I am thinking of upgrading a small tower site that currently has Airmax AC to a single 4500C w/4 90deg horns in a N/S/E/W (omni) arrangement. Any issues or quirks I should be aware of? I have never done a split sector AP before.
I have not tried it myself and am intested as well. I have seen others post that they have done so and it appears to work. Not sure what if anything it does with mu-mimo.
I would recommend looking at the RFElements coverage simulator and making sure the 90 degree horns will work for your application. We have deployed a few 90 degree horns and the drop off on the edges is real.
The only other concern would be frequency usage at this location, as you only use one channel. So if you have noise from the north, you have to change the channel for the north sector to work well. Now it does depend on how small of a tower site we are talking, as having signals really high signals makes that a non-issue. My understanding is that the spectrum analyzer only works on two chains of the eight chains as well, so tracking down the noise when you can only see one direction would also be difficult.
I have been tempted to do the same at one of our sites where we are also converting from Airmax AC, but I instead decided to go with two 4500c’s in dual sector mode with two 60 degree 4x4 antennas each as the noise is too high from our other equipment along with other providers in the area. We also get the 3db of beamfoarming or multi-mimo on each of the four 60 degree sectors. We will also have a single 4500l for the last little bit of overlap as 60x4 leaves 120 degrees, but there is only around 5 current clients in that 120 span so it should do just fine.
Not sure about the costing, but doing 4x 4500L’s with ABAB frequency reuse might be a safer idea when it comes to noise and picking better frequencies (also much easier to install as wiring and taping up a 4500c with 8 antenna leads sucks while on the tower)
1 Like
Tower currently has one Rocket5AC Omni and two 90deg Prismstation 5ACs covering the NW and NE directions. Total clients on tower are 39 with 34 of them under 3 miles and the rest no more than 4.5 miles from the tower. Of those if I just moved my best clients based off their existing Airmax Signals I would be moving around 24 of them to the 4500 and leaving the rest distributed on the existing AirmaxAC APs.
I first tried a single 4600L with a 90deg WB horn as a test, but found that the antenna pattern of the WB horn was not great for my application and that was compounded by the limited EIRP in the UL direction of the 6ghz radios.
I am thinking the 5ghz will work better due to the higher power EIRP allowed in the UL direction.
The intent is to move all the good clients and heavy user clients to the 4500 and leave the AirmaxAC for poorer clients.
I am checking CPE distribution but it looks like a NW/SW/NE/SW arrangement is the better setup.
Follow up question, but is MU-MIMO supposed to function at all in an Omni arragement or does that only function if you have antennas facing the same direction/or overlapping the same or similar direction?
So I guess the question is what Antenna modes is MU-MIMO/Beamforming available for.
- Sector (8x8 sectors)
- Omni/Split 2x2 - (Four 2x2 antennas is with no overlap)
- Split 4x4 (Two 4x4 or Four 2x2 antennas that overlap or cover same direction?)
MU-MIMO is only available in 2x 4x4 (back to back) or 8x8 mode. It’s not available in 4x 2x2 “omni” mode.
Is that a hardware limitation or a current software limitation? If software will they eventually make it work in 4x 2x2 mode?
I would think you could do two groupings by making it so MU-MIMO only functions when talking to clients that are connected to back to back 2x2 antennas as long as the antennas do not overlap on azimuth.
Also I guess what the benefit is then to using a 4500c for onmi coverage? Why not instead just use a 4500L connected to an onmi antenna?
I stand corrected… when the 4500c is being used in omni mode, it does not simultaneously send data to each sector. SM’s are getting allocated time slots one by one. You won’t get any beamforming gains in this configuration, but that’s beside the point. For MU-MIMO to work the radio needs to be used in 8x8 or 2x 4x4 mode.
As to why to use a 4500c vs. a 4500L connected to an omni… I think the main reason to use a 4500c is that you have much more granular control over the antennas used and physical characteristics (direction, tilt, mount location, F/B ratio, etc), along with being able to use much higher gain antennas… as opposed to a single omni antenna. Basically, better noise rejection, and concentrating energy direction.
I was just corrected by the ePMP dev’s… and NO, MU-MIMO does not work in omni mode. SM time slots are allocated one by one. For MU-MIMO to work the radio needs to be used in 8x8 or 2x 4x4 mode.
ALSO, due to chipset limitations, at the moment, only a maximum of 3 groupings can occur.
Thank you for the update. Is this published somewhere so it is clear what modes do what when using 4500c?
I may rethink my setup then in regard to using 4500c in an omni setup now knowing these limitations.
So in 2x 4x4 mode when using 2x2 sectors to make up the 4x4 antenna do they have to cover the exact same azimuth or can they be offset and if so by how much to still be considered a 4x4 for beamforming/or MU-MIMO to work?
No. Mostly because I think that it’s a moving target and as the firmware matures, additional features are being added.
There have been other posts in this forum with operators discussing different MU-MIMO antenna strategies. I’d suggest doing some searching.