Thanks for the tips, Matt.

On Tue, Oct 4, 2016 at 5:36 PM, Matthew Carpenter <
[email protected]> wrote:

>
> On the UE’s dropping from 50-60Mbit to 2Mbit, what I have seen is its an
> upload issue.  The Upload QAM is SO low that it cannot perform well on the
> download.
> A reboot brings it around and the speeds go back to normal.
> What I have been told is the UE/eNB will lower the QAM rate due to
> interference or other issues and it will lower the QAM faster then its
> rises it back up.
>

We haven't been able to prove interference in these matters. Low upload QAM
could cause this, I agree. However, particularly with the new 8000s, I'm
getting consistently good upload speeds and erratic download speeds, even
taking sector loading in to account. (There's 50-60 megs of free capacity
on most sectors during the mornings.)

The iperf implementation doesn't appear to work on the 8000 (and it only
marginally worked on the 7000). Will see if I can get it running via ssh.


>
> What are the “other issues” you ask?  On the CPE7000 it was bad boards
> that would cause high bit rate errors inside the radio causing it to think
> it has interference.
> Replacing the UE would fix it.  I also think the old crappy POE injectors
> causes issues as well.. cannot prove it..
>

We have done some UE swaps (no change), haven't done POE swaps.


>
> I have a UE that did this yesterday, I will watch it and if it continues
> to lower is QAM over the period of a day or so then we will replace the UE.
>  (CPE7000)
>

How are you watching upload QAM?

It seems to be a rolling target; doesn't usually strike the same customer
more than a day or two in a row, which is why we haven't defaulted to just
replacing hardware. (I am reluctant to do that without clearly
understanding what is going on.) There are customers who are consistently
more inconsistent, if you know what I mean.

We lost a customer a third of mile from the tower due to this, even after a
full hardware replacement.


>
> Also watching the tlsyslog on the EPC will give you all sorts of good
> information as to what a UE is doing, or not doing correctly.  I highly
> suggest you watch it for bad UE’s.
>

There needs to be a practical way to do this in realtime. Standard syslog,
for example.

It seems that some of our issues are concentrated on sectors where there
are more than about 10 UEs attached. Whether this is due to higher L3
density, higher RF usage, or simply UE density, is still to be proven.

-- 
Jeremy Austin

(907) 895-2311
(907) 803-5422
[email protected]

Heritage NetWorks
Whitestone Power & Communications
Vertical Broadband, LLC

Schedule a meeting: http://doodle.com/jermudgeon
_______________________________________________
Telrad mailing list
[email protected]
http://lists.wispa.org/mailman/listinfo/telrad

Reply via email to