LTE Dominance Optimization aside from Physical changes

[anita_2k]

hello friends,


would like to inquire if you experienced LTE Dominance issues in your past/current LTE project.
in UMTS, i recall tuning Soft handover Overhead to speed-up or delay the Handover between cells.
as well as (in NSN) we put on adjacencies EcN0 offset.


do we have such thing in LTE?


Fact:
in our current project, Central Business District, LTE sites are deployed contiguously.
LTE sites are very near to each other with a site-to-site distance of 100-150meters only (in CBD).


Issue:
we have 2 LTE sites which is not serving/dominating upon drivetest verification.
thus, sites/cells DL/UL throughput performance cannot be verified on street-level (but on BTS room only)
Areas are mixed of Huawei and NSN LTE Sites where dominance issue are experienced (Huawei more dominant)




Did so far:


modified the following parameters:
- A3 offset (a3Offset) from 3 to 5dB


Definition: Handover margin for better cell HO. Used in measurement
event type A3 where the event is triggered
when the neighbor cell becomes better than
the serving cell by the value of the A3 offset.


- Cell individual offset of own serving cells PCI (cellIndOffServ) from 0 to 6dB


Definition: Cell Individual Offset (CIO) of own served cell for
insertion into the CIO Cell List of EUTRA Serving
Carrier Measurement Object
Only if CIO is different from zero, will the eNB
consider it for Measurement Configuration.




Results: some sectors did served after modification of the said parameters but Huawei cells are still dominating.
i don't know if the method of drivetest did affect the results / lucky only.
i wonder if increasing the offset really is the answer or maybe some other parameters can be adjusted?
limited drivetest resources for trial and error though.



any comments or parameters suggestion/for mondification are very much appreciated.


thanks
br/Anita

[incoincov]

Hi Anita,

Did you check the settings of power related parameters?

[edinburgh]

For anita_2k' case, his LTE site might be RF sharing with 2G, so he can not change power setting.

Is that true, anita_2k??

In your project, Central Business District looks like very dense?? Like Singapore??? :lol

[anita_2k]

Yes, LTE is RF Sharing but it is with 3G.
Yes similar to Singapore. hehe..


For anita_2k' case, his LTE site might be RF sharing with 2G, so he can not change power setting.

Is that true, anita_2k??

In your project, Central Business District looks like very dense?? Like Singapore??? :lol

[Philippe]

Have you checked tilts and parameter settings?

[anita_2k]

tilts is one thing, i can recommend physical change repan/downtilt.
but approval would take long.
thus i am more keen on parameter changes if any. I guess will have to set more aggressive values.

yes, everything are aligned (on NSN LTE perspective).

on Huawei LTE sites (which is more dominant), is difficult to verify.

thanks
br/anita


Have you checked tilts and parameter settings?

[giegala]

Hi,

there was a very old e/// training course on 2G functionality/optimisation. If i recall it correctly, it started by saying that parameter settings do not fix cell planning errors. I believe it still the same for LTE.

If you keep UE on weaker cell for longer time (due to delayed HO from weak cell, or early HO to weak cell), you will impact User throughput, because of the same frequency band operation.

As you state that HW cells are overshooting, the main reasons can be 2:
1. not enough tilts
2. much higher power on Rs on HW when compared to NSN.

So did you check Rs power values on NSN and HW? Actually you do not need to check OSS to verify it. Rs power values are broadcasted in SIB2 on both vendors and compare. In Pdsch-config section check both "reference signal power" and "p-b" value.

[Philippe]

Do you have clear separated areas between Huawei and NSN? or do you have 2 layers in a same area? Or 2 different bands for Huawei and NSN?
Are Huawei and NSN macrocells? or is it a mix of macrocell from 1 vendor and femto/metrocell from the other vendor?
Have you checked the power settings of Huawei and NSN eNodeB? are they aligned?

[anita_2k]

Do you have clear separated areas between Huawei and NSN? or do you have 2 layers in a same area? Or 2 different bands for Huawei and NSN?
Are Huawei and NSN macrocells? or is it a mix of macrocell from 1 vendor and femto/metrocell from the other vendor?
Have you checked the power settings of Huawei and NSN eNodeB? are they aligned?

phillipe, both Huawei and NSN are using similar bands.
the area (CBD) is very dense.
site to site distance is ~100 to 120 meters only.

[anita_2k]

Hi,

there was a very old e/// training course on 2G functionality/optimisation. If i recall it correctly, it started by saying that parameter settings do not fix cell planning errors. I believe it still the same for LTE.

If you keep UE on weaker cell for longer time (due to delayed HO from weak cell, or early HO to weak cell), you will impact User throughput, because of the same frequency band operation.

As you state that HW cells are overshooting, the main reasons can be 2:
1. not enough tilts
2. much higher power on Rs on HW when compared to NSN.

