- This topic has 76 replies, 1 voice, and was last updated 8 years, 2 months ago by IAN.
1st December 2009 at 22:11 #598582112Guest
In one cluster of our Network AMR was enabled and after enabling AMR DTX for both AMR FR and HR was enabled as soon as DTX was enabled Interference HOs increase for all cells .
Reason provided by Vendor is that after enabling DTX RXQUAL measured by Mobile Station is over only 12 TDMA frames(SID and SAACH) resulting in decreases accuracy as desacribed in 05.08 series of 3GPP Recommendation.
Vendor is Alcatel, so question is for Alcatel experts if they have encountered anything like this in their networks?1st December 2009 at 23:04 #59859PixGuest
It is true that DTX is reducing the accuracy of the measurements, but not as badly as you described.
If DTX had such an impact, you would see an increase in interference HO even without AMR.
Indeed, for non AMR calls, I guess the DTX is enabled in your network, isn’t it ? And yet, you don’t see high interference HO for non AMR calls 🙂
I’ll try to see if there is any history about such misbehaviour, I’ll let you know. Until then, you should probably avoid DTX for AMR calls. THe problem appears with UL DTX or DL DTX ?
Pix2nd December 2009 at 04:42 #598602112Guest
Degradation is in both Downlink Interference and Quality HOs although in percentage stats it is severe for DL Interference HOs which increased 5 Times .
You have made good point that DTX was already working in tne network for normal HR and FR and DL INTF HOs were around 1% of overall HOs.3rd December 2009 at 08:55 #59861PixGuest
It seems that the ETSI (which is in the 3GPP to define the standard of GSM) has made a mistake : the algorithm used in DTX is not compatible with the AMR. The voice quality is seriously degraded when AMR + DTX are used together (but not only in alcatel ! it is an ETSI mistake)
Please confirm that with Alcatel first ! Confirm that the problem still exists in your release, and if it can be fixed with a patch.
Haven’t you done a drive-test to hear it by yourself ? I’m sure you would really hear the proble. It is real and not just due to bad measurements…
Let me know 🙂
Pix3rd December 2009 at 12:57 #59862PixGuest
SlimOTT : here !!5th December 2009 at 07:41 #598632112Guest
Incompatible in what sense??
Same DTX was working fine with normal HR and FR as you pointed out earlier.
On the other hand, we have our motorola network which is serving a big city having DTX enabled and AMR working and not a sinle BSCs has DL INTF HOs more than 0.5% while these ALcatel BSCs after implementation of AMR and DTX for AMR are having DL INTF HOs about 5% for each BSC.
SO this degrdation is not there in Motorola Network .
And is it possible through any parametric settings that less weightage is given to measurements while DTX is being used in DL??5th December 2009 at 08:16 #59864PixGuest
The question is : how did Motorola implemented the DTX ? Did they follow the ETSI recommendations ?
I don’t know all the details obviously, but I gave you your answer. If you want to investigate, have a look on the ETSI website, see how they implemented the VAD2 with the AMR codecs, start questioning them. Perhaps Motorola is not using VAD2 ? I don’t know.
Now it is up to you to work on your vendor 🙂 I can’t share all details on this forum and I wouldn’t want to start assuming things on Motorola or Alcatel systems publicly…
Pix22nd January 2010 at 05:27 #598652112Guest
Here are few things we have done to investigate this issue,
DTX was disabled for AMR FR and HR and Interference HOs restored to their orignal value and as soon as it was enabled again INTF HOs increased.
According to ALU this degrdation is simply due to inaccurate measurement reporting by MS because of DTX and nothing else.
ALU is using VAD2 for DTX with AMR.
Vendor is suggesting to use W_QUAL_HO parameter to give more weightage to non DTX measurements in averaging process so that averages are less impacted by inaccurate DTX measurements.
What do you say???????
