ePMP 4525 vs UI Powerbeam 5AC Gen 2 (bench tests)

[terintamel]

While waiting for my ePMP 4500 Aps to arrive I did some lab tests with a simple PtP link between a pair of 4525 SMs and Powerbeam 5AC Gen 2 to see how performance differs and the affect of wifi interference on them.

The 4525 SMs are configured using WLR mode while the Powerbeams are set in Flexible framing mode. Both are 20mhz channels and set to around the same RSSI.

One end of the link is a Mtik hap AC2 on the SM side and a Mtik RB5009 on the AP side.

Both the 4525 and PBE 5AC are set in NAT mode with the AP in Bridge mode.

Around 3ft from the SM I have the hap AC2 set to 5180mhz @ 20mhz. Around 30ft way is a Unifi AC Pro AP with it set to 5180 @ 40mhz. I have my cell phone connected to the Unifi AP and sitting right next to the SM. To test I would run an internet speed test on my cellphone to create interference.

I will put photos below to show but to summarizie the Powerbeam 5AC link would work well but only if I had ReSE feature enabled. This feature has the CPE say that the channel is busy which causes anything running the same channel in wifi to backoff. Basically it is a wifi jammer. However since to my knowledge Cambium does not have this feature I tested with it off to be equal.

With ReSE off the Powerbeam link tanked with horrible latency and unusable performance. The 4525 link took much less of a hit and kept on running with a some acceptable performance drop when my cellphone was doing a speed test.

Now this is a somewhat unfair test as the Powerbeam 5AC Gen 2 is built on a wifi chipset that is AC Wave 1 along with being developed in ~2016 while the 4525 is the latest chips. I will probably later do a test vs LTU using a pair of LTU-LRs, but I started with the Powerbeams as LTU seems to handle interference even worse then AirmaxAC from what I have heard.

[terintamel]

Here is the UI link doing a Mtik bandwidth test through it on a clean channel

Here is the same UI link doing the same test but this time on the same wifi channel as the Unifi AP and hap AC2 while my cellphone is doing a speedtest connected to the Unifi.

[terintamel]

Here is the 4525 SM link doing a Mtik bandwidth test on a clear channel.

[terintamel]

Here is the 4525 link with the SM set to MCS 9 (256QAM) to make it more equal to the capabilities of the Powerbeams.

With active interference from my cellphone.

[terintamel]

Here is the 4525 without active interference from my cellphone but on the same channel as the wifi routers nearby.

[Eric_Ozrelic]

Interesting tests! I have a few suggestions…

1. Try setting direction to send and receive
2. Try setting connection count to 10
3. Set to random data
4. Don’t use NAT on the SM… just use bridge mode so the radios aren’t touching every packet
5. On the 4500 setup use ePTP mode
6. Monitor the CPU usage on the Mikrotik’s to make sure you’re not overloading them.

[terintamel]

1. Try setting direction to send and receive - I did that but did not record the results. Was pretty much similar in regards to interference, but the aggregate capacity of the 4525 link was higher than just doing a DL test.
2. Try setting connection count to 10 - I tried default, 6, and 20 connections and saw little performance difference.
3. Set to random data - I have never seen much difference with this setting.
4. Don’t use NAT on the SM… just use bridge mode so the radios aren’t touching every packet - I use NAT because that is how my UI CPEs are configured and I plan to replicate that should I switch my network from UI to ePMP 4000. Not a fan of CPEs or SMs in bridge mode. One reason that Mimosa is not high on my list of a possible platform switch.
5. On the 4500 setup use ePTP mode - I am sure I could probably get more performance but I was doing a comparative tests and UI CPEs were not configured in PtP mode.
6. Monitor the CPU usage on the Mikrotik’s to make sure you’re not overloading them. - Neither the RB5009 or the hap AC2 anywhere close to even 50% CPU usage.