Hi All,
I've been reading through the forums including a number of posts which address antenna co-location (on a single tower or mast) to provide 360 degree coverage. Lots of pointers on using RF shield kits, physical separation, channel spacing, etc.
A lot of the discussion seems to focus on using 5GHz in a point to point configuration with well spaced channels. This works when you're generally planning a network and coordinating channels in advance.
The use case that I want to focus on however is how to better support the uncoordinated linking of novice users who are likely to be using 2.4 equipment (ie. NSM2) on a very typical setup with channel -2. In this case, we don't have the luxury of choosing different bands or channels and there isn't any advance coordination or planning. Assume everyone will have the same SSID (i.e. likely BroadbandHamnet).
Scenario
-many users on 2.4 (ch -2) at home node locations with same SSID
-home node to home node is spotty or unlikely due to tree / building interference (thus the need for central array at height)
-opportunity to place a centrally located 360 degree array at significant height
-assume the central array is a typical tower or mast (not the ability to use 4 well separated sides of a building)
-all array radios on ch -2 to enable reach to users with typical configurations
-distance from home nodes to central array varies from 2-7 miles
So my question is this, what is the best way to approach a central 2.4 array which is all on the same channel/SSID? Maybe there isn't a great solution, but here are the options as I see it:
Option 1 - Omni (AMO 2G13) + rocket
Pro: Cost effective
Pro: No self-interference
Con: Lower performance of an omni
Option 2 - 3x 120 degree sectors, rockets, and shield kits
Pro: Best individual technical performance characteristics
Con: High likelihood of self interference even with shield kits
Con: Most costly solution
Option 3 - Several NSM2 (4 to 6) with shield kits (knowing I wouldn't get full 360 degree coverage if I went with 4)
Pro: Easy & relatively cheap
Pro: Easily repurposable for other applications (if this doesn't work well)
Con: High likelihood of self interference even with shield kits
Here are my additional questions:
1 - What do you think the real life performance of the Omni solution might be?
2- If you think option 2 or 3 would be better, what are your thoughts on vert & horz separation distance (obviously more is better....but what do you think the minimum realistic distance might be)
3 - Is there a plainly obvious solution that I'm not seeing?
Thanks in advance!!
Brian
Brian
I may be wrong but my guess is that the sheilding would be pointless. If you are on the same channel there is almost certainly no way that one device could recieve a remote signal while another local device is transmitting. However I don't think that it means that options 2 and 3 are wrong. The only dissadvantage of options 2 or 3 is that there will be a chance of missing some transmissions due to a remote node transmitting when a local node in another direction is transmitting. But even option 1 is not immune from missing transmmisions when two remote nodes that can't hear each other transmit together.
This problem increases as the number of remote users increases and as you try and increase the data throughput. Your decision boils down to whether you want to chase maximum range or highest data throughput. I suspect maximum range may be more of an interest for your case as it would help to connect users with less equipment at their end (in terms of gain or antenna height to clear local obstacles), and for users that do want to get higher data throughput add a 5GHz option.
I would be interested in other oppinions though because I could well be wrong in my thoughts here and want to learn more.
Regards
David
Question 2) My experience is there's no practical way to shield multiple nodes on the same channel at a given tower, I can't even shield from a tower 150+ yards away pointing a sector in the opposite direction with RF Armor. I'd consider this route if there were clients more than 8 maybe 10 miles out. You can get up to 3dB higher gain sector antenna over the omni with the 90deg (+2dB with the 120 deg) and receives less interfering signals with the narrowed range. You'd have to decide if the much higher cost to get the marginal increase in SNR is worth it. Vertical separation is always helpful if you decide to break up into 2 x 5MHz adjacent channels on ch -1 and -2 and can achieve higher thoughput overall.
Question 3) Not that I know of.
Joe AE6XE
This is actually by RF theory harder to do. you have to consider your not in as deep a null as a direct vertical seperation deployment would be.
Not saying that will successfully shield on the same tower, just that comparing shielding against a remote tower isn't necessarily and easier scenario.
More details are in my post on this thread (warning: lots of math) http://www.aredn.org/content/nanobridge-m2-multiple-mounts
So I've been thinking about this lately (purely from a theoretical perspective). Coupla thoughts:
An omni always interferes with itself, can never transmit to more than one station at a time and can never receive more than one station at a time.
Multiple directional stations, even if they can hear each other, have a chance of transmitting at the same time and a chance of receiving at the same time (although it sounds like the RTS/CTS protocol may prevent them from doing so).
Multiple stations may have to randomly retry to get stepped-on packets through. Maybe there's a bit of "after you, no, after you, no, I insist..." In a single omni it coordinates perfectly with itself and the packets wait in an orderly line to be transmitted.
Also, vertical separation may not help at all, at least for nanostation M2's - looking at the gain plots on the datasheet, the horizontally polarized antenna has huge side lobes in the up and down directions. Likewise for 180 degrees back-to-back. I'm guessing if they are on the same tower on same freq., they will hear each other one way or another - one little reflection will do it.
Finally, an omni receives 360 degrees of noise at high gain whereas a nanostation receives about 1/6 of that noise.
Please let me know if that makes any sense.
With RTS/CTS handshake, which will occur a lot, the sector antennas will not hear clients behind them several miles away--there will be a lot of hidden transmitters. This handshaking propagates to tell everyone to stop transmitting for one node to get through. There is an RTS/CTS trigger based on threshold packet size, the default threshold is a full size packet. This is one of the reason that we see lower LQ % values at tower sites as beacons and OLSR packets (below threshold and/or broadcasted) don't trigger RTS/CTS.
I ran into the specs of 802.11ac the other day. This would help in tower situations in the future as the protocol specification can transmit simultaneously to multiple stations on the different polarities--the different data streams can be meant for different stations. The Atheros chipset and firmware is only supporting up to 802.11n today. A dual polarity omni would then have a big advantage going to 802.11ac.
Joe AE6XE
Just wanted to thank you all for your perspectives. It sounds like this ultimately comes down to managing the tradeoffs between designing the optimal network (pre-planning channel mix/width and band usage) vs. a common standard for channel/ch width/band/SSID that would maximize the potential number of ad-hoc users. Given the use case for MESH, it would be great to identify and document best practices to get the best achievable performance using a common standard. Antenna co-location is obviously aspect to consider, but I'm sure there are many more things to cover. I'll continue to post findings as I experiment more with this.
Brian