We are evaluating ePMP 1000 as a potential to replace UBNT AUs and SUs (nanobridges/nanobeams) becasue -
a. UBNT AUs cant handle more than 30 Mbps thruput , and no more than 40 SUs max
b. Cambium AUs it appers , have a higher capacity and can handle more Subscribers
We have started to test Cambium AUs and integrated subscribers on 4 sites, and getting mixed results.
On a couple of sites we are setting quite alot of packet loss and retransmits (Monitor>Performance) where the integrated subscribers are less than 6 kms from the AU. The envonment is quite noisy and we have selected the best available channel. The SNRS and RSSI levels seems fine. However when the Uplink RSSI goes beyond -70 , the unit drops off.
Nanobeam 300's at these distance have no problems at all.
Would appreciate if someone could give some guidance from their experience so we can get reduce the packet loss and have a solid AU-SS association..
The SNR is just a measure of the signal over the noise floor. It doesnt give you indication of the interference (CINR). Have you done a spectrum analysis at the AP. I say AP because the DL RSSIs are good but the UL RSSIs are lower. It may be because there is interference at the AP. Things you can do:
1. Run eDetect on the AP under Tools->eDetect to see if there any 802.11 interferers on the channel at the AP
2. Run Spectrum Analyzer to see if there are other interferers at the AP on that channel
3. Increase the UL RSSI. On the AP change, Configuration->Radio->Power Control->Subscriber Module Target Receive Level to something higher than -60.
Thanks for the information. The site is noisy. I have attached a screenshot from the ACS. We have also switched to a 10 Mhz channel.
The UBNT equipment sees to be quite tolerant of noise, so we were hoping the Cambium equipment would be also.
CAn you give us an idea of the performance we can expect on a 10 Mhz channel vs 20 Mhz - numbers of subscribers, throughputs etc when we find the best channels.
Also , would additional shielding on the AU and Antenna help ?
That ACS chart shows shows quite some noise across the band.
I would still recommend you to run eDetect (Tools->eDetect) to determine where the interference is coming from and at what level. Same with a spectrum analysis (Tools->Spectrum Analyzer). This will give you a good idea of how high the interference is.
A lot of work has gone into ePMP to make it resilient to noise and has performed very well under harsh interference in many deployments around the world. From your screenshots, I see that the bottom 3 SMs are the ones having trouble on the UL and, as expected, all 3 of them have an UL RSSI of low 60s and one with -70. And the fact that these SMs drop off if the RSSI goes below -70 tells me that the interference is quite a bit above -70, which gives us a negative C/I. That is a tough RF environment indeed.
The number of subscribers supported for each channel bandwidth is as follows:
20 MHz / 40 MHz: 120 SMs
10 MHz: 60 SMs
5 MHz: 30 SMs
The performance of a 10 MHz channel is roughly half of a 20 MHz channel. For exact values and more performance characterics of ePMP, there are two free tools available for you:
2. LINKPlanner program - A more comprehensive software application to help plan and predict your ePMP deployment
ePMP comes with built in shielding and does not need additional shielding such as an RF amour. Both the radio and Cambium sector antenna have good adjacent channel rejection characteristics and you should not have to use third party accessories. However, it is up to you if you would like to give some accessories a try.
Thanks for your reply. Plenty of information to work with there.
We'll try your suggestions and revert back.
On two other sites where we have deployed Cambium AUs, with better signals and less noise, we are getting great results with the Cambium integrated units. Real speeds over 70 Mbps over 5 kms surprised us. We are very keen to see what happens as we load of a sector to around the 60 subscriber mark in terms of stability and throughput.
The UBNT Rockets can only handle about 35 customers, and process no more than 30 Mbits which has forced us to look elsewhere.
a quick extra to deal with noise, check your sector tilt and go as far as you can without taking more than a few points off of your -70 sub. you may see a reasonble improvement in CSNR and MCS even though your SM might drop a few points due to some noise insulatoin. the ubnt sectors are pre-tilted and cambiums are not so without checking and planning tilt, you may run into extra noise.
we've changed around 100 rockets to epmp and have seem a genral improblem in link stability and performance.
also the force 110s are amazing. we use them a lot. you'll likely see that -70 sub get to -58 ;)
