'interop' mode for Mikrotik SXT CPE's


#1

On the moment ‘interop’ mode is the only available working mode and for the bulk of migrating operators towards Mimosa brand they have to go this route anyway. (Unless you have the resources to change overnight…)

We work with one A5-14 that has now some 55 CPE’s able to connect to it where 50 are Mikrotik SXT’s (both ‘n’ as ‘ac’ models) and one Mimosa C5 and 4 other Mikrotik boards…

All SXT’s or other boards are within some 400mtrs distance to the tower and ‘hit’ the AP with a signal of -52 to - 58dB.
We adjusted the output power of the SXT’s to that level to make sure they all still can ‘hear’ each other. A very important issue to have the RTS/CTS system working in the CSMA (=legacy 802.11) protocol!
We’d tried lower output on the CPE’s to reduce a ‘noise’ level at the AP but it seems the CPE’s couldn’t ‘hear’ eachother all anymore and some units that ran a data stream brought the whole network down. Typical ‘hidden node’ issue. (Network collisions).

In the tdma world it was and still is, important to have good directional CPE antenna’s to get the best signals from the AP and at the same time to isolate as much as possible unwanted signal for other directions. After all, the AP directs when CPE is to send and receive. CPE’s have no need to ‘listen’ to other CPE’s in the same PtMP network. The better the signal coming from AP is isolated from anything else the better (S/N!)
Hence many metal dish type of CPE’s probably worked better then a standard Mikrotik SXT since they only have a plastic case.

But imho this disadvantage in tdma networks now works out to be an advantage in a basic CSMA system where it is needed that all CPE’s can hear eachother’s RTS/CTS messages to avoid network collisions!!

Off course spectrum around and background noise at the working frequency best should be as low as possible but since the Mimosa AP can do with lower S/N ratios as the Mikrotik AP’s to achieve certain MCS rates and overall the Mimosa chipset and software seems to deal better with the spectrum then the Mikrotik the pro of this outweighs the cons.
We saw that the replacement of two Omnitiks for one A5-14 for the same clients with lower signals deliver higher mcs rates. Due this and the fact with re-designing the 2 x 20Mhz channel (that had to be separated) we now work with one 40Mhz channel (well, we already tried 80Mhz too!) and we have an better overall network.


#2

Lots of good info here thanks for sharing!


#3

I have to correct myself here a bit in the RTS / CTS protocol add-on. Actually the rts/cts is made just for these situations where the ‘hidden node’ situation exist. Aka there is no more need all stations ‘hear eachother’ as long as they all here the main one that will have contact with all. This one, the AP, is the one that answers all RTS from stations with a CTS or ‘channel occupied’ frame. As long as all stations can hear that collisions in the network are avoided and the network runs fine.

Imho this would give the csma rts/cts an advantage over tdma since there is no time-spectrum waste the network. TDMA reserves time slots for all stations, even if they are not using it. A csma rts/cts can use all time-spectrum available as long as other stations don’t reserver time.
This probably will also be the caveat of it. What happens when many clients demand airtime? TDMA became widely used since it increased capacity in the PtMP networks with many clients and made it more stable.
But now with the heavily increased performance of the Mimosa chipset-software combo with high power cpu in a low use but high number client network it might well be csma comes a long way in outperforming tdma.

Off course, when Mimosa comes with its TDMA we might see an improvement again but so far it is not there and when it will be, only against a considerable extra investment for the operator.

Nevertheless, we need to test the A5 with more streaming video clients to see how far we can stretch the A5 csma before it collapses…