Poor throughput experienced on Uplink for 2km LoS ePMP link

A pair of integrated ePMPs was deployed for a wifi hotspot for a marathon event recently. I have attached a photo of the subscriber end of the link with the integrated ePMP. As you can see this has a steep elevated angle back to the high site.

This was for a 20MHz channel, with 29dB SNR over a 2.7km LoS path, RSSI -69dBm, MCS 12-14 in both directions, ver 2.3.4, WPA2, 23dBm power each end.

My planning in LINKplanner agreed very well with actual results, to within 1dB, being a line of sight path.  Also the MCS and max downlink throughput of 34Mb/s agreed with LINKPLanner’s 50:50 estimate.  Also when changing to Flexible, I observed a downlink of 53Mb/s which agreed with LINKPLanner’s max throughput of 50Mb/s.  Also I tried 30:70 where I observed 17Mb/s down close to LINKPlanner’s 18Mb/s predicted downlink throughput.  The WiFi APs were not on during these tests.


The actual downlink MCS was around 12-13-14, and uplink similar around 12-13.  Both SNR were 29dB.


I’m just unsure of why the uplink throughput was not balanced as expected, when the modulation and SNR were balanced.  We used channel 5800MHz. We did not get a chance to change to an alternative channel, however the area is rural with no known interference. I was not able to test alternative channels to rule out the high site potentially having interference issues. If there was interference at the high site end of the link I would have expected the MCS to be much lower than 13.


When we tested with external speedtest.net (measured at the router connected to the ePMP - with the WiFI APs off), we observed the max circuit speed of 20Mb/s downlink, but the upload test was in the order of 6Mb/s instead of 20Mb/s.  This was unexpected for the uplink as we were connected to a 20Mb/s circuit (both up and download).


The customer was pleased with the outcome, as their streaming was only 550kb/s and typically only expected around 1.5Mb/s.  An we only had 40-50 public Wifi users served by two APs throttled back to 0.5Mb/s, so traffic demands were not high.


Please let me know if more info is required.


1 Like

Sounds like a great deployment, we are checking on the throughput.

We were also surprised and not satisfied with the Uplink performance in this case. We have consistently seen higher in many deployments. Our development team talked with Jason about the link performance, and this is what was found:

  • The STA end was in a quiet park away from interference sources.
  • At the AP end up on the high site tower, it cannot ruled out there being other wireless equipment, as it is a broadcast site. Snce the uplink was poor, any interference likely would have been at the AP end and not the STA.

  • eDetect was not run at installation to identify any interferers. Find out how to use eDetect in a video HERE.

In this case, there were many things that were examples of best practices. Tips for network operators installing a temporary network on short notice:

  • Understand the connectivity need, the number of users and the throughputs they need
  • Consider using different SSIDs for different groups. Here, there was one for media, and a separate one for public.
  • Use LINKPlanner to plan the links
  • Use eDetect when installing the AP to check for interference, and select the cleanest spectrum
  • Test the network prior to the event, with as much advance time as you can
  • Get feedback from users during and after the event