We have a sector that is starting to reach capacity. Is it frowned upon to overlap an existing sector's coverage with another access point (likely using RFE horns) and then just manually load balance customers across the two? If so, how big of a guard band should there be between the two overlapping sectors?
We would use multiple horns, but we need to retain MUMIMO as much as we can, which isn't possible using the 30* horns.
Josh,
We will soon be releasing a 60 degree horn that is optimized for MU-MIMO operation. Below is the part number and description. To help today can you use freq diversity say if AP one is in 5.8ghz can you use 5.1 etc? This depends on your local regulatory but an example. If you so this you'll want to use the GPS to sync both AP's.
| C050900D025A | ePMP Dual Horn MU-MIMO Antenna, 5 GHz, 60 degree |
Ok - so the separation needs to be completely different bands? 5.1/5.8/etc?
Do we need to enable "colocation" mode and choose a front/back sector? Isn't that just for frequency re-use?
Hi Josh,
The optimal separation is a channel size so it should be 20MHz for 20MHz channel bandwidth. The front and back settings are usually required for two sectors working on the same frequency.
Thanks,
Dmitry
@JoshB wrote:Ok - so the separation needs to be completely different bands? 5.1/5.8/etc?
No. Assuming you're using sync, with ePMP you can use as little as a 5MHz guard band between channels. Horns also have excellent F/B ratios and tight beam patterns with no side-lobes, so you can sometimes get away without using any guard band, depending on physical spacing and coverage overlap.
Do we need to enable "colocation" mode and choose a front/back sector? Isn't that just for frequency re-use?
No, you only use colocation mode when sync'ing with PMP100/PMP450 radios. No, you only choose front/back sector modes when you're using back to back frequency reuse on two radios.
If you are using overlapping AP's, even with GPS sync enabled, you will need to use 20Mhz guard bands as Dmitry advised. The reason is if there is a complete overlap of the APs, then every SM connected to either AP, will see -both- APs at the same RSSI, including the adjacent channel power. This adjacent channel power will reduce MCS rates and overall throughput.
I doctored an online photo to hopefully give a visual of why the 20Mhz guard band is needed for overlapping APs. It uses 2.4ghz, but the you can subsitute 5ghz freqs as well. Same concept, just different bands/freqs. Sectors that do not overlap can use 5mhz guard bands as Eric stated.
Edit: The 20Mhz guard channel below is actually a 30Mhz guard channel due to there already being a 5mhz guard channel between 1 and 6 and another 5mhz guard channel between 6 and 11. Hopefully you still get the gist without this being too confusing.
This is false. All antennas have side lobes.
For more than a decade, RF Elements falsely claimed that their antennas magically had “no side lobes.” They supported this self-serving, misleading marketing by hand-drawing radiation patterns on their datasheets and providing low-detail patterns for only one polarization and frequency point.
They also told WISPs that they could save time and money and didn’t need to implement other isolation techniques—such as physical isolation, GPS sync, or active filtering—when co-locating antennas, because their antennas supposedly had “no side lobes.”
This caused many WISPs to deploy densely co-located towers with radios that were insufficiently isolated, resulting in colocated radio interference, poor performance, and degraded service to their subscribers, ultimately hurting their businesses and damaging the entire industry’s reputation.
Here is one documented example of this issue where a trusting WISP in Italy colocates many MikroTik radios with RF Elements horn antennas, believing that he does not need GPS sync because the horn antennas won’t interfere with each other: https://www.youtube.com/watch?v=frPdl_GlB1M&t=23m20s
RF Elements lists this as a testimonial on their website. There are countless other examples.
As a consequence of not disclosing their sidelobes, WISPs can not determine the actual locations of side lobes and nulls, can not place co-located antennas in those nulls, and can not accurately map coverages or create effective plans for frequency reuse.
This misleading marketing also started a race to the bottom: competitors who honestly and accurately represent their side lobes lose sales to RF Elements.
On their ePMP 3000 Dual Horn MU-MIMO Antenna, Cambium Networks provides computer generated radiation patterns with 45 dB of detail at 3 frequency points, for both polarizations, honestly representing the antennas sidelobes. This honest representation undoubtably causes Cambium to lose sales because RF Elements had persuaded much of the WISP industry that side lobes are problematic.
Getting back to the original topic, I wrote an article about overlapping sector coverages and how it is best to offset the coverage patterns by half the beam width to improve average signal strength. You might find this helpful:
