- This topic has 29 replies, 1 voice, and was last updated 9 years, 5 months ago by BOUBA.
6th April 2010 at 20:24 #62007lilyGuest
Good evening, dear all!
I would be very grateful for some explanation, if you please, about situation described below…
It has already been mentioned here but I can’t add a post to that thread (even the smallest) so I’m starting a new one. I’m facing the problem with RxLevel = -125dBm and high C/I = 21 right now. And the most unpleasant fact is that it’s not unique here – I found 7 or 8 places with such symptoms in one logfile (drivetest of a medium size city).Weird situation…
– at first, I doubt very much – are RxLevel measurements correct? How can we have good C/I with such a low signal? As I remember, -121 dBm is a level of typical *thermal* noise for 200 kHz channel. Correct me, please, if I’m mistaken. Moreover, the voice quality is really good (we measured MOS so had plenty of audio samples and could estimate them by our own ears). So, from these two values: RxLevel and C/I the first one seems to be incorrect and the second – on the contrary – very credible.
– not only TEMS behaves itself peculiar in these cells – in one place we also made experiments with Motorola L7 (with Netmonitor) and Nokia N95(with FieldTest). They all showed the same -110dBm and it also was accompanied by an excellent speech quality. I can’t say exactly which MAIO and TS we had when we use Motorola and Nokia, but from TEMS logfile – it was always MAIO = 0 and HSN = 0. All cells works in BBH (not RH).
– trx replacing and re-creating the cell in OMC-R are useless
– watching those TRXs (MAIO = 0) in NPO (RxQuality – Rxlevel DL matrix, I see that Rxlevel DL distribution is abnormal – 15 – 25% of samples have RxLevel < -104dBm, while other trx*s in the same cell has absolutely normal distribution of RxLevel, and the situation is stable from day to day.. So, lots of MSs (or all MSs) measures RxLevel on MAIO = 0 incorrectly there. What could be a reason of such mass inadequate MSs’s behavior? Of course, tomorrow I'll change HSNs from 0 to other values as it was recommended, but really I don't believe it can help. I just can't see logical interconnection between the order of frequency change (HSN) and the process of measuring signal strength by MS… Thank you very much in advance Best regards, Lily8th April 2010 at 05:15 #62008AliAsgherGuest
I hope that you are talking about RxLevel(Sub) values in your post. Can you kindly send some of the DT logs to the below ID. I am very much interested in seeing this problem.
Regards,8th April 2010 at 08:38 #62009BijoyGuest
I’m also quite curious as why should this happen.Keep posting for any sort of findings.
Bijoy8th April 2010 at 13:29 #62010ALex_RGuest
i faced the same problem but with no hopping so changing HSN will not help, sorry Lily
by the way which vender you work with ?
i faced this problem with Huawei BTS30129th April 2010 at 08:55 #62011lilyGuest
Good day, AliAsgher, Bijoy and Alex_R!
I’ve just sent you some logs I have at hand now. They were made with TEMS 7 and TEMS 8.
Two big ones – from road drivests and the smallest logs – it was investigation of one site with such a problem.
It’s excellent that you’ve joined this discussion 🙂
Sory, but I have no experience with Huawei at all. I work with Alcatel (B10 now).
But the situation you’ve described is strange… Does it happen on certain TS (or TSs)? Or on all TSs on certain TRX?
Some interesting moments from our observation: we saw the problem only in case of using BBH (i.e. it’s always BCCH-trx because MAIO = 0 ), never – in RH. And it’s always subchannel 0. I didn’t see it (yet maybe) if we work in FR.
Correct it, please, if you have additional info.
But the most important thing – in our case all 7 experimental cells were treated by changing HSN from 0 to 0..63 values…
I hope all together we’ll find the explanation 🙂
Best regards, Lily9th April 2010 at 11:08 #62012AliAsgherGuest
Just downloaded the files and am checking it right now. One quick observation.
DTX on the cell is OFF. And Rxlev (Full) is giving valid values whereas Rxlev(Sub) is fixed at -126.
I think it may be coded in TEMS that if DTX is off then it does not calculate Rxlev(Sub) ??? Not sure though. Will update soon.
For those who want to see a quick picture, please visit
Regards,9th April 2010 at 11:20 #62013AliAsgherGuest
Well I think the reason is the same as I stated earlier. On cells with DTX disabled, RXLEV(Sub) is constant on -126, but on other cells it is OK.9th April 2010 at 11:34 #62014PixGuest
So let’s recap :
MS measures a very low DL RXLEV on the TCH timeslot :
only in BBH
only in HR
only on subchannel 0 (subchannel 1 is fine)
only MAIO = 0 (so that’s the 1st TRX)
only HSN = 0 (cyclic)
In those cells, the number of hopping frequencies = ?
Since it is on BCCH-TRX, then the DL DTX does not apply. Therefore, RXLEV FULL should always show valid measures.
There could be some issue in RXLEV SUB, due to cyclic usage of frequencies, but since BTS will always use RXLEV FULL (DTX is never on) then that probably is not worth investigating.
