I have a sector (RBMetal +Panel 23 db) with 16 clients connected on my BTS.
One of the customers connected with a -73/-69 dBm to 8km (groove52HPn + 29dB grid) has a download bandwidth of about 3-5Mbps (BWTest UDP).
Another customer connected with a -53/-52 dBm get to about 30-32Mbps (BWTest UDP)
When I run the test from both: the first customer reaches 2-3Mbps while the second reaches only to 5-6Mbps
Technical details:
Protocol NV2 (best performance with TDMA-Period-Size=8 and Cell-Radius=26)
Frequency is the best
Tried with ROS 5.24 and 6.7 (no difference)
Why this behavior? as it seems anomalous that a customer who traffics, degrade the performance of the whole cell
A CPE with low modulation needs a lot of airtime to transmit it’s data.
So the sector capacity goes down with every transmitting low-performer.
You’ve to increase signal of the bad cpe or limit it’s bandwidth down.
We install additional high gain sectors for weak customers to increase
their signal and free the main sector. Use e.g. SXT-HG as additional sector.
We use 19dbi Mars Antennas with RB411AH so far.
administrator controlled media access policy
synchronization between Nv2 APs
some other features…
The problem you’ve described is a real world problem and there is no explanation or guide how to solve it.
There is no explanation how that problem even exist in TDMA based protocol - by definition of TDMA it shouldn’t exist at all.
One of the most important features of TDMA is not to have this kind of problem!
Mikrotik guys, this is a serious issue, please allow us to change time-allocation between clients or
at least to allocate/force each of them equal percent of time t[%]=100/n (n is a number of clients connected to AP)
Regards,
M.
Using 23db sector and 8kms 29db grid signal = -73/-69 dBm???
Using 16db sector and 23db panel my signal = 60-65dbm -
So unless the XR5’s give out a lot more transmit power than both RBMetal + groove52HPn, or the gain of either 23 or 29 antennas is not that when using the “best” frequency - or both ?
We like the concept we see with another gear called time-slots. We can assign fixed timeslots to cpes. The remaining timeslots are distributed on demand. Additionaly we can assign max bandwidth/cpe. A weak CPE gets less bandwidth/timeslot. Of course the aggregated bandwidth is not higher but we can do more finetuning.