So did you check Rs power values on NSN and HW? Actually you do not need to check OSS to verify it. Rs power values are broadcasted in SIB2 on both vendors and compare. In Pdsch-config section check both "reference signal power" and "p-b" value.

giegala, yes, i do not questioned the planning basics. i am sure as well that doing physical tuning will surely solve the dominance issue.
just checking if other way is possible.
sorry for late reply regarding your inquiry on rs power.

here are the details:
1. in Huawei, i can see on the logfiles 15 (dBm) as rs power (p-b = 1)
2. in NSN, i cannot see any rs power on logfiles as well as in parameter dictionary (i can only see pMax - Maximum Output Power). Which can be modified per cell as well

regards,
Anita

[anita_2k]

giegala, here is the power settings in NSN as well in ALL our current LTE (on-aired sites).
pMax on all LTE sites/cells = 43.0 dBm (20 W)

but i think this one is eNodeB power (not rs power itselft). it is too high and somewhat similar to 3G 20Watter sites (nodeB).

sites are all RF Shared.

regards,
Anita

[giegala]

Hello anita_2K,

your collected HW setting show that
(p-b = 1) -> Rs symbol power is by 3dB higher than normal symbol;
Single Rs symbol power is 15 (dBm)

I do not see what is bandwidth of carrier. Considering 20Mhz, the total output power from HW is:
15dBm -> 31.6 mW
31.6 W * 1200 /2 = 18973.7 mW =~20W as compared to NSN Pmax
I am dividing here by 2 because Rs symbol has 3dB gain compared to normal symbol.

Now there are 2 things:
1. NSN should also broadcast both p-b and Rs power, because it is needed for UE power control
2. by default NSN does not add 3dB gain to Rs symbol - this should be obvious p-b value in the logfile.

When collecting logfiles make sure, that you make several reselections inside NSN area, because if you are staying inside 1 cell SIBs do not change, and UE does not read them.

[anita_2k]

Hello anita_2K,

your collected HW setting show that
(p-b = 1) -> Rs symbol power is by 3dB higher than normal symbol;
Single Rs symbol power is 15 (dBm)

I do not see what is bandwidth of carrier. Considering 20Mhz, the total output power from HW is:
15dBm -> 31.6 mW
31.6 W * 1200 /2 = 18973.7 mW =~20W as compared to NSN Pmax
I am dividing here by 2 because Rs symbol has 3dB gain compared to normal symbol.

Now there are 2 things:
1. NSN should also broadcast both p-b and Rs power, because it is needed for UE power control
2. by default NSN does not add 3dB gain to Rs symbol - this should be obvious p-b value in the logfile.

When collecting logfiles make sure, that you make several reselections inside NSN area, because if you are staying inside 1 cell SIBs do not change, and UE does not read them.


hello giegala, thanks for your time and detailed explanation.
btw, BW = 10MHz (dl earfcn = 150), fyr.

i still do not have the logfiles now, maybe later i will post again.
i saw your comment last night and search for it in the dictionary.
somewhat, i saw some related parameter.

dlpcMimoComp (MIMO power compensation)
it is set to 0dB.
ranges are 0dB, 1.77dB, 3dB, 4.77dB and 6dB.

i think by modifying it from 0 to 3dB, i can equalize the area (mismatch area bet NSN & HW) what do you think?

thanks alot
anita

[giegala]

Hi,

i would suggest you to collect the logs first, before doing any changes. Then it would be possible to verify if you have found correct parameters and if they do any good. Besides dlpcMimoComp is not a correct parameter. There should be RS letters in the parameter name.

Considering HW power and 10 MHz bandwidth, that would make 10W per PA branch.

Good luck

[anita_2k]

hi giegala,
just want to mention what happened and what i saw:
i collected some logs
1. after modification of dlpcMimoComp (MIMO power compensation) from "0" to "6dB", i saw on referencesignalpower on logs (served by NSN site) = 18dBm
2. as you mentioned, this might not be the correct parameter for adjusting RS Power. I saw another parameter in the dictionary
a. dlPcfichBoost (Downlink PCFICH transmission power boost) - Transmission power of the Physical Control Format Indicator Channel is boosted by set value
b. dlPhichBoost (Downlink PHICH transmission power boost) - Transmission power of the Physical Hybrid ARQ Indicator Channel is boosted by set value
c. dlRsBoost (Downlink reference signals transmission power boost) - Transmission power of the downlink reference signals in a cell is boosted by set value compared to PDSCH

all parameters ranges are 0 to 6dB 90dB (0), 1.77dB (1), 3dB (2), 4.77dB (3), 6dB (4)

i would go on dlRsBoost... I will try to change this + offsets if can change the outcome.

will let this forum know as well.

thanks
anita


34505


Hi,

i would suggest you to collect the logs first, before doing any changes. Then it would be possible to verify if you have found correct parameters and if they do any good. Besides dlpcMimoComp is not a correct parameter. There should be RS letters in the parameter name.

Considering HW power and 10 MHz bandwidth, that would make 10W per PA branch.

Good luck

[anita_2k]

btw giegala, do you know as well the difference between
p-a and p-b as mentioned in 3gpp docs?

tnx
anita

34506

[giegala]

Hi,

P-a tells what is relative power on PDSCH per subcarrier on symbol that carries Rs (in normal CP it would be symbols 0 & 4)
P-b tells what is relative power on PDSCH per subcarrier on symbol that does not carry Rs (in normal CP it would be symbols 1,2,3,5 & 6)

Actually P-b is sent relative to Pa. Pb=1 means that the power of Pa and Pb are equal.

Normal Mimo 2x2 setup should have P-a= -3dB and P-b =1 (i.e. -3dB as well).

[Kamatchou]

Hi Guys,

Yes, dlpcMimoComp doesn't have an impact on the RS power. It reduces the power per antenna when using 2Tx and it applies to all channels except the RS.
The recommendation is to set the parameter to 0 dB independently of the pMax used i.e. MIMO Compensation is disabled, as better throughput are achieved.

It is possible to directly boost the RS starting RL30, through ‘RS boosting’ feature.

I hoep it helps :)