802.11 suffers from hidden node problems, network stalls, at least if trying it in some distribution network. Here counts nv2, and in old days nstreme.
We have around 100Mbit/s P2MP using "modern devices" and current firmware and in legacy device environment (RN52 radio boards) we have 20-30Mbit/s P2MP using bug fix only. Well planned, avoiding populated 5GHz bands, and with space between channels to avoid scatter from nv2, all learned from the gurus in this forum and vendor recommendations. Just mikrotik fixes the mipsbe wireless issue with newer RoS versions like this, I am happy.
With a properly configure RTS/CTS the 'hidden node' issue doesn't exist anymore. And in 'ac' this system is actually improved 'per standard'.
My region is full with other 5Ghz usage. In fact, my 30+ AP-sectors at times have to use same of close to same frequency as other towers of mine. Off course we try to keep spectral distance but with 7 other operator and some 50+ sectors in the same limited 5Ghz band this is just a hell of a task to find a relative 'free' frequency anyway.
Some of my Omni-directional AP's have clients that for 100% sure can't 'see' (so probably 'hear') each other are working fine with RTS/CTS configured like it should. (Always 'on' and AP RTS/CTS to istself)
Last night I did a test of a relative new AP. A NetMetal connected to a 30º 18dBi RF elements horn that directs to a relative quiet region. It only serves some 12 clients. Apart from one client all 'ac' clients.
This AP I tested with all 3 protocols (NV2, nstreme and 802.11) and withing each of the protocols I did some fine tuning.
This is the first AP for a long time that gave NV2 better results.
With 4 clients running the bandwidth test (to a fast router 'behind' the AP) throughput over the AP would be around 100Mbps and the best clients would both get some 35Mbps were the two other ones with some lesser signal would get the rest..
When I stopped the clients test but one, the best one would get just over 100Mbps and with some little traffic from regular users the AP showed maximal 115Mbps. This in a 40Mhz channel with CCQ into the 80+ range.
I even tried a 80Mhz channel but it didn't bring up the speeds hardly. I saw a peak of 120Mbps where in 40Mhz channel peaks reached 115Mbps....
But NV2 was the best. In 802.11 (still better then 'nstreme') all speeds where some 10-20% less and total throughput hardly pushed through the 100Mbps.....
Over the last weeks I did similar tests with similar and heavier populated AP's, all Omnitiks (both the new 'ac' models as the old 'n' types) and in all intances the 802.11 protocol gave best results.
One of these AP's is with its frequency 'squeezed' between other AP's that either 'touched' the band or even overlapped partially and again 802.11 is best. NV2 is by far worse... some 50% worse!
I also have two OmniTiks tested that definitely have clients that cannot 'see' nor hear each-other and yet again 802.11 give me 50 to 100% better results then NV2.
Show your appreciation of this post by giving me Karma! Thanks.
Rudy R. Puister
WISP operator based on MT routerboard & ROS.