able to override interference from other users - as you well know, Mt. Wilson is a high profile site.  This would be used only as a point-to-point control link and not linked to any other system.

I'm thinking of Ubiquiti equipment.  How much through-put do I want?...  As much as I can get.  :>

I would like your recommendations and hopefully, specific model numbers and links to specific equipment suggestions would be appreciated.

Thanks in advance,

Burt, K6OQK
Glendale, CA

AE6XE
AE6XE's picture
Burt,   Assuming 3GHz stays
Burt,   Assuming 3GHz stays around for part 97 use, a couple RocketDish M3 devices would be ideal, these are 2' dishes.  3GHz should be the quietest space at Mt Wilson.  Alternatively, go to 5GHz Rocket Dishes, also 2'.  If $$ are tight, then I'd fall back to PBE-M5-620 iso devices -- much more cost effective with minimal reduction in EIRP.    The RocketDishes are achieving 65Mbps, 10Mhz channels, over 40 mi links in SoCal.  At the distance you are considering, these devices would give the best option of success in this price range.  

A dish in the 4' ballpark would be significantly higher $$$$, and would certainly cut though a lot more noise.   These may be harder to source as they would exceed part 15 power emissions and not commonly used. 

There are AREDN nodes at Mt Wilson already.   Reach out to Orv W6BI to compare notes, if not already.  People are connecting in to 120deg sector panels on Mt Wilson on 3Ghz today at 15+ mi.   I don't recall what is covering your area, but maybe there is a way to test from your QTH today?

Joe AE6XE
N7JYS
Tx Mpbs

Joe, Im curious if your 65 Mbps is true measured thru-put speed or is the what OLSR is indicating, ie LQ (100%) NLQ (100%) TxMpbs (65).
The reason Im asking is Im using two MikroTik RBLHG-5HPNd-XL for a 20 mile back haul  and OLSR reports a 103.5 TxMbps with (100%) LQ & NLQ
using 5.825 @ 20mhz. Im feeding it from my home ISP with  20 Mbps and only get a thru-put speed at the other node end of 10-12 Mbps.
Am I missing something or is the equipment your using that much better?

Eric     

k1ky
k1ky's picture
TX Mbps
Joe's suggestion of the UBNT PBE ISO units is a good one, providing isolation from the sides at congested sites.  How are you measuring the throughput? (IPerfSpeed?) 
AE6XE
AE6XE's picture
Let me diverge a little and
Let me diverge a little and go into some detail if this is new to some reading this post.

2 common ways to describe the link in terms of bps:
A) the raw protocol rate at layer 2 of the communication stack
B) the application data throughput  at layer 3 of the communication stack

For 'A', the protocol rates are in the 802.11n specification and called MCS0 to MCS15.  In commonly found tables of these speeds, be sure to use the Long Guard Interval (LGI) timing of symbols.  Although on short distance links it might be able to select and use the Short Guard Interval (SGI) and have slightly higher link rates.  Note, what is shown in mesh status is adjusted:  (MCS rate) * (actual packet success rate).   Thus in mesh status it might show, say ~98% of the MCS rate.   It's an attempt to account for known packet loss.   This rate is to a specific neighbor and the packet success is only the unicast or traffic addressed directly to that neighbor at the optimal MCS rate. 

For 'B', this is generally measured using iperf.  It will be considerably lower than 'A'.   But put into perspective, a stand HD video stream from NetFlix will be 1mbps or less.   If you have achieved 10mbps iperf measurement, this is 10! HD video streams.  That's a functional link that will meet a lot of demand.  Most home internet users with 100mbps premium service, rarely go over 10mbps use. 

LQ% and NLQ% or not very good indicators.  This is based on broadcast packets (addressed to everyone on the channel) at the lowest MCS0 rate, the common rate all neighbors can best receive.  There is no use of RTS/CTS for these transmissions and packets are easily lost.  When an iperf test is done, the protocol will use RTS/CTS and can reserve time slots on the channel, which includes hidden nodes, to minimize loss of data.

Back to the question...
In my case, the 65mbps is the top MCS15 protocol rate @ 10MHz channel.  It's been a while since I did an iperf test on this link, it was over 10mbps.    3 or 4 hams have connect in to this P2P backbone link, so iperf rates may not be as high in recent times :) . 
 
Joe AE6XE
w6bi
w6bi's picture
HD Video Streams
Joe, I found this - I think's it's correct, based on what I've seen:

Netflix recommends a 3 Mbps connection for one standard-quality stream and 5 Mbps for a high-definition stream. Two simultaneous HD quality streams would need around 10 Mbps, and so on.

So a 10 Mbps mesh network stream could maybe support two HD streams.   Of course, running 720p resolution at 10-15 frames/second vastly reduces the required bandwidth.

But props for the great explanation - I learned something!

Orv W6BI
 
AE6XE
AE6XE's picture
Yes, the 5Mbps is the "A"
Yes, the 5Mbps is the "A" basic or minimum internet connection speed NetFlix is recommending consumers have to watch HD video.   The marketing hat selling internet services always uses "A", the bigger number :) .   The 1 mbps or less is to compare directly with "B" of actual data thoughput or iperf measure. 

Netflix uses adaptive bitrate streaming to ensure the video can be enjoyed verses dropping out.   At times we might see the image getting a bit blurry, but can still enjoy the movie.   Netflix might go down to 256kbps ('B') or lower if the internet connection is saturated or degraded for any reason, not sure how low it will go.  

ipCams will have (we need one that has) a config setting for this 'B' video stream rate.  I set all the ipCams for events to ~750kbs, 15 fps, and 720p, which is still a great picture.   Too often I've been bit because I forgot to change a default setting and the camera is at 4Mbps, 30 fps, and 1080p -- floods the network with an ipCam on this setting.

Joe AE6XE
AE6XE
AE6XE's picture
MCS rates and channel width
MCS rates and channel width

If looking up the MCS rate, these are always in context to the 20MHz channel width 802.11n specification.   If using a 10MHz channel width, then the table rates need to be cut in half.   If on a 5MHz channel width, the rates are divided by 4 from the table.
    
Joe AE6XE
AE6XE
AE6XE's picture
Eric, you might try an iperf
Eric, you might try an iperf test @ 10MHz channel width on your link.    The max TxMbps will be 65 or less (the 'A' measure).   However, the iperf ('B' measure) that the 10MHz channel will support may be comparable and in some cases better than the 20MHz channel.   This has to do with the fading issues on long distance links.
   
On a 10MHz channel width a symbol will be transmitted in 2x the time it is transmitted on a 20MHz channel width.  Thus the receiver in 10MHz channel has more time receiving each symbol so a delayed reflected signal of the prior symbol doesn't corrupt the next symbol.   The 802.11n 20MHz specification is primarily designed for short distances, 10s to 100s of feet.  More time is needed proportionally as the link gets longer to mitigate reflected signals or fading and avoid inter-symbol interference (ISI).

Joe AE6XE
N7JYS
Yes that is correct.
Yes that is correct.

Theme by Danetsoft and Danang Probo Sayekti inspired by Maksimer