You didn’t answer Ali’s question : problem is in RXLEV FULL or RXLEV SUB ?
Lily, you are the brain twister 🙂
Cheers 🙂9th April 2010 at 11:35 #62015pixGuest
hahaha, just read your replies, ali. Seems like we are both locking on the DTX 🙂9th April 2010 at 11:41 #62016m121Guest
We are facing similar problem in our network (Vendor Siemens)but with little bit different simptoms and we coudn’t “catch” it on TEMS, only customer traces So, Lily, can I ask you some questions?
Does this happen in only 900Mhz cells or in 1800Mhz cells? Does BS power reduction goes up to max values? Does this situation happens after HO to this cell from another or immediatly after call setup?
I am asking this because in our network some calls after HO to 1800Mhz cells stops sending valid (or at all) SUB measurements (sends only first one valid) and BSC then uses this one for all measurements (from BSC point of view MS sends all MesReports with constant level and quality ) and BS power goes immediatly down to max values and no HO is possible, because BSC thinks that DL level and quality are great and that results in droped call.
If simptoms of your problem are similar or somebody else have any ideas how to solve problem in our network, please, don’t be shy and give some advice 🙂
PS.At this moment I am on vacation so HSN I will try changing later, when I will be at work and if that helps I will share ir with you.9th April 2010 at 11:47 #62017AliAsgherGuest
Actually your case is simply a trace where MS got lost. Due to this as you are not receiving MRs, on CTR you will continue to see TA=63, Rxlev=-110 etc etc. After RLINKT the call will be dropped too.
We both locked to DTX right, but you did it even without analyzing the DT logs 🙂9th April 2010 at 13:14 #62018m121Guest
Thanks for replay AliAsgher, but this is not the case: TA is normal and changes if MS is moving. With UL measurements everything is ok. If MS doesn’t move, than call can go one as long as suscriber wants (we have some calls that are active for half an hour with normal call clearing).
Only thing that is different for these calls are that DL measurements are the same with every report (for example if MS access the cell with -60 dBm level and quality 0, these values stays for all duration of call even of UL drops to -110) and dropped calls happens only to moving MS that drives out of coverage area and there are no HO possible, because BSC thinks that MS have good DL coverage and there are no cells that are good enougth to fulfill DL requements and only in that case radio link failure happens.9th April 2010 at 14:08 #62019lilyGuest
Hello, dear AliAsgher, Pix and m121!
Oh, so much interesting happened here during last hours 🙂
Actually, there is nothing to add …
Sorry, I’ve just missed – yes, of course, it’s RxLevSub… but it’s obvious of course now
The question about DTX … super – the universal mind should exist – my colleague and me were discussing almost the same on our way back to office…
Dear Pix, as usual, you’re the best:
– you put everything in order
– summed up
and get the answer 🙂
Dear Ali, it seems to be the end of active investigation 🙂
HSN = 0 is universal evil…
Let’s change it and then be happy 🙂 As Pix said it isn’t worth further investigating 🙂
Dear m121, answers to your questions:
– either in 900 or 1800 bands
– either after Call Setup or HO (always if you’re allocated on subch 0, MAIO=0 and HSN = 0)
So these problems aren’t similar. Lots of my colleagues work with Siemens BSS, I’ll ask them about your problem, maybe they could help.
Pix, it wasn’t so funny at first…
We were *clever* enough to implement hopping in one region with HSN=0 everywhere, so after the next comparative drivetest (we and other operators) we got there the worst average RxLevSub amongst others … Our management had a tantrum, and so on…
But the best was my boss, he said “Only women can afford themselves light-headed behavior, equipment must work correctly. The problem should be solved before the next drivetest”. But he obviously forgot that Alcatel HW has female temper 🙂
Best wishes from my colleagues from Kiev (your trainees) Inspired by you, they are going to implement lots of features – NC2, DTM 🙂
Lily10th April 2010 at 19:14 #62020PixGuest
Hi Lily :))
There is still something that doesn’t click :
RxLev Sub is bad, ok, but the BTS and BSC should not care about the “sub” value, because DTX is OFF.
It is only when DTX is On for the current measurement that the BTS and BSC take the Sub measure within their computation.
Apart from the rxlev sub in the drive tests, did you see :
a low RMS DL_RXLEV_AVG on TRX1 ?
a high amount of DL HO in the cell ?
I wonder why you are seeing the RxLevSub in the RMS, rather than the RxLev Full… Is it normal ? Is it a bug ? I’ll check it out.
I’m glad your colleagues will implement the NC2, but I’m less enthousiastic about DTM 🙂
I can’t see the traces unfortunately.. i don’t have TEMS installed, I don’t even have the software available. But the good thing is that my eyes are perpetually burnt with the screen layout of the TEMS, thanks to spending so many hours doing drivetests around S-E Asia ;))
pix11th April 2010 at 04:25 #62021duocdtvtGuest
can u help me? i need Erlang table for design