Thanks.22nd January 2010 at 11:56 #59866PixGuest
That sounds like a good workaround ! You should start experimenting, and please share the results with us. I am very interested about this (no time to study it though…)
Thanks for your feedback,
-pix25th January 2010 at 10:06 #59867sergioGuest
I’ll be grateful if you explain please what parameter W_QUAL_HO means exactly.
i can’t understand how it work throught docs i have
thanx in advance
br26th January 2010 at 06:07 #59868PixGuest
It is the weight given to a measurement done during a multiframe (480ms) where DTX was not used. It is FULL.
If DTX was applied, the measurement is a “SUB” measurement, taking into account only measuring of the SACCH bursts (there are no TCH bursts, they are all empty). Here the weight is always 1.
For example : W_QUAL_HO = 3.
samples 1 and 2 are “FULL”, and 3 and 4 are “SUB”.
AV_RXQUAL = [3*RXQUAL(1)+ 3*RXQUAL(2) + RXQUAL(3)+RXQUAL(4)] / 8
Here you see, you must define the A QUAL HO (averagin window) = 8. Because 1 and 2 weights 3 times more, the actual number of samples in the average is only 4.
So when you increase the Weight, be sure that you increase the Averaging Window too. Multiply it by “0.75*weight”, that should be not too bad 🙂27th January 2010 at 10:20 #59869sergioGuest
Thanks alot for your just-in-time and detailed answer. At the end it’s so simple!
We use the next parameters set now:
A_LEV_HO = 6
W_LEV_HO = 1
A_QUAL_HO = 4
W_QUAL_HO = 1
A_LEV_PC = 4
W_LEV_PC = 1
A_QUAL_PC = 3
W_QUAL_PC = 1
I want to increase W_xx_xx value – twice. What is your opinion (from experience)?
Many thanx dear Pix
BR, Sergio27th January 2010 at 17:40 #59870PixGuest
You just need to increase the W_QUAL_HO and W_QUAL_PC, not the others.
I have no experience with the weight, but I would start with something like this :
W_QUAL_HO = 3
W_QUAL_PC = 3
A_QUAL_HO = 8
A_QUAL_PC = 6
And check for the results. Can’t you just :
Downlink_DTX_enable_AMR_FR = DISABLE
Downlink_DTX_enable_AMR_HR = DISABLE
DTX_FR_INDICATOR_AMR = DISABLE
DTX_HR_INDICATOR_AMR = DISABLE
DTX is still used for non-AMR calls, but is disabled for AMR calls.
That looks SO MUCH simpler than changing the weight… And because the DTX is disabled, to balance the negative impact on overall interferences, I would push the DL and UL PC by few dB’s:
L_RXLEV_DL_P : apply -2dB
U_RXLEV_DL_P : apply -2dB
pix15th February 2010 at 22:29 #59871sergioGuest
26 I tried to post this message once more (26th).
Lots of thanks i applied parameters you recommended on 2 BSC. Results are following: decreasing number of HO and decreasing % of TCH blocking. It’s very good, I’m very grateful to you.
I would say that TCH blocking “redistributes” between cells. Before that I had some heavily congested cell (due to fails for incoming HOs MC541, 551, 561). For those cells we could see strong misbalance between HOs. Outgoing HOs < incoming ones. Now they are almost equal. For surrounding cells THC blocking became a little bit higher, but average value decreased as I said from 2,01% to 1,59%. Thanx, dear Pix! For surrounding cells THC blocking became a little bit higher, but average value decreased as I said from 2,01% to 1,59%.16th February 2010 at 09:49 #59872PixGuest
Thanks for the feedback !
What did you do finally, change the weight or disable DTX for AMR calls ?
As i said, I would strongly recommend the 2nd solution.
I am surprised those parameters are affecting your congestion. I thought it was just a Quality Issue ?
The average congestion of your network is rather high, but I’m glad it decreased a little. What is your RTCH_Duration_Average ?
To decrease congestion, have you activated EN_FDR and EN_DR in all your cells ? You could tune the FDR parameters, they offer a very effective way to distribute traffic with neighbour cells. However, you told me the traffic is already shared…
So now you could increase the HR usage…